diff --git a/README.md b/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..622b1afbf67513c6d5b974cf6a1b6d5ad79c52e7
--- /dev/null
+++ b/README.md
@@ -0,0 +1,202 @@
+---
+base_model: liuhaotian/llava-v1.5-13b
+library_name: peft
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.13.2
\ No newline at end of file
diff --git a/adapter_config.json b/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..31d77354128d962ce655ffa50a52c067d2b8a463
--- /dev/null
+++ b/adapter_config.json
@@ -0,0 +1,34 @@
+{
+ "alpha_pattern": {},
+ "auto_mapping": null,
+ "base_model_name_or_path": "liuhaotian/llava-v1.5-13b",
+ "bias": "none",
+ "fan_in_fan_out": false,
+ "inference_mode": true,
+ "init_lora_weights": true,
+ "layer_replication": null,
+ "layers_pattern": null,
+ "layers_to_transform": null,
+ "loftq_config": {},
+ "lora_alpha": 16,
+ "lora_dropout": 0.05,
+ "megatron_config": null,
+ "megatron_core": "megatron.core",
+ "modules_to_save": null,
+ "peft_type": "LORA",
+ "r": 8,
+ "rank_pattern": {},
+ "revision": null,
+ "target_modules": [
+ "up_proj",
+ "k_proj",
+ "v_proj",
+ "gate_proj",
+ "o_proj",
+ "down_proj",
+ "q_proj"
+ ],
+ "task_type": "CAUSAL_LM",
+ "use_dora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/adapter_model.safetensors b/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..47909a01fda6c865b132a061ebf9081e42f0d0fb
--- /dev/null
+++ b/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0ca6fa3684a839a6dd096790f5cb429d7bac913ab314c609dc0399ab43390ad
+size 62660864
diff --git a/checkpoint-224/README.md b/checkpoint-224/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..622b1afbf67513c6d5b974cf6a1b6d5ad79c52e7
--- /dev/null
+++ b/checkpoint-224/README.md
@@ -0,0 +1,202 @@
+---
+base_model: liuhaotian/llava-v1.5-13b
+library_name: peft
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.13.2
\ No newline at end of file
diff --git a/checkpoint-224/adapter_config.json b/checkpoint-224/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b267ddc6d75edbd5b7129d5d37c47ac7274bc433
--- /dev/null
+++ b/checkpoint-224/adapter_config.json
@@ -0,0 +1,34 @@
+{
+ "alpha_pattern": {},
+ "auto_mapping": null,
+ "base_model_name_or_path": "liuhaotian/llava-v1.5-13b",
+ "bias": "none",
+ "fan_in_fan_out": false,
+ "inference_mode": true,
+ "init_lora_weights": true,
+ "layer_replication": null,
+ "layers_pattern": null,
+ "layers_to_transform": null,
+ "loftq_config": {},
+ "lora_alpha": 16,
+ "lora_dropout": 0.05,
+ "megatron_config": null,
+ "megatron_core": "megatron.core",
+ "modules_to_save": null,
+ "peft_type": "LORA",
+ "r": 8,
+ "rank_pattern": {},
+ "revision": null,
+ "target_modules": [
+ "v_proj",
+ "down_proj",
+ "up_proj",
+ "q_proj",
+ "o_proj",
+ "k_proj",
+ "gate_proj"
+ ],
+ "task_type": "CAUSAL_LM",
+ "use_dora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-224/adapter_model.safetensors b/checkpoint-224/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..47909a01fda6c865b132a061ebf9081e42f0d0fb
--- /dev/null
+++ b/checkpoint-224/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0ca6fa3684a839a6dd096790f5cb429d7bac913ab314c609dc0399ab43390ad
+size 62660864
diff --git a/checkpoint-224/global_step224/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-224/global_step224/zero_pp_rank_0_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..73fea0c043783e60ae87178982a1dcd3a25ec4b0
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_0_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:809c10f935f479188e529d8e510db416031e71c61c901246cbff27407af9f9a4
+size 593618
diff --git a/checkpoint-224/global_step224/zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-224/global_step224/zero_pp_rank_0_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8bd9dd0a9452c127924c3590d184883daafbf8b0
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_0_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:137299d61c0b7289153a620a6f2c7773a828e56c86835b354d73fcb65a3949d1
+size 188286957
diff --git a/checkpoint-224/global_step224/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-224/global_step224/zero_pp_rank_1_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..33bd25ff41a2aecc57b906db28dc40c6e86bd7ac
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_1_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5cdba4711b4b0d505ef85e7f5337c47e7f5e2da1639a74c451238d88a5025df
+size 593618
diff --git a/checkpoint-224/global_step224/zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-224/global_step224/zero_pp_rank_1_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..36844818fb0572b405449ca20fc2b5e266c92845
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_1_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:344f6a277d819456eefe986b75f61b1caef404db8f10ce1a1aaf23e6ad464af7
+size 188286957
diff --git a/checkpoint-224/global_step224/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-224/global_step224/zero_pp_rank_2_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6e40f3ac622df2fee7d342a5d31978530e06b544
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_2_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e32eaa7a0b327ca2f2cb690a2aed3e10e6c9155107947b0563bd7b1f17c58cee
+size 593618
diff --git a/checkpoint-224/global_step224/zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-224/global_step224/zero_pp_rank_2_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..13a24b398e1b74d0b86276c08c4eb35fc0527702
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_2_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b1f7a9eacac7373d9db9f2eb64e400b5f6c7767db0ec3bccb20f1292f75fd0d
+size 188286957
diff --git a/checkpoint-224/global_step224/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-224/global_step224/zero_pp_rank_3_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bab17376be9506a18ac3e85b86e3906a12b5807f
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_3_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad6eef1ecdff0f96311700f48b433a118a173b7c92e643450314d1e702f2503d
+size 593618
diff --git a/checkpoint-224/global_step224/zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-224/global_step224/zero_pp_rank_3_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a3e8c36714db7dd0be888396c853259edffe8c9a
--- /dev/null
+++ b/checkpoint-224/global_step224/zero_pp_rank_3_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:759ec840d614ccc1ef432f6d7ef14d5cf0c039a1c98e13bb80c6778f7d6ab6a8
+size 188286957
diff --git a/checkpoint-224/latest b/checkpoint-224/latest
new file mode 100644
index 0000000000000000000000000000000000000000..c2e6fcae2a3afef93707542af52323196e00f7b9
--- /dev/null
+++ b/checkpoint-224/latest
@@ -0,0 +1 @@
+global_step224
\ No newline at end of file
diff --git a/checkpoint-224/rng_state_0.pth b/checkpoint-224/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..a50621454578bf73f5804cb75a8dff9ff90ed73c
--- /dev/null
+++ b/checkpoint-224/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b32a0d78d2c8a662384fae768e5baa216e31a9e0266798371b91daacdc50f66b
+size 14960
diff --git a/checkpoint-224/rng_state_1.pth b/checkpoint-224/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..568648a3c1d1a2064e1dc99b1efe9c9f4c0aa862
--- /dev/null
+++ b/checkpoint-224/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b04c2612dcb2b3d3fd7c114e9a5fdd7ec6de1ea0b8cc1ac278cbf363b67e5eb5
+size 14960
diff --git a/checkpoint-224/rng_state_2.pth b/checkpoint-224/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..84cbcee450e57c4816bca2e0565a16a6afe78c2f
--- /dev/null
+++ b/checkpoint-224/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:652571be01e2aa6a8f0308d56ef329284a66de86ae1f13d2a777a17d7a5fbea8
+size 14960
diff --git a/checkpoint-224/rng_state_3.pth b/checkpoint-224/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..01f6f5afb070d91e8aaf74110ee2083849e9bb5a
--- /dev/null
+++ b/checkpoint-224/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7f2cb749c8a1d3504339650d317841f232205be1d955c4f0dab1072cd2ecfe7
+size 14960
diff --git a/checkpoint-224/special_tokens_map.json b/checkpoint-224/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..14761dcf1466dc232bd41de9c21d4c617b15755e
--- /dev/null
+++ b/checkpoint-224/special_tokens_map.json
@@ -0,0 +1,24 @@
+{
+ "bos_token": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ "eos_token": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ "pad_token": "",
+ "unk_token": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ }
+}
diff --git a/checkpoint-224/tokenizer.model b/checkpoint-224/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899
--- /dev/null
+++ b/checkpoint-224/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723
diff --git a/checkpoint-224/tokenizer_config.json b/checkpoint-224/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..26c65df1bf794f101c1dd54c908180dc0d880fe3
--- /dev/null
+++ b/checkpoint-224/tokenizer_config.json
@@ -0,0 +1,43 @@
+{
+ "add_bos_token": true,
+ "add_eos_token": false,
+ "add_prefix_space": true,
+ "added_tokens_decoder": {
+ "0": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "1": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "2": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ }
+ },
+ "bos_token": "",
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "",
+ "legacy": false,
+ "model_max_length": 2048,
+ "pad_token": "",
+ "padding_side": "right",
+ "sp_model_kwargs": {},
+ "spaces_between_special_tokens": false,
+ "tokenizer_class": "LlamaTokenizer",
+ "unk_token": "",
+ "use_default_system_prompt": false
+}
diff --git a/checkpoint-224/trainer_state.json b/checkpoint-224/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..5a1ed44795343b36af9a1d262bb4c1787e400221
--- /dev/null
+++ b/checkpoint-224/trainer_state.json
@@ -0,0 +1,3393 @@
+{
+ "best_metric": 0.6895740032196045,
+ "best_model_checkpoint": "./checkpoints/llava-v1.5-13b/checkpoint-224",
+ "epoch": 7.0,
+ "eval_steps": 1.0,
+ "global_step": 224,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.03125,
+ "grad_norm": 0.2380081706918525,
+ "learning_rate": 0.0,
+ "loss": 1.2458,
+ "step": 1
+ },
+ {
+ "epoch": 0.03125,
+ "eval_loss": 1.3161638975143433,
+ "eval_runtime": 50.8995,
+ "eval_samples_per_second": 3.929,
+ "eval_steps_per_second": 0.255,
+ "step": 1
+ },
+ {
+ "epoch": 0.0625,
+ "grad_norm": 0.20429495268987705,
+ "learning_rate": 8.613531161467863e-06,
+ "loss": 1.2003,
+ "step": 2
+ },
+ {
+ "epoch": 0.0625,
+ "eval_loss": 1.3161638975143433,
+ "eval_runtime": 47.4818,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 2
+ },
+ {
+ "epoch": 0.09375,
+ "grad_norm": 0.20616215800420787,
+ "learning_rate": 1.3652123889719709e-05,
+ "loss": 1.2622,
+ "step": 3
+ },
+ {
+ "epoch": 0.09375,
+ "eval_loss": 1.309991478919983,
+ "eval_runtime": 47.4152,
+ "eval_samples_per_second": 4.218,
+ "eval_steps_per_second": 0.274,
+ "step": 3
+ },
+ {
+ "epoch": 0.125,
+ "grad_norm": 0.20155595022101944,
+ "learning_rate": 1.7227062322935725e-05,
+ "loss": 1.2845,
+ "step": 4
+ },
+ {
+ "epoch": 0.125,
+ "eval_loss": 1.3013781309127808,
+ "eval_runtime": 47.4814,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 4
+ },
+ {
+ "epoch": 0.15625,
+ "grad_norm": 0.21113117474989132,
+ "learning_rate": 2e-05,
+ "loss": 1.246,
+ "step": 5
+ },
+ {
+ "epoch": 0.15625,
+ "eval_loss": 1.2892160415649414,
+ "eval_runtime": 47.7209,
+ "eval_samples_per_second": 4.191,
+ "eval_steps_per_second": 0.272,
+ "step": 5
+ },
+ {
+ "epoch": 0.1875,
+ "grad_norm": 0.21377946631015488,
+ "learning_rate": 2e-05,
+ "loss": 1.2684,
+ "step": 6
+ },
+ {
+ "epoch": 0.1875,
+ "eval_loss": 1.2754532098770142,
+ "eval_runtime": 47.5781,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 6
+ },
+ {
+ "epoch": 0.21875,
+ "grad_norm": 0.2284268997618767,
+ "learning_rate": 2e-05,
+ "loss": 1.2681,
+ "step": 7
+ },
+ {
+ "epoch": 0.21875,
+ "eval_loss": 1.2605774402618408,
+ "eval_runtime": 47.5326,
+ "eval_samples_per_second": 4.208,
+ "eval_steps_per_second": 0.273,
+ "step": 7
+ },
+ {
+ "epoch": 0.25,
+ "grad_norm": 0.23585343568544442,
+ "learning_rate": 2e-05,
+ "loss": 1.2407,
+ "step": 8
+ },
+ {
+ "epoch": 0.25,
+ "eval_loss": 1.244718313217163,
+ "eval_runtime": 47.5001,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 8
+ },
+ {
+ "epoch": 0.28125,
+ "grad_norm": 0.23051191992462533,
+ "learning_rate": 2e-05,
+ "loss": 1.2766,
+ "step": 9
+ },
+ {
+ "epoch": 0.28125,
+ "eval_loss": 1.2285138368606567,
+ "eval_runtime": 47.4631,
+ "eval_samples_per_second": 4.214,
+ "eval_steps_per_second": 0.274,
+ "step": 9
+ },
+ {
+ "epoch": 0.3125,
+ "grad_norm": 0.22726394327484983,
+ "learning_rate": 2e-05,
+ "loss": 1.2024,
+ "step": 10
+ },
+ {
+ "epoch": 0.3125,
+ "eval_loss": 1.2118008136749268,
+ "eval_runtime": 47.4991,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 10
+ },
+ {
+ "epoch": 0.34375,
+ "grad_norm": 0.25404890894461285,
+ "learning_rate": 2e-05,
+ "loss": 1.2742,
+ "step": 11
+ },
+ {
+ "epoch": 0.34375,
+ "eval_loss": 1.1942989826202393,
+ "eval_runtime": 49.2609,
+ "eval_samples_per_second": 4.06,
+ "eval_steps_per_second": 0.264,
+ "step": 11
+ },
+ {
+ "epoch": 0.375,
+ "grad_norm": 0.26336210916526287,
+ "learning_rate": 2e-05,
+ "loss": 1.2258,
+ "step": 12
+ },
+ {
+ "epoch": 0.375,
+ "eval_loss": 1.176426649093628,
+ "eval_runtime": 49.0639,
+ "eval_samples_per_second": 4.076,
+ "eval_steps_per_second": 0.265,
+ "step": 12
+ },
+ {
+ "epoch": 0.40625,
+ "grad_norm": 0.29637148470746666,
+ "learning_rate": 2e-05,
+ "loss": 1.2345,
+ "step": 13
+ },
+ {
+ "epoch": 0.40625,
+ "eval_loss": 1.1577811241149902,
+ "eval_runtime": 49.1352,
+ "eval_samples_per_second": 4.07,
+ "eval_steps_per_second": 0.265,
+ "step": 13
+ },
+ {
+ "epoch": 0.4375,
+ "grad_norm": 0.2841880377627424,
+ "learning_rate": 2e-05,
+ "loss": 1.0765,
+ "step": 14
+ },
+ {
+ "epoch": 0.4375,
+ "eval_loss": 1.1381279230117798,
+ "eval_runtime": 49.25,
+ "eval_samples_per_second": 4.061,
+ "eval_steps_per_second": 0.264,
+ "step": 14
+ },
+ {
+ "epoch": 0.46875,
+ "grad_norm": 0.2773140636191091,
+ "learning_rate": 2e-05,
+ "loss": 1.1812,
+ "step": 15
+ },
+ {
+ "epoch": 0.46875,
+ "eval_loss": 1.1178216934204102,
+ "eval_runtime": 49.0879,
+ "eval_samples_per_second": 4.074,
+ "eval_steps_per_second": 0.265,
+ "step": 15
+ },
+ {
+ "epoch": 0.5,
+ "grad_norm": 0.3568607365552051,
+ "learning_rate": 2e-05,
+ "loss": 1.1327,
+ "step": 16
+ },
+ {
+ "epoch": 0.5,
+ "eval_loss": 1.0954149961471558,
+ "eval_runtime": 48.6546,
+ "eval_samples_per_second": 4.111,
+ "eval_steps_per_second": 0.267,
+ "step": 16
+ },
+ {
+ "epoch": 0.53125,
+ "grad_norm": 0.32574391414112897,
+ "learning_rate": 2e-05,
+ "loss": 1.1162,
+ "step": 17
+ },
+ {
+ "epoch": 0.53125,
+ "eval_loss": 1.071275234222412,
+ "eval_runtime": 48.5618,
+ "eval_samples_per_second": 4.118,
+ "eval_steps_per_second": 0.268,
+ "step": 17
+ },
+ {
+ "epoch": 0.5625,
+ "grad_norm": 0.4256864144638081,
+ "learning_rate": 2e-05,
+ "loss": 1.1138,
+ "step": 18
+ },
+ {
+ "epoch": 0.5625,
+ "eval_loss": 1.0455905199050903,
+ "eval_runtime": 48.4981,
+ "eval_samples_per_second": 4.124,
+ "eval_steps_per_second": 0.268,
+ "step": 18
+ },
+ {
+ "epoch": 0.59375,
+ "grad_norm": 0.31230014132112643,
+ "learning_rate": 2e-05,
+ "loss": 1.0011,
+ "step": 19
+ },
+ {
+ "epoch": 0.59375,
+ "eval_loss": 1.0208789110183716,
+ "eval_runtime": 48.4675,
+ "eval_samples_per_second": 4.126,
+ "eval_steps_per_second": 0.268,
+ "step": 19
+ },
+ {
+ "epoch": 0.625,
+ "grad_norm": 0.3025724039243594,
+ "learning_rate": 2e-05,
+ "loss": 1.109,
+ "step": 20
+ },
+ {
+ "epoch": 0.625,
+ "eval_loss": 1.002480149269104,
+ "eval_runtime": 48.5265,
+ "eval_samples_per_second": 4.121,
+ "eval_steps_per_second": 0.268,
+ "step": 20
+ },
+ {
+ "epoch": 0.65625,
+ "grad_norm": 0.27787879590501874,
+ "learning_rate": 2e-05,
+ "loss": 1.0291,
+ "step": 21
+ },
+ {
+ "epoch": 0.65625,
+ "eval_loss": 0.9933492541313171,
+ "eval_runtime": 50.0369,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 21
+ },
+ {
+ "epoch": 0.6875,
+ "grad_norm": 0.4231294067130801,
+ "learning_rate": 2e-05,
+ "loss": 1.0779,
+ "step": 22
+ },
+ {
+ "epoch": 0.6875,
+ "eval_loss": 0.9850385785102844,
+ "eval_runtime": 50.0062,
+ "eval_samples_per_second": 4.0,
+ "eval_steps_per_second": 0.26,
+ "step": 22
+ },
+ {
+ "epoch": 0.71875,
+ "grad_norm": 0.42130097437373987,
+ "learning_rate": 2e-05,
+ "loss": 1.0897,
+ "step": 23
+ },
+ {
+ "epoch": 0.71875,
+ "eval_loss": 0.9758670330047607,
+ "eval_runtime": 50.1031,
+ "eval_samples_per_second": 3.992,
+ "eval_steps_per_second": 0.259,
+ "step": 23
+ },
+ {
+ "epoch": 0.75,
+ "grad_norm": 0.27711808063263893,
+ "learning_rate": 2e-05,
+ "loss": 1.0739,
+ "step": 24
+ },
+ {
+ "epoch": 0.75,
+ "eval_loss": 0.9674506187438965,
+ "eval_runtime": 50.0337,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 24
+ },
+ {
+ "epoch": 0.78125,
+ "grad_norm": 0.2879649409281791,
+ "learning_rate": 2e-05,
+ "loss": 1.0182,
+ "step": 25
+ },
+ {
+ "epoch": 0.78125,
+ "eval_loss": 0.9592065215110779,
+ "eval_runtime": 50.0709,
+ "eval_samples_per_second": 3.994,
+ "eval_steps_per_second": 0.26,
+ "step": 25
+ },
+ {
+ "epoch": 0.8125,
+ "grad_norm": 0.19327450826076825,
+ "learning_rate": 2e-05,
+ "loss": 1.0413,
+ "step": 26
+ },
+ {
+ "epoch": 0.8125,
+ "eval_loss": 0.9518552422523499,
+ "eval_runtime": 50.0572,
+ "eval_samples_per_second": 3.995,
+ "eval_steps_per_second": 0.26,
+ "step": 26
+ },
+ {
+ "epoch": 0.84375,
+ "grad_norm": 0.19707021382445633,
+ "learning_rate": 2e-05,
+ "loss": 0.9525,
+ "step": 27
+ },
+ {
+ "epoch": 0.84375,
+ "eval_loss": 0.9449941515922546,
+ "eval_runtime": 50.0515,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 27
+ },
+ {
+ "epoch": 0.875,
+ "grad_norm": 0.2420270757641518,
+ "learning_rate": 2e-05,
+ "loss": 0.9658,
+ "step": 28
+ },
+ {
+ "epoch": 0.875,
+ "eval_loss": 0.9378474354743958,
+ "eval_runtime": 49.9299,
+ "eval_samples_per_second": 4.006,
+ "eval_steps_per_second": 0.26,
+ "step": 28
+ },
+ {
+ "epoch": 0.90625,
+ "grad_norm": 0.18074632782127534,
+ "learning_rate": 2e-05,
+ "loss": 0.9866,
+ "step": 29
+ },
+ {
+ "epoch": 0.90625,
+ "eval_loss": 0.93099045753479,
+ "eval_runtime": 50.0096,
+ "eval_samples_per_second": 3.999,
+ "eval_steps_per_second": 0.26,
+ "step": 29
+ },
+ {
+ "epoch": 0.9375,
+ "grad_norm": 0.1936051126921734,
+ "learning_rate": 2e-05,
+ "loss": 1.0128,
+ "step": 30
+ },
+ {
+ "epoch": 0.9375,
+ "eval_loss": 0.9244199991226196,
+ "eval_runtime": 50.2469,
+ "eval_samples_per_second": 3.98,
+ "eval_steps_per_second": 0.259,
+ "step": 30
+ },
+ {
+ "epoch": 0.96875,
+ "grad_norm": 0.26164254459782943,
+ "learning_rate": 2e-05,
+ "loss": 0.88,
+ "step": 31
+ },
+ {
+ "epoch": 0.96875,
+ "eval_loss": 0.9175177216529846,
+ "eval_runtime": 50.1695,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 31
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.18677152741688485,
+ "learning_rate": 2e-05,
+ "loss": 0.9569,
+ "step": 32
+ },
+ {
+ "epoch": 1.0,
+ "eval_loss": 0.9108598828315735,
+ "eval_runtime": 50.0387,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 32
+ },
+ {
+ "epoch": 1.03125,
+ "grad_norm": 0.20486279036126417,
+ "learning_rate": 2e-05,
+ "loss": 1.0208,
+ "step": 33
+ },
+ {
+ "epoch": 1.03125,
+ "eval_loss": 0.9042049646377563,
+ "eval_runtime": 50.1472,
+ "eval_samples_per_second": 3.988,
+ "eval_steps_per_second": 0.259,
+ "step": 33
+ },
+ {
+ "epoch": 1.0625,
+ "grad_norm": 0.2004946169291112,
+ "learning_rate": 2e-05,
+ "loss": 0.9931,
+ "step": 34
+ },
+ {
+ "epoch": 1.0625,
+ "eval_loss": 0.8980298042297363,
+ "eval_runtime": 50.245,
+ "eval_samples_per_second": 3.98,
+ "eval_steps_per_second": 0.259,
+ "step": 34
+ },
+ {
+ "epoch": 1.09375,
+ "grad_norm": 0.1645872432258401,
+ "learning_rate": 2e-05,
+ "loss": 1.0184,
+ "step": 35
+ },
+ {
+ "epoch": 1.09375,
+ "eval_loss": 0.8924428820610046,
+ "eval_runtime": 50.3703,
+ "eval_samples_per_second": 3.971,
+ "eval_steps_per_second": 0.258,
+ "step": 35
+ },
+ {
+ "epoch": 1.125,
+ "grad_norm": 0.18293519304435016,
+ "learning_rate": 2e-05,
+ "loss": 1.0026,
+ "step": 36
+ },
+ {
+ "epoch": 1.125,
+ "eval_loss": 0.8870412707328796,
+ "eval_runtime": 50.0483,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 36
+ },
+ {
+ "epoch": 1.15625,
+ "grad_norm": 0.17712548516246762,
+ "learning_rate": 2e-05,
+ "loss": 0.9387,
+ "step": 37
+ },
+ {
+ "epoch": 1.15625,
+ "eval_loss": 0.881915271282196,
+ "eval_runtime": 49.9751,
+ "eval_samples_per_second": 4.002,
+ "eval_steps_per_second": 0.26,
+ "step": 37
+ },
+ {
+ "epoch": 1.1875,
+ "grad_norm": 0.21472689311609464,
+ "learning_rate": 2e-05,
+ "loss": 0.958,
+ "step": 38
+ },
+ {
+ "epoch": 1.1875,
+ "eval_loss": 0.8768754601478577,
+ "eval_runtime": 50.1204,
+ "eval_samples_per_second": 3.99,
+ "eval_steps_per_second": 0.259,
+ "step": 38
+ },
+ {
+ "epoch": 1.21875,
+ "grad_norm": 0.21117297910005806,
+ "learning_rate": 2e-05,
+ "loss": 0.9922,
+ "step": 39
+ },
+ {
+ "epoch": 1.21875,
+ "eval_loss": 0.8718628883361816,
+ "eval_runtime": 50.1732,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 39
+ },
+ {
+ "epoch": 1.25,
+ "grad_norm": 0.17835587003909165,
+ "learning_rate": 2e-05,
+ "loss": 0.9776,
+ "step": 40
+ },
+ {
+ "epoch": 1.25,
+ "eval_loss": 0.8669865131378174,
+ "eval_runtime": 50.1148,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 40
+ },
+ {
+ "epoch": 1.28125,
+ "grad_norm": 0.2092736372483734,
+ "learning_rate": 2e-05,
+ "loss": 0.9731,
+ "step": 41
+ },
+ {
+ "epoch": 1.28125,
+ "eval_loss": 0.8619834780693054,
+ "eval_runtime": 50.052,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 41
+ },
+ {
+ "epoch": 1.3125,
+ "grad_norm": 0.2338857391910308,
+ "learning_rate": 2e-05,
+ "loss": 0.9319,
+ "step": 42
+ },
+ {
+ "epoch": 1.3125,
+ "eval_loss": 0.8572126030921936,
+ "eval_runtime": 50.1212,
+ "eval_samples_per_second": 3.99,
+ "eval_steps_per_second": 0.259,
+ "step": 42
+ },
+ {
+ "epoch": 1.34375,
+ "grad_norm": 0.19168719284572813,
+ "learning_rate": 2e-05,
+ "loss": 0.9083,
+ "step": 43
+ },
+ {
+ "epoch": 1.34375,
+ "eval_loss": 0.8525611758232117,
+ "eval_runtime": 50.1733,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 43
+ },
+ {
+ "epoch": 1.375,
+ "grad_norm": 0.20004868138433377,
+ "learning_rate": 2e-05,
+ "loss": 0.9118,
+ "step": 44
+ },
+ {
+ "epoch": 1.375,
+ "eval_loss": 0.8483461141586304,
+ "eval_runtime": 50.1083,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 44
+ },
+ {
+ "epoch": 1.40625,
+ "grad_norm": 0.19012965506122342,
+ "learning_rate": 2e-05,
+ "loss": 0.8888,
+ "step": 45
+ },
+ {
+ "epoch": 1.40625,
+ "eval_loss": 0.8446614742279053,
+ "eval_runtime": 50.1171,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 45
+ },
+ {
+ "epoch": 1.4375,
+ "grad_norm": 0.21187005706805245,
+ "learning_rate": 2e-05,
+ "loss": 0.9319,
+ "step": 46
+ },
+ {
+ "epoch": 1.4375,
+ "eval_loss": 0.8412036299705505,
+ "eval_runtime": 50.0918,
+ "eval_samples_per_second": 3.993,
+ "eval_steps_per_second": 0.26,
+ "step": 46
+ },
+ {
+ "epoch": 1.46875,
+ "grad_norm": 0.19673832205926584,
+ "learning_rate": 2e-05,
+ "loss": 0.9359,
+ "step": 47
+ },
+ {
+ "epoch": 1.46875,
+ "eval_loss": 0.8380417823791504,
+ "eval_runtime": 50.2214,
+ "eval_samples_per_second": 3.982,
+ "eval_steps_per_second": 0.259,
+ "step": 47
+ },
+ {
+ "epoch": 1.5,
+ "grad_norm": 0.21712294106174318,
+ "learning_rate": 2e-05,
+ "loss": 0.8511,
+ "step": 48
+ },
+ {
+ "epoch": 1.5,
+ "eval_loss": 0.8353021740913391,
+ "eval_runtime": 50.1617,
+ "eval_samples_per_second": 3.987,
+ "eval_steps_per_second": 0.259,
+ "step": 48
+ },
+ {
+ "epoch": 1.53125,
+ "grad_norm": 0.2138924779700934,
+ "learning_rate": 2e-05,
+ "loss": 0.8695,
+ "step": 49
+ },
+ {
+ "epoch": 1.53125,
+ "eval_loss": 0.8327407836914062,
+ "eval_runtime": 50.1442,
+ "eval_samples_per_second": 3.988,
+ "eval_steps_per_second": 0.259,
+ "step": 49
+ },
+ {
+ "epoch": 1.5625,
+ "grad_norm": 0.22387442384578618,
+ "learning_rate": 2e-05,
+ "loss": 0.8518,
+ "step": 50
+ },
+ {
+ "epoch": 1.5625,
+ "eval_loss": 0.8301742076873779,
+ "eval_runtime": 50.1867,
+ "eval_samples_per_second": 3.985,
+ "eval_steps_per_second": 0.259,
+ "step": 50
+ },
+ {
+ "epoch": 1.59375,
+ "grad_norm": 0.1975577146517192,
+ "learning_rate": 2e-05,
+ "loss": 0.8868,
+ "step": 51
+ },
+ {
+ "epoch": 1.59375,
+ "eval_loss": 0.8275265693664551,
+ "eval_runtime": 51.2257,
+ "eval_samples_per_second": 3.904,
+ "eval_steps_per_second": 0.254,
+ "step": 51
+ },
+ {
+ "epoch": 1.625,
+ "grad_norm": 0.21474817057286624,
+ "learning_rate": 2e-05,
+ "loss": 0.767,
+ "step": 52
+ },
+ {
+ "epoch": 1.625,
+ "eval_loss": 0.824796736240387,
+ "eval_runtime": 51.276,
+ "eval_samples_per_second": 3.9,
+ "eval_steps_per_second": 0.254,
+ "step": 52
+ },
+ {
+ "epoch": 1.65625,
+ "grad_norm": 0.21105651676755652,
+ "learning_rate": 2e-05,
+ "loss": 0.9219,
+ "step": 53
+ },
+ {
+ "epoch": 1.65625,
+ "eval_loss": 0.8221166729927063,
+ "eval_runtime": 51.141,
+ "eval_samples_per_second": 3.911,
+ "eval_steps_per_second": 0.254,
+ "step": 53
+ },
+ {
+ "epoch": 1.6875,
+ "grad_norm": 0.20706475184742085,
+ "learning_rate": 2e-05,
+ "loss": 0.8873,
+ "step": 54
+ },
+ {
+ "epoch": 1.6875,
+ "eval_loss": 0.819589376449585,
+ "eval_runtime": 51.0045,
+ "eval_samples_per_second": 3.921,
+ "eval_steps_per_second": 0.255,
+ "step": 54
+ },
+ {
+ "epoch": 1.71875,
+ "grad_norm": 0.21722220033855957,
+ "learning_rate": 2e-05,
+ "loss": 0.8956,
+ "step": 55
+ },
+ {
+ "epoch": 1.71875,
+ "eval_loss": 0.8176340460777283,
+ "eval_runtime": 51.1941,
+ "eval_samples_per_second": 3.907,
+ "eval_steps_per_second": 0.254,
+ "step": 55
+ },
+ {
+ "epoch": 1.75,
+ "grad_norm": 0.20669001221665667,
+ "learning_rate": 2e-05,
+ "loss": 0.9506,
+ "step": 56
+ },
+ {
+ "epoch": 1.75,
+ "eval_loss": 0.8158826231956482,
+ "eval_runtime": 52.1162,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 56
+ },
+ {
+ "epoch": 1.78125,
+ "grad_norm": 0.22189732090066341,
+ "learning_rate": 2e-05,
+ "loss": 0.8955,
+ "step": 57
+ },
+ {
+ "epoch": 1.78125,
+ "eval_loss": 0.814656674861908,
+ "eval_runtime": 52.1361,
+ "eval_samples_per_second": 3.836,
+ "eval_steps_per_second": 0.249,
+ "step": 57
+ },
+ {
+ "epoch": 1.8125,
+ "grad_norm": 0.2030113892848459,
+ "learning_rate": 2e-05,
+ "loss": 0.9108,
+ "step": 58
+ },
+ {
+ "epoch": 1.8125,
+ "eval_loss": 0.813343346118927,
+ "eval_runtime": 52.2552,
+ "eval_samples_per_second": 3.827,
+ "eval_steps_per_second": 0.249,
+ "step": 58
+ },
+ {
+ "epoch": 1.84375,
+ "grad_norm": 0.2123201057569791,
+ "learning_rate": 2e-05,
+ "loss": 0.8779,
+ "step": 59
+ },
+ {
+ "epoch": 1.84375,
+ "eval_loss": 0.8116877675056458,
+ "eval_runtime": 52.1233,
+ "eval_samples_per_second": 3.837,
+ "eval_steps_per_second": 0.249,
+ "step": 59
+ },
+ {
+ "epoch": 1.875,
+ "grad_norm": 0.211551126937912,
+ "learning_rate": 2e-05,
+ "loss": 0.9294,
+ "step": 60
+ },
+ {
+ "epoch": 1.875,
+ "eval_loss": 0.8098442554473877,
+ "eval_runtime": 52.1091,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 60
+ },
+ {
+ "epoch": 1.90625,
+ "grad_norm": 0.24981344981629752,
+ "learning_rate": 2e-05,
+ "loss": 0.8409,
+ "step": 61
+ },
+ {
+ "epoch": 1.90625,
+ "eval_loss": 0.8070770502090454,
+ "eval_runtime": 53.4187,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 61
+ },
+ {
+ "epoch": 1.9375,
+ "grad_norm": 0.2341550589775159,
+ "learning_rate": 2e-05,
+ "loss": 0.888,
+ "step": 62
+ },
+ {
+ "epoch": 1.9375,
+ "eval_loss": 0.8040286898612976,
+ "eval_runtime": 53.2197,
+ "eval_samples_per_second": 3.758,
+ "eval_steps_per_second": 0.244,
+ "step": 62
+ },
+ {
+ "epoch": 1.96875,
+ "grad_norm": 0.2336241775649256,
+ "learning_rate": 2e-05,
+ "loss": 0.913,
+ "step": 63
+ },
+ {
+ "epoch": 1.96875,
+ "eval_loss": 0.8013430833816528,
+ "eval_runtime": 53.1784,
+ "eval_samples_per_second": 3.761,
+ "eval_steps_per_second": 0.244,
+ "step": 63
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.2414390628081758,
+ "learning_rate": 2e-05,
+ "loss": 0.8754,
+ "step": 64
+ },
+ {
+ "epoch": 2.0,
+ "eval_loss": 0.7985894680023193,
+ "eval_runtime": 53.2454,
+ "eval_samples_per_second": 3.756,
+ "eval_steps_per_second": 0.244,
+ "step": 64
+ },
+ {
+ "epoch": 2.03125,
+ "grad_norm": 0.2484104465653703,
+ "learning_rate": 2e-05,
+ "loss": 0.8497,
+ "step": 65
+ },
+ {
+ "epoch": 2.03125,
+ "eval_loss": 0.7954932451248169,
+ "eval_runtime": 53.3794,
+ "eval_samples_per_second": 3.747,
+ "eval_steps_per_second": 0.244,
+ "step": 65
+ },
+ {
+ "epoch": 2.0625,
+ "grad_norm": 0.23859744120942086,
+ "learning_rate": 2e-05,
+ "loss": 0.8567,
+ "step": 66
+ },
+ {
+ "epoch": 2.0625,
+ "eval_loss": 0.7929843068122864,
+ "eval_runtime": 55.517,
+ "eval_samples_per_second": 3.602,
+ "eval_steps_per_second": 0.234,
+ "step": 66
+ },
+ {
+ "epoch": 2.09375,
+ "grad_norm": 0.24584758647855462,
+ "learning_rate": 2e-05,
+ "loss": 0.8489,
+ "step": 67
+ },
+ {
+ "epoch": 2.09375,
+ "eval_loss": 0.7903321981430054,
+ "eval_runtime": 55.4151,
+ "eval_samples_per_second": 3.609,
+ "eval_steps_per_second": 0.235,
+ "step": 67
+ },
+ {
+ "epoch": 2.125,
+ "grad_norm": 0.2484917818304153,
+ "learning_rate": 2e-05,
+ "loss": 0.9122,
+ "step": 68
+ },
+ {
+ "epoch": 2.125,
+ "eval_loss": 0.7877185344696045,
+ "eval_runtime": 55.4069,
+ "eval_samples_per_second": 3.61,
+ "eval_steps_per_second": 0.235,
+ "step": 68
+ },
+ {
+ "epoch": 2.15625,
+ "grad_norm": 0.2184614083026819,
+ "learning_rate": 2e-05,
+ "loss": 0.8355,
+ "step": 69
+ },
+ {
+ "epoch": 2.15625,
+ "eval_loss": 0.7852210998535156,
+ "eval_runtime": 55.3381,
+ "eval_samples_per_second": 3.614,
+ "eval_steps_per_second": 0.235,
+ "step": 69
+ },
+ {
+ "epoch": 2.1875,
+ "grad_norm": 0.24978410070800153,
+ "learning_rate": 2e-05,
+ "loss": 0.7968,
+ "step": 70
+ },
+ {
+ "epoch": 2.1875,
+ "eval_loss": 0.7827157378196716,
+ "eval_runtime": 55.3708,
+ "eval_samples_per_second": 3.612,
+ "eval_steps_per_second": 0.235,
+ "step": 70
+ },
+ {
+ "epoch": 2.21875,
+ "grad_norm": 0.23059883325890385,
+ "learning_rate": 2e-05,
+ "loss": 0.8783,
+ "step": 71
+ },
+ {
+ "epoch": 2.21875,
+ "eval_loss": 0.7805906534194946,
+ "eval_runtime": 55.6033,
+ "eval_samples_per_second": 3.597,
+ "eval_steps_per_second": 0.234,
+ "step": 71
+ },
+ {
+ "epoch": 2.25,
+ "grad_norm": 0.23261007334915096,
+ "learning_rate": 2e-05,
+ "loss": 0.7956,
+ "step": 72
+ },
+ {
+ "epoch": 2.25,
+ "eval_loss": 0.7786691784858704,
+ "eval_runtime": 55.0913,
+ "eval_samples_per_second": 3.63,
+ "eval_steps_per_second": 0.236,
+ "step": 72
+ },
+ {
+ "epoch": 2.28125,
+ "grad_norm": 0.25779598356574085,
+ "learning_rate": 2e-05,
+ "loss": 0.8426,
+ "step": 73
+ },
+ {
+ "epoch": 2.28125,
+ "eval_loss": 0.7771151661872864,
+ "eval_runtime": 55.0698,
+ "eval_samples_per_second": 3.632,
+ "eval_steps_per_second": 0.236,
+ "step": 73
+ },
+ {
+ "epoch": 2.3125,
+ "grad_norm": 0.2288243335971112,
+ "learning_rate": 2e-05,
+ "loss": 0.8381,
+ "step": 74
+ },
+ {
+ "epoch": 2.3125,
+ "eval_loss": 0.7756838202476501,
+ "eval_runtime": 54.8412,
+ "eval_samples_per_second": 3.647,
+ "eval_steps_per_second": 0.237,
+ "step": 74
+ },
+ {
+ "epoch": 2.34375,
+ "grad_norm": 0.24235644907977733,
+ "learning_rate": 2e-05,
+ "loss": 0.887,
+ "step": 75
+ },
+ {
+ "epoch": 2.34375,
+ "eval_loss": 0.7739972472190857,
+ "eval_runtime": 54.9718,
+ "eval_samples_per_second": 3.638,
+ "eval_steps_per_second": 0.236,
+ "step": 75
+ },
+ {
+ "epoch": 2.375,
+ "grad_norm": 0.23666820017867402,
+ "learning_rate": 2e-05,
+ "loss": 0.8007,
+ "step": 76
+ },
+ {
+ "epoch": 2.375,
+ "eval_loss": 0.7724328637123108,
+ "eval_runtime": 55.0225,
+ "eval_samples_per_second": 3.635,
+ "eval_steps_per_second": 0.236,
+ "step": 76
+ },
+ {
+ "epoch": 2.40625,
+ "grad_norm": 0.22815737396609181,
+ "learning_rate": 2e-05,
+ "loss": 0.8529,
+ "step": 77
+ },
+ {
+ "epoch": 2.40625,
+ "eval_loss": 0.7710004448890686,
+ "eval_runtime": 55.321,
+ "eval_samples_per_second": 3.615,
+ "eval_steps_per_second": 0.235,
+ "step": 77
+ },
+ {
+ "epoch": 2.4375,
+ "grad_norm": 0.2701264871470739,
+ "learning_rate": 2e-05,
+ "loss": 0.8515,
+ "step": 78
+ },
+ {
+ "epoch": 2.4375,
+ "eval_loss": 0.7695322632789612,
+ "eval_runtime": 55.3045,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 78
+ },
+ {
+ "epoch": 2.46875,
+ "grad_norm": 0.24363813951328234,
+ "learning_rate": 2e-05,
+ "loss": 0.8587,
+ "step": 79
+ },
+ {
+ "epoch": 2.46875,
+ "eval_loss": 0.7689024209976196,
+ "eval_runtime": 55.3009,
+ "eval_samples_per_second": 3.617,
+ "eval_steps_per_second": 0.235,
+ "step": 79
+ },
+ {
+ "epoch": 2.5,
+ "grad_norm": 0.30924701355253065,
+ "learning_rate": 2e-05,
+ "loss": 0.9076,
+ "step": 80
+ },
+ {
+ "epoch": 2.5,
+ "eval_loss": 0.7676254510879517,
+ "eval_runtime": 55.2365,
+ "eval_samples_per_second": 3.621,
+ "eval_steps_per_second": 0.235,
+ "step": 80
+ },
+ {
+ "epoch": 2.53125,
+ "grad_norm": 0.2665188280221636,
+ "learning_rate": 2e-05,
+ "loss": 0.8445,
+ "step": 81
+ },
+ {
+ "epoch": 2.53125,
+ "eval_loss": 0.7661146521568298,
+ "eval_runtime": 55.2775,
+ "eval_samples_per_second": 3.618,
+ "eval_steps_per_second": 0.235,
+ "step": 81
+ },
+ {
+ "epoch": 2.5625,
+ "grad_norm": 0.24674191720675534,
+ "learning_rate": 2e-05,
+ "loss": 0.8882,
+ "step": 82
+ },
+ {
+ "epoch": 2.5625,
+ "eval_loss": 0.76513671875,
+ "eval_runtime": 55.0857,
+ "eval_samples_per_second": 3.631,
+ "eval_steps_per_second": 0.236,
+ "step": 82
+ },
+ {
+ "epoch": 2.59375,
+ "grad_norm": 0.2736689405531704,
+ "learning_rate": 2e-05,
+ "loss": 0.8336,
+ "step": 83
+ },
+ {
+ "epoch": 2.59375,
+ "eval_loss": 0.764373779296875,
+ "eval_runtime": 55.2069,
+ "eval_samples_per_second": 3.623,
+ "eval_steps_per_second": 0.235,
+ "step": 83
+ },
+ {
+ "epoch": 2.625,
+ "grad_norm": 0.290841287198557,
+ "learning_rate": 2e-05,
+ "loss": 0.795,
+ "step": 84
+ },
+ {
+ "epoch": 2.625,
+ "eval_loss": 0.7632084488868713,
+ "eval_runtime": 55.1009,
+ "eval_samples_per_second": 3.63,
+ "eval_steps_per_second": 0.236,
+ "step": 84
+ },
+ {
+ "epoch": 2.65625,
+ "grad_norm": 0.2912051076836381,
+ "learning_rate": 2e-05,
+ "loss": 0.772,
+ "step": 85
+ },
+ {
+ "epoch": 2.65625,
+ "eval_loss": 0.7618446350097656,
+ "eval_runtime": 55.3717,
+ "eval_samples_per_second": 3.612,
+ "eval_steps_per_second": 0.235,
+ "step": 85
+ },
+ {
+ "epoch": 2.6875,
+ "grad_norm": 0.3169908538809109,
+ "learning_rate": 2e-05,
+ "loss": 0.8148,
+ "step": 86
+ },
+ {
+ "epoch": 2.6875,
+ "eval_loss": 0.7599577307701111,
+ "eval_runtime": 55.3931,
+ "eval_samples_per_second": 3.611,
+ "eval_steps_per_second": 0.235,
+ "step": 86
+ },
+ {
+ "epoch": 2.71875,
+ "grad_norm": 0.28780549186847426,
+ "learning_rate": 2e-05,
+ "loss": 0.8154,
+ "step": 87
+ },
+ {
+ "epoch": 2.71875,
+ "eval_loss": 0.7583369612693787,
+ "eval_runtime": 55.1679,
+ "eval_samples_per_second": 3.625,
+ "eval_steps_per_second": 0.236,
+ "step": 87
+ },
+ {
+ "epoch": 2.75,
+ "grad_norm": 0.30695250620091474,
+ "learning_rate": 2e-05,
+ "loss": 0.9032,
+ "step": 88
+ },
+ {
+ "epoch": 2.75,
+ "eval_loss": 0.7571613192558289,
+ "eval_runtime": 55.1779,
+ "eval_samples_per_second": 3.625,
+ "eval_steps_per_second": 0.236,
+ "step": 88
+ },
+ {
+ "epoch": 2.78125,
+ "grad_norm": 0.2693887416759828,
+ "learning_rate": 2e-05,
+ "loss": 0.8106,
+ "step": 89
+ },
+ {
+ "epoch": 2.78125,
+ "eval_loss": 0.7566004991531372,
+ "eval_runtime": 55.1107,
+ "eval_samples_per_second": 3.629,
+ "eval_steps_per_second": 0.236,
+ "step": 89
+ },
+ {
+ "epoch": 2.8125,
+ "grad_norm": 0.2887583627563198,
+ "learning_rate": 2e-05,
+ "loss": 0.8518,
+ "step": 90
+ },
+ {
+ "epoch": 2.8125,
+ "eval_loss": 0.7558963298797607,
+ "eval_runtime": 55.2153,
+ "eval_samples_per_second": 3.622,
+ "eval_steps_per_second": 0.235,
+ "step": 90
+ },
+ {
+ "epoch": 2.84375,
+ "grad_norm": 0.3059402168979351,
+ "learning_rate": 2e-05,
+ "loss": 0.7727,
+ "step": 91
+ },
+ {
+ "epoch": 2.84375,
+ "eval_loss": 0.7545350790023804,
+ "eval_runtime": 55.3225,
+ "eval_samples_per_second": 3.615,
+ "eval_steps_per_second": 0.235,
+ "step": 91
+ },
+ {
+ "epoch": 2.875,
+ "grad_norm": 0.3096260477909968,
+ "learning_rate": 2e-05,
+ "loss": 0.8477,
+ "step": 92
+ },
+ {
+ "epoch": 2.875,
+ "eval_loss": 0.7526452541351318,
+ "eval_runtime": 55.4311,
+ "eval_samples_per_second": 3.608,
+ "eval_steps_per_second": 0.235,
+ "step": 92
+ },
+ {
+ "epoch": 2.90625,
+ "grad_norm": 0.31498884686525297,
+ "learning_rate": 2e-05,
+ "loss": 0.7982,
+ "step": 93
+ },
+ {
+ "epoch": 2.90625,
+ "eval_loss": 0.7510760426521301,
+ "eval_runtime": 55.4361,
+ "eval_samples_per_second": 3.608,
+ "eval_steps_per_second": 0.235,
+ "step": 93
+ },
+ {
+ "epoch": 2.9375,
+ "grad_norm": 0.31302830623184313,
+ "learning_rate": 2e-05,
+ "loss": 0.871,
+ "step": 94
+ },
+ {
+ "epoch": 2.9375,
+ "eval_loss": 0.7500898838043213,
+ "eval_runtime": 55.3025,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 94
+ },
+ {
+ "epoch": 2.96875,
+ "grad_norm": 0.3132608568779145,
+ "learning_rate": 2e-05,
+ "loss": 0.8094,
+ "step": 95
+ },
+ {
+ "epoch": 2.96875,
+ "eval_loss": 0.7498895525932312,
+ "eval_runtime": 55.2402,
+ "eval_samples_per_second": 3.621,
+ "eval_steps_per_second": 0.235,
+ "step": 95
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 0.298645350091386,
+ "learning_rate": 2e-05,
+ "loss": 0.7673,
+ "step": 96
+ },
+ {
+ "epoch": 3.0,
+ "eval_loss": 0.7493192553520203,
+ "eval_runtime": 54.8718,
+ "eval_samples_per_second": 3.645,
+ "eval_steps_per_second": 0.237,
+ "step": 96
+ },
+ {
+ "epoch": 3.03125,
+ "grad_norm": 0.34042584783125357,
+ "learning_rate": 2e-05,
+ "loss": 0.7336,
+ "step": 97
+ },
+ {
+ "epoch": 3.03125,
+ "eval_loss": 0.7476670742034912,
+ "eval_runtime": 54.9305,
+ "eval_samples_per_second": 3.641,
+ "eval_steps_per_second": 0.237,
+ "step": 97
+ },
+ {
+ "epoch": 3.0625,
+ "grad_norm": 0.293099043801068,
+ "learning_rate": 2e-05,
+ "loss": 0.8088,
+ "step": 98
+ },
+ {
+ "epoch": 3.0625,
+ "eval_loss": 0.745802640914917,
+ "eval_runtime": 55.2051,
+ "eval_samples_per_second": 3.623,
+ "eval_steps_per_second": 0.235,
+ "step": 98
+ },
+ {
+ "epoch": 3.09375,
+ "grad_norm": 0.3042839507858426,
+ "learning_rate": 2e-05,
+ "loss": 0.787,
+ "step": 99
+ },
+ {
+ "epoch": 3.09375,
+ "eval_loss": 0.7439618110656738,
+ "eval_runtime": 55.0065,
+ "eval_samples_per_second": 3.636,
+ "eval_steps_per_second": 0.236,
+ "step": 99
+ },
+ {
+ "epoch": 3.125,
+ "grad_norm": 0.32992077073227005,
+ "learning_rate": 2e-05,
+ "loss": 0.8296,
+ "step": 100
+ },
+ {
+ "epoch": 3.125,
+ "eval_loss": 0.7424842715263367,
+ "eval_runtime": 55.1254,
+ "eval_samples_per_second": 3.628,
+ "eval_steps_per_second": 0.236,
+ "step": 100
+ },
+ {
+ "epoch": 3.15625,
+ "grad_norm": 0.2798839747424062,
+ "learning_rate": 2e-05,
+ "loss": 0.7642,
+ "step": 101
+ },
+ {
+ "epoch": 3.15625,
+ "eval_loss": 0.7414796948432922,
+ "eval_runtime": 49.183,
+ "eval_samples_per_second": 4.066,
+ "eval_steps_per_second": 0.264,
+ "step": 101
+ },
+ {
+ "epoch": 3.1875,
+ "grad_norm": 0.3046631191964983,
+ "learning_rate": 2e-05,
+ "loss": 0.8203,
+ "step": 102
+ },
+ {
+ "epoch": 3.1875,
+ "eval_loss": 0.7410265207290649,
+ "eval_runtime": 48.1541,
+ "eval_samples_per_second": 4.153,
+ "eval_steps_per_second": 0.27,
+ "step": 102
+ },
+ {
+ "epoch": 3.21875,
+ "grad_norm": 0.3117517214859861,
+ "learning_rate": 2e-05,
+ "loss": 0.8222,
+ "step": 103
+ },
+ {
+ "epoch": 3.21875,
+ "eval_loss": 0.7405675649642944,
+ "eval_runtime": 47.7145,
+ "eval_samples_per_second": 4.192,
+ "eval_steps_per_second": 0.272,
+ "step": 103
+ },
+ {
+ "epoch": 3.25,
+ "grad_norm": 0.3412709249466801,
+ "learning_rate": 2e-05,
+ "loss": 0.7459,
+ "step": 104
+ },
+ {
+ "epoch": 3.25,
+ "eval_loss": 0.7395681738853455,
+ "eval_runtime": 47.5855,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 104
+ },
+ {
+ "epoch": 3.28125,
+ "grad_norm": 0.2917443566507923,
+ "learning_rate": 2e-05,
+ "loss": 0.7849,
+ "step": 105
+ },
+ {
+ "epoch": 3.28125,
+ "eval_loss": 0.7387100458145142,
+ "eval_runtime": 47.6344,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 105
+ },
+ {
+ "epoch": 3.3125,
+ "grad_norm": 0.3054484743574741,
+ "learning_rate": 2e-05,
+ "loss": 0.8354,
+ "step": 106
+ },
+ {
+ "epoch": 3.3125,
+ "eval_loss": 0.7384718060493469,
+ "eval_runtime": 47.8373,
+ "eval_samples_per_second": 4.181,
+ "eval_steps_per_second": 0.272,
+ "step": 106
+ },
+ {
+ "epoch": 3.34375,
+ "grad_norm": 0.34986630381114014,
+ "learning_rate": 2e-05,
+ "loss": 0.7069,
+ "step": 107
+ },
+ {
+ "epoch": 3.34375,
+ "eval_loss": 0.737342357635498,
+ "eval_runtime": 47.5763,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 107
+ },
+ {
+ "epoch": 3.375,
+ "grad_norm": 0.32324403145716496,
+ "learning_rate": 2e-05,
+ "loss": 0.767,
+ "step": 108
+ },
+ {
+ "epoch": 3.375,
+ "eval_loss": 0.7360101938247681,
+ "eval_runtime": 47.5774,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 108
+ },
+ {
+ "epoch": 3.40625,
+ "grad_norm": 0.3795969851258545,
+ "learning_rate": 2e-05,
+ "loss": 0.7556,
+ "step": 109
+ },
+ {
+ "epoch": 3.40625,
+ "eval_loss": 0.7339167594909668,
+ "eval_runtime": 47.5818,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 109
+ },
+ {
+ "epoch": 3.4375,
+ "grad_norm": 0.34401062275458993,
+ "learning_rate": 2e-05,
+ "loss": 0.7494,
+ "step": 110
+ },
+ {
+ "epoch": 3.4375,
+ "eval_loss": 0.7321068644523621,
+ "eval_runtime": 47.7643,
+ "eval_samples_per_second": 4.187,
+ "eval_steps_per_second": 0.272,
+ "step": 110
+ },
+ {
+ "epoch": 3.46875,
+ "grad_norm": 0.3248480010385237,
+ "learning_rate": 2e-05,
+ "loss": 0.8103,
+ "step": 111
+ },
+ {
+ "epoch": 3.46875,
+ "eval_loss": 0.7309197783470154,
+ "eval_runtime": 49.5841,
+ "eval_samples_per_second": 4.034,
+ "eval_steps_per_second": 0.262,
+ "step": 111
+ },
+ {
+ "epoch": 3.5,
+ "grad_norm": 0.3572409124813593,
+ "learning_rate": 2e-05,
+ "loss": 0.7972,
+ "step": 112
+ },
+ {
+ "epoch": 3.5,
+ "eval_loss": 0.7301727533340454,
+ "eval_runtime": 49.3728,
+ "eval_samples_per_second": 4.051,
+ "eval_steps_per_second": 0.263,
+ "step": 112
+ },
+ {
+ "epoch": 3.53125,
+ "grad_norm": 0.37348522775103665,
+ "learning_rate": 2e-05,
+ "loss": 0.88,
+ "step": 113
+ },
+ {
+ "epoch": 3.53125,
+ "eval_loss": 0.7292957305908203,
+ "eval_runtime": 49.2192,
+ "eval_samples_per_second": 4.063,
+ "eval_steps_per_second": 0.264,
+ "step": 113
+ },
+ {
+ "epoch": 3.5625,
+ "grad_norm": 0.37667450960329546,
+ "learning_rate": 2e-05,
+ "loss": 0.7518,
+ "step": 114
+ },
+ {
+ "epoch": 3.5625,
+ "eval_loss": 0.728556215763092,
+ "eval_runtime": 49.0971,
+ "eval_samples_per_second": 4.074,
+ "eval_steps_per_second": 0.265,
+ "step": 114
+ },
+ {
+ "epoch": 3.59375,
+ "grad_norm": 0.3163628607304638,
+ "learning_rate": 2e-05,
+ "loss": 0.7948,
+ "step": 115
+ },
+ {
+ "epoch": 3.59375,
+ "eval_loss": 0.7287828326225281,
+ "eval_runtime": 49.0213,
+ "eval_samples_per_second": 4.08,
+ "eval_steps_per_second": 0.265,
+ "step": 115
+ },
+ {
+ "epoch": 3.625,
+ "grad_norm": 0.3038899302084592,
+ "learning_rate": 2e-05,
+ "loss": 0.7791,
+ "step": 116
+ },
+ {
+ "epoch": 3.625,
+ "eval_loss": 0.7294514179229736,
+ "eval_runtime": 51.9137,
+ "eval_samples_per_second": 3.853,
+ "eval_steps_per_second": 0.25,
+ "step": 116
+ },
+ {
+ "epoch": 3.65625,
+ "grad_norm": 0.3746448663122327,
+ "learning_rate": 2e-05,
+ "loss": 0.7863,
+ "step": 117
+ },
+ {
+ "epoch": 3.65625,
+ "eval_loss": 0.7289304137229919,
+ "eval_runtime": 51.3023,
+ "eval_samples_per_second": 3.898,
+ "eval_steps_per_second": 0.253,
+ "step": 117
+ },
+ {
+ "epoch": 3.6875,
+ "grad_norm": 0.4058937381299434,
+ "learning_rate": 2e-05,
+ "loss": 0.7907,
+ "step": 118
+ },
+ {
+ "epoch": 3.6875,
+ "eval_loss": 0.7281011343002319,
+ "eval_runtime": 50.8635,
+ "eval_samples_per_second": 3.932,
+ "eval_steps_per_second": 0.256,
+ "step": 118
+ },
+ {
+ "epoch": 3.71875,
+ "grad_norm": 0.31608065583227885,
+ "learning_rate": 2e-05,
+ "loss": 0.8348,
+ "step": 119
+ },
+ {
+ "epoch": 3.71875,
+ "eval_loss": 0.7280247211456299,
+ "eval_runtime": 50.4903,
+ "eval_samples_per_second": 3.961,
+ "eval_steps_per_second": 0.257,
+ "step": 119
+ },
+ {
+ "epoch": 3.75,
+ "grad_norm": 0.3375768031046084,
+ "learning_rate": 2e-05,
+ "loss": 0.7783,
+ "step": 120
+ },
+ {
+ "epoch": 3.75,
+ "eval_loss": 0.7281913757324219,
+ "eval_runtime": 50.5906,
+ "eval_samples_per_second": 3.953,
+ "eval_steps_per_second": 0.257,
+ "step": 120
+ },
+ {
+ "epoch": 3.78125,
+ "grad_norm": 0.36047493494859845,
+ "learning_rate": 2e-05,
+ "loss": 0.765,
+ "step": 121
+ },
+ {
+ "epoch": 3.78125,
+ "eval_loss": 0.7269737124443054,
+ "eval_runtime": 53.4722,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 121
+ },
+ {
+ "epoch": 3.8125,
+ "grad_norm": 0.389743860171921,
+ "learning_rate": 2e-05,
+ "loss": 0.8269,
+ "step": 122
+ },
+ {
+ "epoch": 3.8125,
+ "eval_loss": 0.7251996397972107,
+ "eval_runtime": 53.4986,
+ "eval_samples_per_second": 3.738,
+ "eval_steps_per_second": 0.243,
+ "step": 122
+ },
+ {
+ "epoch": 3.84375,
+ "grad_norm": 0.33850935145960215,
+ "learning_rate": 2e-05,
+ "loss": 0.7497,
+ "step": 123
+ },
+ {
+ "epoch": 3.84375,
+ "eval_loss": 0.723595142364502,
+ "eval_runtime": 53.4196,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 123
+ },
+ {
+ "epoch": 3.875,
+ "grad_norm": 0.3166770012114478,
+ "learning_rate": 2e-05,
+ "loss": 0.7648,
+ "step": 124
+ },
+ {
+ "epoch": 3.875,
+ "eval_loss": 0.7223578095436096,
+ "eval_runtime": 52.6143,
+ "eval_samples_per_second": 3.801,
+ "eval_steps_per_second": 0.247,
+ "step": 124
+ },
+ {
+ "epoch": 3.90625,
+ "grad_norm": 0.41948670305268276,
+ "learning_rate": 2e-05,
+ "loss": 0.8306,
+ "step": 125
+ },
+ {
+ "epoch": 3.90625,
+ "eval_loss": 0.7206680774688721,
+ "eval_runtime": 52.3885,
+ "eval_samples_per_second": 3.818,
+ "eval_steps_per_second": 0.248,
+ "step": 125
+ },
+ {
+ "epoch": 3.9375,
+ "grad_norm": 0.35580041105853477,
+ "learning_rate": 2e-05,
+ "loss": 0.7945,
+ "step": 126
+ },
+ {
+ "epoch": 3.9375,
+ "eval_loss": 0.7196171283721924,
+ "eval_runtime": 55.1225,
+ "eval_samples_per_second": 3.628,
+ "eval_steps_per_second": 0.236,
+ "step": 126
+ },
+ {
+ "epoch": 3.96875,
+ "grad_norm": 0.38411890663257114,
+ "learning_rate": 2e-05,
+ "loss": 0.7466,
+ "step": 127
+ },
+ {
+ "epoch": 3.96875,
+ "eval_loss": 0.7188088297843933,
+ "eval_runtime": 55.3068,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 127
+ },
+ {
+ "epoch": 4.0,
+ "grad_norm": 0.3682220575203032,
+ "learning_rate": 2e-05,
+ "loss": 0.6752,
+ "step": 128
+ },
+ {
+ "epoch": 4.0,
+ "eval_loss": 0.7181470990180969,
+ "eval_runtime": 53.9116,
+ "eval_samples_per_second": 3.71,
+ "eval_steps_per_second": 0.241,
+ "step": 128
+ },
+ {
+ "epoch": 4.03125,
+ "grad_norm": 0.34160763542661665,
+ "learning_rate": 2e-05,
+ "loss": 0.7788,
+ "step": 129
+ },
+ {
+ "epoch": 4.03125,
+ "eval_loss": 0.717949390411377,
+ "eval_runtime": 53.8446,
+ "eval_samples_per_second": 3.714,
+ "eval_steps_per_second": 0.241,
+ "step": 129
+ },
+ {
+ "epoch": 4.0625,
+ "grad_norm": 0.35709301353799944,
+ "learning_rate": 2e-05,
+ "loss": 0.8002,
+ "step": 130
+ },
+ {
+ "epoch": 4.0625,
+ "eval_loss": 0.7179380655288696,
+ "eval_runtime": 53.9299,
+ "eval_samples_per_second": 3.709,
+ "eval_steps_per_second": 0.241,
+ "step": 130
+ },
+ {
+ "epoch": 4.09375,
+ "grad_norm": 0.3503147340749238,
+ "learning_rate": 2e-05,
+ "loss": 0.7789,
+ "step": 131
+ },
+ {
+ "epoch": 4.09375,
+ "eval_loss": 0.7180312871932983,
+ "eval_runtime": 53.4091,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 131
+ },
+ {
+ "epoch": 4.125,
+ "grad_norm": 0.3931715546229069,
+ "learning_rate": 2e-05,
+ "loss": 0.762,
+ "step": 132
+ },
+ {
+ "epoch": 4.125,
+ "eval_loss": 0.717825710773468,
+ "eval_runtime": 53.6366,
+ "eval_samples_per_second": 3.729,
+ "eval_steps_per_second": 0.242,
+ "step": 132
+ },
+ {
+ "epoch": 4.15625,
+ "grad_norm": 0.36864033862644363,
+ "learning_rate": 2e-05,
+ "loss": 0.829,
+ "step": 133
+ },
+ {
+ "epoch": 4.15625,
+ "eval_loss": 0.7178698182106018,
+ "eval_runtime": 53.4891,
+ "eval_samples_per_second": 3.739,
+ "eval_steps_per_second": 0.243,
+ "step": 133
+ },
+ {
+ "epoch": 4.1875,
+ "grad_norm": 0.41393587587462155,
+ "learning_rate": 2e-05,
+ "loss": 0.7624,
+ "step": 134
+ },
+ {
+ "epoch": 4.1875,
+ "eval_loss": 0.7181968092918396,
+ "eval_runtime": 53.5395,
+ "eval_samples_per_second": 3.736,
+ "eval_steps_per_second": 0.243,
+ "step": 134
+ },
+ {
+ "epoch": 4.21875,
+ "grad_norm": 0.36727603900023204,
+ "learning_rate": 2e-05,
+ "loss": 0.7572,
+ "step": 135
+ },
+ {
+ "epoch": 4.21875,
+ "eval_loss": 0.7187527418136597,
+ "eval_runtime": 53.4818,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 135
+ },
+ {
+ "epoch": 4.25,
+ "grad_norm": 0.3684078795455007,
+ "learning_rate": 2e-05,
+ "loss": 0.7352,
+ "step": 136
+ },
+ {
+ "epoch": 4.25,
+ "eval_loss": 0.7194793820381165,
+ "eval_runtime": 53.4694,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 136
+ },
+ {
+ "epoch": 4.28125,
+ "grad_norm": 0.42414766562621153,
+ "learning_rate": 2e-05,
+ "loss": 0.7433,
+ "step": 137
+ },
+ {
+ "epoch": 4.28125,
+ "eval_loss": 0.7189603447914124,
+ "eval_runtime": 53.8049,
+ "eval_samples_per_second": 3.717,
+ "eval_steps_per_second": 0.242,
+ "step": 137
+ },
+ {
+ "epoch": 4.3125,
+ "grad_norm": 0.40420796619211563,
+ "learning_rate": 2e-05,
+ "loss": 0.7466,
+ "step": 138
+ },
+ {
+ "epoch": 4.3125,
+ "eval_loss": 0.7173956036567688,
+ "eval_runtime": 53.4014,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 138
+ },
+ {
+ "epoch": 4.34375,
+ "grad_norm": 0.36419740641344456,
+ "learning_rate": 2e-05,
+ "loss": 0.7045,
+ "step": 139
+ },
+ {
+ "epoch": 4.34375,
+ "eval_loss": 0.7153105139732361,
+ "eval_runtime": 53.285,
+ "eval_samples_per_second": 3.753,
+ "eval_steps_per_second": 0.244,
+ "step": 139
+ },
+ {
+ "epoch": 4.375,
+ "grad_norm": 0.384927357409491,
+ "learning_rate": 2e-05,
+ "loss": 0.7437,
+ "step": 140
+ },
+ {
+ "epoch": 4.375,
+ "eval_loss": 0.7135314345359802,
+ "eval_runtime": 53.4056,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 140
+ },
+ {
+ "epoch": 4.40625,
+ "grad_norm": 0.37218579680263697,
+ "learning_rate": 2e-05,
+ "loss": 0.7693,
+ "step": 141
+ },
+ {
+ "epoch": 4.40625,
+ "eval_loss": 0.7120725512504578,
+ "eval_runtime": 53.5467,
+ "eval_samples_per_second": 3.735,
+ "eval_steps_per_second": 0.243,
+ "step": 141
+ },
+ {
+ "epoch": 4.4375,
+ "grad_norm": 0.38541382926033946,
+ "learning_rate": 2e-05,
+ "loss": 0.708,
+ "step": 142
+ },
+ {
+ "epoch": 4.4375,
+ "eval_loss": 0.7110380530357361,
+ "eval_runtime": 53.4119,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 142
+ },
+ {
+ "epoch": 4.46875,
+ "grad_norm": 0.4028726453247759,
+ "learning_rate": 2e-05,
+ "loss": 0.7263,
+ "step": 143
+ },
+ {
+ "epoch": 4.46875,
+ "eval_loss": 0.7100683450698853,
+ "eval_runtime": 53.4337,
+ "eval_samples_per_second": 3.743,
+ "eval_steps_per_second": 0.243,
+ "step": 143
+ },
+ {
+ "epoch": 4.5,
+ "grad_norm": 0.3736204162232246,
+ "learning_rate": 2e-05,
+ "loss": 0.698,
+ "step": 144
+ },
+ {
+ "epoch": 4.5,
+ "eval_loss": 0.7093971371650696,
+ "eval_runtime": 53.4582,
+ "eval_samples_per_second": 3.741,
+ "eval_steps_per_second": 0.243,
+ "step": 144
+ },
+ {
+ "epoch": 4.53125,
+ "grad_norm": 0.4179284798304916,
+ "learning_rate": 2e-05,
+ "loss": 0.7611,
+ "step": 145
+ },
+ {
+ "epoch": 4.53125,
+ "eval_loss": 0.7089446783065796,
+ "eval_runtime": 53.4752,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 145
+ },
+ {
+ "epoch": 4.5625,
+ "grad_norm": 0.4038858950888911,
+ "learning_rate": 2e-05,
+ "loss": 0.6652,
+ "step": 146
+ },
+ {
+ "epoch": 4.5625,
+ "eval_loss": 0.7089542150497437,
+ "eval_runtime": 53.4741,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 146
+ },
+ {
+ "epoch": 4.59375,
+ "grad_norm": 0.41740068710674544,
+ "learning_rate": 2e-05,
+ "loss": 0.7319,
+ "step": 147
+ },
+ {
+ "epoch": 4.59375,
+ "eval_loss": 0.7090431451797485,
+ "eval_runtime": 53.2419,
+ "eval_samples_per_second": 3.756,
+ "eval_steps_per_second": 0.244,
+ "step": 147
+ },
+ {
+ "epoch": 4.625,
+ "grad_norm": 0.4288335811568808,
+ "learning_rate": 2e-05,
+ "loss": 0.6837,
+ "step": 148
+ },
+ {
+ "epoch": 4.625,
+ "eval_loss": 0.7088204026222229,
+ "eval_runtime": 53.3614,
+ "eval_samples_per_second": 3.748,
+ "eval_steps_per_second": 0.244,
+ "step": 148
+ },
+ {
+ "epoch": 4.65625,
+ "grad_norm": 0.399955010119186,
+ "learning_rate": 2e-05,
+ "loss": 0.7989,
+ "step": 149
+ },
+ {
+ "epoch": 4.65625,
+ "eval_loss": 0.7084855437278748,
+ "eval_runtime": 53.4923,
+ "eval_samples_per_second": 3.739,
+ "eval_steps_per_second": 0.243,
+ "step": 149
+ },
+ {
+ "epoch": 4.6875,
+ "grad_norm": 0.41794643164255846,
+ "learning_rate": 2e-05,
+ "loss": 0.7194,
+ "step": 150
+ },
+ {
+ "epoch": 4.6875,
+ "eval_loss": 0.7080708146095276,
+ "eval_runtime": 53.639,
+ "eval_samples_per_second": 3.729,
+ "eval_steps_per_second": 0.242,
+ "step": 150
+ },
+ {
+ "epoch": 4.71875,
+ "grad_norm": 0.40953367303148197,
+ "learning_rate": 2e-05,
+ "loss": 0.7354,
+ "step": 151
+ },
+ {
+ "epoch": 4.71875,
+ "eval_loss": 0.7077429890632629,
+ "eval_runtime": 53.3837,
+ "eval_samples_per_second": 3.746,
+ "eval_steps_per_second": 0.244,
+ "step": 151
+ },
+ {
+ "epoch": 4.75,
+ "grad_norm": 0.5012282841513718,
+ "learning_rate": 2e-05,
+ "loss": 0.7662,
+ "step": 152
+ },
+ {
+ "epoch": 4.75,
+ "eval_loss": 0.7064151167869568,
+ "eval_runtime": 53.3549,
+ "eval_samples_per_second": 3.748,
+ "eval_steps_per_second": 0.244,
+ "step": 152
+ },
+ {
+ "epoch": 4.78125,
+ "grad_norm": 0.4210784420989087,
+ "learning_rate": 2e-05,
+ "loss": 0.7133,
+ "step": 153
+ },
+ {
+ "epoch": 4.78125,
+ "eval_loss": 0.7052726745605469,
+ "eval_runtime": 53.5059,
+ "eval_samples_per_second": 3.738,
+ "eval_steps_per_second": 0.243,
+ "step": 153
+ },
+ {
+ "epoch": 4.8125,
+ "grad_norm": 0.43520348530514996,
+ "learning_rate": 2e-05,
+ "loss": 0.729,
+ "step": 154
+ },
+ {
+ "epoch": 4.8125,
+ "eval_loss": 0.7045274972915649,
+ "eval_runtime": 53.8352,
+ "eval_samples_per_second": 3.715,
+ "eval_steps_per_second": 0.241,
+ "step": 154
+ },
+ {
+ "epoch": 4.84375,
+ "grad_norm": 0.4287647569802656,
+ "learning_rate": 2e-05,
+ "loss": 0.6727,
+ "step": 155
+ },
+ {
+ "epoch": 4.84375,
+ "eval_loss": 0.7041358947753906,
+ "eval_runtime": 53.7435,
+ "eval_samples_per_second": 3.721,
+ "eval_steps_per_second": 0.242,
+ "step": 155
+ },
+ {
+ "epoch": 4.875,
+ "grad_norm": 0.41883715320456333,
+ "learning_rate": 2e-05,
+ "loss": 0.7755,
+ "step": 156
+ },
+ {
+ "epoch": 4.875,
+ "eval_loss": 0.7037128210067749,
+ "eval_runtime": 53.8035,
+ "eval_samples_per_second": 3.717,
+ "eval_steps_per_second": 0.242,
+ "step": 156
+ },
+ {
+ "epoch": 4.90625,
+ "grad_norm": 0.40617584505395354,
+ "learning_rate": 2e-05,
+ "loss": 0.7776,
+ "step": 157
+ },
+ {
+ "epoch": 4.90625,
+ "eval_loss": 0.703965425491333,
+ "eval_runtime": 53.8731,
+ "eval_samples_per_second": 3.712,
+ "eval_steps_per_second": 0.241,
+ "step": 157
+ },
+ {
+ "epoch": 4.9375,
+ "grad_norm": 0.4085802225532245,
+ "learning_rate": 2e-05,
+ "loss": 0.7628,
+ "step": 158
+ },
+ {
+ "epoch": 4.9375,
+ "eval_loss": 0.7040860056877136,
+ "eval_runtime": 53.9059,
+ "eval_samples_per_second": 3.71,
+ "eval_steps_per_second": 0.241,
+ "step": 158
+ },
+ {
+ "epoch": 4.96875,
+ "grad_norm": 0.418039298119887,
+ "learning_rate": 2e-05,
+ "loss": 0.7221,
+ "step": 159
+ },
+ {
+ "epoch": 4.96875,
+ "eval_loss": 0.7039948105812073,
+ "eval_runtime": 53.7323,
+ "eval_samples_per_second": 3.722,
+ "eval_steps_per_second": 0.242,
+ "step": 159
+ },
+ {
+ "epoch": 5.0,
+ "grad_norm": 0.46118870048713073,
+ "learning_rate": 2e-05,
+ "loss": 0.7029,
+ "step": 160
+ },
+ {
+ "epoch": 5.0,
+ "eval_loss": 0.703814685344696,
+ "eval_runtime": 53.8975,
+ "eval_samples_per_second": 3.711,
+ "eval_steps_per_second": 0.241,
+ "step": 160
+ },
+ {
+ "epoch": 5.03125,
+ "grad_norm": 0.431474386110294,
+ "learning_rate": 2e-05,
+ "loss": 0.6772,
+ "step": 161
+ },
+ {
+ "epoch": 5.03125,
+ "eval_loss": 0.7034456133842468,
+ "eval_runtime": 51.1105,
+ "eval_samples_per_second": 3.913,
+ "eval_steps_per_second": 0.254,
+ "step": 161
+ },
+ {
+ "epoch": 5.0625,
+ "grad_norm": 0.39618929325750435,
+ "learning_rate": 2e-05,
+ "loss": 0.8219,
+ "step": 162
+ },
+ {
+ "epoch": 5.0625,
+ "eval_loss": 0.7042189240455627,
+ "eval_runtime": 47.2927,
+ "eval_samples_per_second": 4.229,
+ "eval_steps_per_second": 0.275,
+ "step": 162
+ },
+ {
+ "epoch": 5.09375,
+ "grad_norm": 0.4489132713249424,
+ "learning_rate": 2e-05,
+ "loss": 0.6387,
+ "step": 163
+ },
+ {
+ "epoch": 5.09375,
+ "eval_loss": 0.7061256170272827,
+ "eval_runtime": 47.387,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 163
+ },
+ {
+ "epoch": 5.125,
+ "grad_norm": 0.5100329637159183,
+ "learning_rate": 2e-05,
+ "loss": 0.7677,
+ "step": 164
+ },
+ {
+ "epoch": 5.125,
+ "eval_loss": 0.708121657371521,
+ "eval_runtime": 47.3311,
+ "eval_samples_per_second": 4.226,
+ "eval_steps_per_second": 0.275,
+ "step": 164
+ },
+ {
+ "epoch": 5.15625,
+ "grad_norm": 0.525511631981176,
+ "learning_rate": 2e-05,
+ "loss": 0.5956,
+ "step": 165
+ },
+ {
+ "epoch": 5.15625,
+ "eval_loss": 0.7091134786605835,
+ "eval_runtime": 47.2978,
+ "eval_samples_per_second": 4.229,
+ "eval_steps_per_second": 0.275,
+ "step": 165
+ },
+ {
+ "epoch": 5.1875,
+ "grad_norm": 0.534675354231597,
+ "learning_rate": 2e-05,
+ "loss": 0.7097,
+ "step": 166
+ },
+ {
+ "epoch": 5.1875,
+ "eval_loss": 0.7097848653793335,
+ "eval_runtime": 47.4095,
+ "eval_samples_per_second": 4.219,
+ "eval_steps_per_second": 0.274,
+ "step": 166
+ },
+ {
+ "epoch": 5.21875,
+ "grad_norm": 0.47286903698857446,
+ "learning_rate": 2e-05,
+ "loss": 0.7371,
+ "step": 167
+ },
+ {
+ "epoch": 5.21875,
+ "eval_loss": 0.7090296745300293,
+ "eval_runtime": 47.4487,
+ "eval_samples_per_second": 4.215,
+ "eval_steps_per_second": 0.274,
+ "step": 167
+ },
+ {
+ "epoch": 5.25,
+ "grad_norm": 0.4734705066820788,
+ "learning_rate": 2e-05,
+ "loss": 0.7652,
+ "step": 168
+ },
+ {
+ "epoch": 5.25,
+ "eval_loss": 0.7079525589942932,
+ "eval_runtime": 47.4101,
+ "eval_samples_per_second": 4.219,
+ "eval_steps_per_second": 0.274,
+ "step": 168
+ },
+ {
+ "epoch": 5.28125,
+ "grad_norm": 0.46209764763985184,
+ "learning_rate": 2e-05,
+ "loss": 0.6852,
+ "step": 169
+ },
+ {
+ "epoch": 5.28125,
+ "eval_loss": 0.7072803974151611,
+ "eval_runtime": 47.3704,
+ "eval_samples_per_second": 4.222,
+ "eval_steps_per_second": 0.274,
+ "step": 169
+ },
+ {
+ "epoch": 5.3125,
+ "grad_norm": 0.4828284708486433,
+ "learning_rate": 2e-05,
+ "loss": 0.6609,
+ "step": 170
+ },
+ {
+ "epoch": 5.3125,
+ "eval_loss": 0.7068901062011719,
+ "eval_runtime": 47.425,
+ "eval_samples_per_second": 4.217,
+ "eval_steps_per_second": 0.274,
+ "step": 170
+ },
+ {
+ "epoch": 5.34375,
+ "grad_norm": 0.5230116179180577,
+ "learning_rate": 2e-05,
+ "loss": 0.6872,
+ "step": 171
+ },
+ {
+ "epoch": 5.34375,
+ "eval_loss": 0.7058187127113342,
+ "eval_runtime": 47.5711,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 171
+ },
+ {
+ "epoch": 5.375,
+ "grad_norm": 0.48081340678536255,
+ "learning_rate": 2e-05,
+ "loss": 0.7694,
+ "step": 172
+ },
+ {
+ "epoch": 5.375,
+ "eval_loss": 0.7044984698295593,
+ "eval_runtime": 47.4233,
+ "eval_samples_per_second": 4.217,
+ "eval_steps_per_second": 0.274,
+ "step": 172
+ },
+ {
+ "epoch": 5.40625,
+ "grad_norm": 0.4787525602476421,
+ "learning_rate": 2e-05,
+ "loss": 0.7342,
+ "step": 173
+ },
+ {
+ "epoch": 5.40625,
+ "eval_loss": 0.7032212018966675,
+ "eval_runtime": 47.3534,
+ "eval_samples_per_second": 4.224,
+ "eval_steps_per_second": 0.275,
+ "step": 173
+ },
+ {
+ "epoch": 5.4375,
+ "grad_norm": 0.4871847582306217,
+ "learning_rate": 2e-05,
+ "loss": 0.7562,
+ "step": 174
+ },
+ {
+ "epoch": 5.4375,
+ "eval_loss": 0.7019696235656738,
+ "eval_runtime": 47.382,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 174
+ },
+ {
+ "epoch": 5.46875,
+ "grad_norm": 0.47999745025553603,
+ "learning_rate": 2e-05,
+ "loss": 0.7534,
+ "step": 175
+ },
+ {
+ "epoch": 5.46875,
+ "eval_loss": 0.7014529705047607,
+ "eval_runtime": 47.4435,
+ "eval_samples_per_second": 4.216,
+ "eval_steps_per_second": 0.274,
+ "step": 175
+ },
+ {
+ "epoch": 5.5,
+ "grad_norm": 0.5168030891996357,
+ "learning_rate": 2e-05,
+ "loss": 0.707,
+ "step": 176
+ },
+ {
+ "epoch": 5.5,
+ "eval_loss": 0.6993884444236755,
+ "eval_runtime": 47.4943,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 176
+ },
+ {
+ "epoch": 5.53125,
+ "grad_norm": 0.536450206978984,
+ "learning_rate": 2e-05,
+ "loss": 0.7318,
+ "step": 177
+ },
+ {
+ "epoch": 5.53125,
+ "eval_loss": 0.6971662640571594,
+ "eval_runtime": 47.4193,
+ "eval_samples_per_second": 4.218,
+ "eval_steps_per_second": 0.274,
+ "step": 177
+ },
+ {
+ "epoch": 5.5625,
+ "grad_norm": 0.45352543205020696,
+ "learning_rate": 2e-05,
+ "loss": 0.7421,
+ "step": 178
+ },
+ {
+ "epoch": 5.5625,
+ "eval_loss": 0.6962605118751526,
+ "eval_runtime": 47.3798,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 178
+ },
+ {
+ "epoch": 5.59375,
+ "grad_norm": 0.5054883443109318,
+ "learning_rate": 2e-05,
+ "loss": 0.6668,
+ "step": 179
+ },
+ {
+ "epoch": 5.59375,
+ "eval_loss": 0.6970357298851013,
+ "eval_runtime": 47.3311,
+ "eval_samples_per_second": 4.226,
+ "eval_steps_per_second": 0.275,
+ "step": 179
+ },
+ {
+ "epoch": 5.625,
+ "grad_norm": 0.49584660418833293,
+ "learning_rate": 2e-05,
+ "loss": 0.6548,
+ "step": 180
+ },
+ {
+ "epoch": 5.625,
+ "eval_loss": 0.6980059146881104,
+ "eval_runtime": 47.299,
+ "eval_samples_per_second": 4.228,
+ "eval_steps_per_second": 0.275,
+ "step": 180
+ },
+ {
+ "epoch": 5.65625,
+ "grad_norm": 0.5114381326491793,
+ "learning_rate": 2e-05,
+ "loss": 0.6691,
+ "step": 181
+ },
+ {
+ "epoch": 5.65625,
+ "eval_loss": 0.6995040774345398,
+ "eval_runtime": 47.3887,
+ "eval_samples_per_second": 4.22,
+ "eval_steps_per_second": 0.274,
+ "step": 181
+ },
+ {
+ "epoch": 5.6875,
+ "grad_norm": 0.48550125668870825,
+ "learning_rate": 2e-05,
+ "loss": 0.6525,
+ "step": 182
+ },
+ {
+ "epoch": 5.6875,
+ "eval_loss": 0.7020326256752014,
+ "eval_runtime": 47.3838,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 182
+ },
+ {
+ "epoch": 5.71875,
+ "grad_norm": 0.5860847796671736,
+ "learning_rate": 2e-05,
+ "loss": 0.674,
+ "step": 183
+ },
+ {
+ "epoch": 5.71875,
+ "eval_loss": 0.7027825713157654,
+ "eval_runtime": 47.3875,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 183
+ },
+ {
+ "epoch": 5.75,
+ "grad_norm": 0.5535582209035479,
+ "learning_rate": 2e-05,
+ "loss": 0.6643,
+ "step": 184
+ },
+ {
+ "epoch": 5.75,
+ "eval_loss": 0.7025408148765564,
+ "eval_runtime": 47.5534,
+ "eval_samples_per_second": 4.206,
+ "eval_steps_per_second": 0.273,
+ "step": 184
+ },
+ {
+ "epoch": 5.78125,
+ "grad_norm": 0.5443574176405931,
+ "learning_rate": 2e-05,
+ "loss": 0.709,
+ "step": 185
+ },
+ {
+ "epoch": 5.78125,
+ "eval_loss": 0.7007840871810913,
+ "eval_runtime": 47.4469,
+ "eval_samples_per_second": 4.215,
+ "eval_steps_per_second": 0.274,
+ "step": 185
+ },
+ {
+ "epoch": 5.8125,
+ "grad_norm": 0.563830259704143,
+ "learning_rate": 2e-05,
+ "loss": 0.6884,
+ "step": 186
+ },
+ {
+ "epoch": 5.8125,
+ "eval_loss": 0.6979361176490784,
+ "eval_runtime": 49.1203,
+ "eval_samples_per_second": 4.072,
+ "eval_steps_per_second": 0.265,
+ "step": 186
+ },
+ {
+ "epoch": 5.84375,
+ "grad_norm": 0.5094956892765212,
+ "learning_rate": 2e-05,
+ "loss": 0.7318,
+ "step": 187
+ },
+ {
+ "epoch": 5.84375,
+ "eval_loss": 0.6962587237358093,
+ "eval_runtime": 49.1831,
+ "eval_samples_per_second": 4.066,
+ "eval_steps_per_second": 0.264,
+ "step": 187
+ },
+ {
+ "epoch": 5.875,
+ "grad_norm": 0.5264819980742595,
+ "learning_rate": 2e-05,
+ "loss": 0.6746,
+ "step": 188
+ },
+ {
+ "epoch": 5.875,
+ "eval_loss": 0.694776713848114,
+ "eval_runtime": 49.1994,
+ "eval_samples_per_second": 4.065,
+ "eval_steps_per_second": 0.264,
+ "step": 188
+ },
+ {
+ "epoch": 5.90625,
+ "grad_norm": 0.4737429304023209,
+ "learning_rate": 2e-05,
+ "loss": 0.664,
+ "step": 189
+ },
+ {
+ "epoch": 5.90625,
+ "eval_loss": 0.6939517855644226,
+ "eval_runtime": 49.2438,
+ "eval_samples_per_second": 4.061,
+ "eval_steps_per_second": 0.264,
+ "step": 189
+ },
+ {
+ "epoch": 5.9375,
+ "grad_norm": 0.494163934813738,
+ "learning_rate": 2e-05,
+ "loss": 0.6978,
+ "step": 190
+ },
+ {
+ "epoch": 5.9375,
+ "eval_loss": 0.6933834552764893,
+ "eval_runtime": 49.3494,
+ "eval_samples_per_second": 4.053,
+ "eval_steps_per_second": 0.263,
+ "step": 190
+ },
+ {
+ "epoch": 5.96875,
+ "grad_norm": 0.4945972278087299,
+ "learning_rate": 2e-05,
+ "loss": 0.6909,
+ "step": 191
+ },
+ {
+ "epoch": 5.96875,
+ "eval_loss": 0.6924250721931458,
+ "eval_runtime": 50.3255,
+ "eval_samples_per_second": 3.974,
+ "eval_steps_per_second": 0.258,
+ "step": 191
+ },
+ {
+ "epoch": 6.0,
+ "grad_norm": 0.48872556688745233,
+ "learning_rate": 2e-05,
+ "loss": 0.6622,
+ "step": 192
+ },
+ {
+ "epoch": 6.0,
+ "eval_loss": 0.6922193765640259,
+ "eval_runtime": 50.4561,
+ "eval_samples_per_second": 3.964,
+ "eval_steps_per_second": 0.258,
+ "step": 192
+ },
+ {
+ "epoch": 6.03125,
+ "grad_norm": 0.5013452255378538,
+ "learning_rate": 2e-05,
+ "loss": 0.7458,
+ "step": 193
+ },
+ {
+ "epoch": 6.03125,
+ "eval_loss": 0.6931161284446716,
+ "eval_runtime": 50.5049,
+ "eval_samples_per_second": 3.96,
+ "eval_steps_per_second": 0.257,
+ "step": 193
+ },
+ {
+ "epoch": 6.0625,
+ "grad_norm": 0.48271161232093784,
+ "learning_rate": 2e-05,
+ "loss": 0.7171,
+ "step": 194
+ },
+ {
+ "epoch": 6.0625,
+ "eval_loss": 0.6959040760993958,
+ "eval_runtime": 50.2441,
+ "eval_samples_per_second": 3.981,
+ "eval_steps_per_second": 0.259,
+ "step": 194
+ },
+ {
+ "epoch": 6.09375,
+ "grad_norm": 0.5414562703154852,
+ "learning_rate": 2e-05,
+ "loss": 0.6419,
+ "step": 195
+ },
+ {
+ "epoch": 6.09375,
+ "eval_loss": 0.7000604271888733,
+ "eval_runtime": 50.4261,
+ "eval_samples_per_second": 3.966,
+ "eval_steps_per_second": 0.258,
+ "step": 195
+ },
+ {
+ "epoch": 6.125,
+ "grad_norm": 0.5074661247335385,
+ "learning_rate": 2e-05,
+ "loss": 0.6881,
+ "step": 196
+ },
+ {
+ "epoch": 6.125,
+ "eval_loss": 0.7039622664451599,
+ "eval_runtime": 51.5214,
+ "eval_samples_per_second": 3.882,
+ "eval_steps_per_second": 0.252,
+ "step": 196
+ },
+ {
+ "epoch": 6.15625,
+ "grad_norm": 0.5603468534764365,
+ "learning_rate": 2e-05,
+ "loss": 0.7085,
+ "step": 197
+ },
+ {
+ "epoch": 6.15625,
+ "eval_loss": 0.7055023312568665,
+ "eval_runtime": 51.7102,
+ "eval_samples_per_second": 3.868,
+ "eval_steps_per_second": 0.251,
+ "step": 197
+ },
+ {
+ "epoch": 6.1875,
+ "grad_norm": 0.5992190802422799,
+ "learning_rate": 2e-05,
+ "loss": 0.7614,
+ "step": 198
+ },
+ {
+ "epoch": 6.1875,
+ "eval_loss": 0.7046856880187988,
+ "eval_runtime": 51.5464,
+ "eval_samples_per_second": 3.88,
+ "eval_steps_per_second": 0.252,
+ "step": 198
+ },
+ {
+ "epoch": 6.21875,
+ "grad_norm": 0.6293684167527106,
+ "learning_rate": 2e-05,
+ "loss": 0.6435,
+ "step": 199
+ },
+ {
+ "epoch": 6.21875,
+ "eval_loss": 0.7021151781082153,
+ "eval_runtime": 51.5328,
+ "eval_samples_per_second": 3.881,
+ "eval_steps_per_second": 0.252,
+ "step": 199
+ },
+ {
+ "epoch": 6.25,
+ "grad_norm": 0.591265449241434,
+ "learning_rate": 2e-05,
+ "loss": 0.688,
+ "step": 200
+ },
+ {
+ "epoch": 6.25,
+ "eval_loss": 0.7002359628677368,
+ "eval_runtime": 51.5812,
+ "eval_samples_per_second": 3.877,
+ "eval_steps_per_second": 0.252,
+ "step": 200
+ },
+ {
+ "epoch": 6.28125,
+ "grad_norm": 0.543141536526749,
+ "learning_rate": 2e-05,
+ "loss": 0.7027,
+ "step": 201
+ },
+ {
+ "epoch": 6.28125,
+ "eval_loss": 0.6986366510391235,
+ "eval_runtime": 52.6956,
+ "eval_samples_per_second": 3.795,
+ "eval_steps_per_second": 0.247,
+ "step": 201
+ },
+ {
+ "epoch": 6.3125,
+ "grad_norm": 0.5679656300203245,
+ "learning_rate": 2e-05,
+ "loss": 0.625,
+ "step": 202
+ },
+ {
+ "epoch": 6.3125,
+ "eval_loss": 0.698679506778717,
+ "eval_runtime": 52.5102,
+ "eval_samples_per_second": 3.809,
+ "eval_steps_per_second": 0.248,
+ "step": 202
+ },
+ {
+ "epoch": 6.34375,
+ "grad_norm": 0.5285839896523021,
+ "learning_rate": 2e-05,
+ "loss": 0.7687,
+ "step": 203
+ },
+ {
+ "epoch": 6.34375,
+ "eval_loss": 0.7005956768989563,
+ "eval_runtime": 52.6067,
+ "eval_samples_per_second": 3.802,
+ "eval_steps_per_second": 0.247,
+ "step": 203
+ },
+ {
+ "epoch": 6.375,
+ "grad_norm": 0.6512964945211068,
+ "learning_rate": 2e-05,
+ "loss": 0.623,
+ "step": 204
+ },
+ {
+ "epoch": 6.375,
+ "eval_loss": 0.7013595104217529,
+ "eval_runtime": 52.5428,
+ "eval_samples_per_second": 3.806,
+ "eval_steps_per_second": 0.247,
+ "step": 204
+ },
+ {
+ "epoch": 6.40625,
+ "grad_norm": 0.5295248631519638,
+ "learning_rate": 2e-05,
+ "loss": 0.5941,
+ "step": 205
+ },
+ {
+ "epoch": 6.40625,
+ "eval_loss": 0.7016547322273254,
+ "eval_runtime": 52.6142,
+ "eval_samples_per_second": 3.801,
+ "eval_steps_per_second": 0.247,
+ "step": 205
+ },
+ {
+ "epoch": 6.4375,
+ "grad_norm": 0.6134157701434021,
+ "learning_rate": 2e-05,
+ "loss": 0.6506,
+ "step": 206
+ },
+ {
+ "epoch": 6.4375,
+ "eval_loss": 0.7009623646736145,
+ "eval_runtime": 52.1942,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 206
+ },
+ {
+ "epoch": 6.46875,
+ "grad_norm": 0.57886797614996,
+ "learning_rate": 2e-05,
+ "loss": 0.6983,
+ "step": 207
+ },
+ {
+ "epoch": 6.46875,
+ "eval_loss": 0.6988092064857483,
+ "eval_runtime": 52.2577,
+ "eval_samples_per_second": 3.827,
+ "eval_steps_per_second": 0.249,
+ "step": 207
+ },
+ {
+ "epoch": 6.5,
+ "grad_norm": 0.5593482836944472,
+ "learning_rate": 2e-05,
+ "loss": 0.6348,
+ "step": 208
+ },
+ {
+ "epoch": 6.5,
+ "eval_loss": 0.698823094367981,
+ "eval_runtime": 52.2296,
+ "eval_samples_per_second": 3.829,
+ "eval_steps_per_second": 0.249,
+ "step": 208
+ },
+ {
+ "epoch": 6.53125,
+ "grad_norm": 0.662802162179718,
+ "learning_rate": 2e-05,
+ "loss": 0.6206,
+ "step": 209
+ },
+ {
+ "epoch": 6.53125,
+ "eval_loss": 0.6990167498588562,
+ "eval_runtime": 52.4316,
+ "eval_samples_per_second": 3.814,
+ "eval_steps_per_second": 0.248,
+ "step": 209
+ },
+ {
+ "epoch": 6.5625,
+ "grad_norm": 0.6874374231122908,
+ "learning_rate": 2e-05,
+ "loss": 0.6033,
+ "step": 210
+ },
+ {
+ "epoch": 6.5625,
+ "eval_loss": 0.699796736240387,
+ "eval_runtime": 52.3193,
+ "eval_samples_per_second": 3.823,
+ "eval_steps_per_second": 0.248,
+ "step": 210
+ },
+ {
+ "epoch": 6.59375,
+ "grad_norm": 0.6625766736772473,
+ "learning_rate": 2e-05,
+ "loss": 0.6398,
+ "step": 211
+ },
+ {
+ "epoch": 6.59375,
+ "eval_loss": 0.6989737153053284,
+ "eval_runtime": 52.1885,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 211
+ },
+ {
+ "epoch": 6.625,
+ "grad_norm": 0.6563419096027812,
+ "learning_rate": 2e-05,
+ "loss": 0.6119,
+ "step": 212
+ },
+ {
+ "epoch": 6.625,
+ "eval_loss": 0.6973609924316406,
+ "eval_runtime": 52.1628,
+ "eval_samples_per_second": 3.834,
+ "eval_steps_per_second": 0.249,
+ "step": 212
+ },
+ {
+ "epoch": 6.65625,
+ "grad_norm": 0.5796353226697397,
+ "learning_rate": 2e-05,
+ "loss": 0.7041,
+ "step": 213
+ },
+ {
+ "epoch": 6.65625,
+ "eval_loss": 0.6957942247390747,
+ "eval_runtime": 52.2028,
+ "eval_samples_per_second": 3.831,
+ "eval_steps_per_second": 0.249,
+ "step": 213
+ },
+ {
+ "epoch": 6.6875,
+ "grad_norm": 0.5711947110504899,
+ "learning_rate": 2e-05,
+ "loss": 0.6465,
+ "step": 214
+ },
+ {
+ "epoch": 6.6875,
+ "eval_loss": 0.696739673614502,
+ "eval_runtime": 52.1849,
+ "eval_samples_per_second": 3.833,
+ "eval_steps_per_second": 0.249,
+ "step": 214
+ },
+ {
+ "epoch": 6.71875,
+ "grad_norm": 0.6619502413653232,
+ "learning_rate": 2e-05,
+ "loss": 0.6563,
+ "step": 215
+ },
+ {
+ "epoch": 6.71875,
+ "eval_loss": 0.6960940361022949,
+ "eval_runtime": 52.0996,
+ "eval_samples_per_second": 3.839,
+ "eval_steps_per_second": 0.25,
+ "step": 215
+ },
+ {
+ "epoch": 6.75,
+ "grad_norm": 0.6587126256919645,
+ "learning_rate": 2e-05,
+ "loss": 0.6505,
+ "step": 216
+ },
+ {
+ "epoch": 6.75,
+ "eval_loss": 0.6959022283554077,
+ "eval_runtime": 52.1062,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 216
+ },
+ {
+ "epoch": 6.78125,
+ "grad_norm": 0.648164277941964,
+ "learning_rate": 2e-05,
+ "loss": 0.5969,
+ "step": 217
+ },
+ {
+ "epoch": 6.78125,
+ "eval_loss": 0.6999121308326721,
+ "eval_runtime": 51.9356,
+ "eval_samples_per_second": 3.851,
+ "eval_steps_per_second": 0.25,
+ "step": 217
+ },
+ {
+ "epoch": 6.8125,
+ "grad_norm": 0.6595860789738482,
+ "learning_rate": 2e-05,
+ "loss": 0.5945,
+ "step": 218
+ },
+ {
+ "epoch": 6.8125,
+ "eval_loss": 0.7028067111968994,
+ "eval_runtime": 52.2232,
+ "eval_samples_per_second": 3.83,
+ "eval_steps_per_second": 0.249,
+ "step": 218
+ },
+ {
+ "epoch": 6.84375,
+ "grad_norm": 0.7116894779822719,
+ "learning_rate": 2e-05,
+ "loss": 0.7027,
+ "step": 219
+ },
+ {
+ "epoch": 6.84375,
+ "eval_loss": 0.7035638689994812,
+ "eval_runtime": 52.1471,
+ "eval_samples_per_second": 3.835,
+ "eval_steps_per_second": 0.249,
+ "step": 219
+ },
+ {
+ "epoch": 6.875,
+ "grad_norm": 0.7581142336087988,
+ "learning_rate": 2e-05,
+ "loss": 0.7171,
+ "step": 220
+ },
+ {
+ "epoch": 6.875,
+ "eval_loss": 0.6981176733970642,
+ "eval_runtime": 52.1366,
+ "eval_samples_per_second": 3.836,
+ "eval_steps_per_second": 0.249,
+ "step": 220
+ },
+ {
+ "epoch": 6.90625,
+ "grad_norm": 0.6261292745909233,
+ "learning_rate": 2e-05,
+ "loss": 0.658,
+ "step": 221
+ },
+ {
+ "epoch": 6.90625,
+ "eval_loss": 0.6939045786857605,
+ "eval_runtime": 52.2211,
+ "eval_samples_per_second": 3.83,
+ "eval_steps_per_second": 0.249,
+ "step": 221
+ },
+ {
+ "epoch": 6.9375,
+ "grad_norm": 0.7256427809370966,
+ "learning_rate": 2e-05,
+ "loss": 0.6576,
+ "step": 222
+ },
+ {
+ "epoch": 6.9375,
+ "eval_loss": 0.6904327273368835,
+ "eval_runtime": 52.1829,
+ "eval_samples_per_second": 3.833,
+ "eval_steps_per_second": 0.249,
+ "step": 222
+ },
+ {
+ "epoch": 6.96875,
+ "grad_norm": 0.6653711103404113,
+ "learning_rate": 2e-05,
+ "loss": 0.6938,
+ "step": 223
+ },
+ {
+ "epoch": 6.96875,
+ "eval_loss": 0.6893274188041687,
+ "eval_runtime": 51.899,
+ "eval_samples_per_second": 3.854,
+ "eval_steps_per_second": 0.25,
+ "step": 223
+ },
+ {
+ "epoch": 7.0,
+ "grad_norm": 0.6730688267524797,
+ "learning_rate": 2e-05,
+ "loss": 0.7397,
+ "step": 224
+ },
+ {
+ "epoch": 7.0,
+ "eval_loss": 0.6895740032196045,
+ "eval_runtime": 52.1977,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 224
+ }
+ ],
+ "logging_steps": 1.0,
+ "max_steps": 224,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 7,
+ "save_steps": 5,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 322567586447360.0,
+ "train_batch_size": 8,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-224/training_args.bin b/checkpoint-224/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6e3191ec71847df102d8e3c538f0f4fea777607a
--- /dev/null
+++ b/checkpoint-224/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21fca50f49cefaafcd1ff13744949a11f0be41ae12da12aa7b74f1b7c0c2d5f2
+size 8184
diff --git a/checkpoint-224/zero_to_fp32.py b/checkpoint-224/zero_to_fp32.py
new file mode 100755
index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8
--- /dev/null
+++ b/checkpoint-224/zero_to_fp32.py
@@ -0,0 +1,604 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+ buffers: dict()
+ param_shapes: dict()
+ shared_params: list
+ ds_version: int
+ frozen_param_shapes: dict()
+ frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+ return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+ '''
+ alist.sort(key=natural_keys) sorts in human order
+ http://nedbatchelder.com/blog/200712/human_sorting.html
+ (See Toothy's implementation in the comments)
+ '''
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+ if not os.path.isdir(checkpoint_dir):
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+ # there should be only one file
+ if zero_stage <= 2:
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+ elif zero_stage == 3:
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+ if not os.path.exists(file):
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+ return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+ # XXX: need to test that this simple glob rule works for multi-node setup too
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+ if len(ckpt_files) == 0:
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+ return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+ zero_model_states = []
+ for file in files:
+ state_dict = torch.load(file, map_location=device)
+
+ if BUFFER_NAMES not in state_dict:
+ raise ValueError(f"{file} is not a model state checkpoint")
+ buffer_names = state_dict[BUFFER_NAMES]
+ if debug:
+ print("Found buffers:", buffer_names)
+
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+ param_shapes = state_dict[PARAM_SHAPES]
+
+ # collect parameters that are included in param_shapes
+ param_names = []
+ for s in param_shapes:
+ for name in s.keys():
+ param_names.append(name)
+
+ # update with frozen parameters
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+ if frozen_param_shapes is not None:
+ if debug:
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+ param_names += list(frozen_param_shapes.keys())
+
+ # handle shared params
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+ ds_version = state_dict.get(DS_VERSION, None)
+
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+ z_model_state = zero_model_state(buffers=buffers,
+ param_shapes=param_shapes,
+ shared_params=shared_params,
+ ds_version=ds_version,
+ frozen_param_shapes=frozen_param_shapes,
+ frozen_param_fragments=frozen_param_fragments)
+ zero_model_states.append(z_model_state)
+
+ return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+ total_files = len(files)
+ state_dicts = []
+ for f in files:
+ state_dict = torch.load(f, map_location=device)
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+ # and also handle the case where it was already removed by another helper script
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+ state_dicts.append(state_dict)
+
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
+ # use the max of the partition_count to get the dp world_size.
+
+ if type(world_size) is list:
+ world_size = max(world_size)
+
+ if world_size != total_files:
+ raise ValueError(
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+ )
+
+ # the groups are named differently in each stage
+ if zero_stage <= 2:
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+ elif zero_stage == 3:
+ fp32_groups_key = FP32_FLAT_GROUPS
+ else:
+ raise ValueError(f"unknown zero stage {zero_stage}")
+
+ if zero_stage <= 2:
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+ elif zero_stage == 3:
+ # if there is more than one param group, there will be multiple flattened tensors - one
+ # flattened tensor per group - for simplicity merge them into a single tensor
+ #
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+ fp32_flat_groups = [
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+ ]
+
+ return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+ """
+ Returns fp32 state_dict reconstructed from ds checkpoint
+
+ Args:
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+ """
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+ optim_files = get_optim_files(ds_checkpoint_dir)
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+ model_files = get_model_state_files(ds_checkpoint_dir)
+
+ zero_model_states = parse_model_states(model_files)
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+ if zero_stage <= 2:
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters)
+ elif zero_stage == 3:
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+ return
+
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+ if debug:
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+ wanted_params = len(frozen_param_shapes)
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+ print(f'Frozen params: Have {avail_numel} numels to process.')
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+ total_params = 0
+ total_numel = 0
+ for name, shape in frozen_param_shapes.items():
+ total_params += 1
+ unpartitioned_numel = shape.numel()
+ total_numel += unpartitioned_numel
+
+ state_dict[name] = frozen_param_fragments[name]
+
+ if debug:
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+ attr = getattr(obj, fn, None)
+ return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+ param_shapes = zero_model_states[0].param_shapes
+
+ # Reconstruction protocol:
+ #
+ # XXX: document this
+
+ if debug:
+ for i in range(world_size):
+ for j in range(len(fp32_flat_groups[0])):
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+ # XXX: memory usage doubles here (zero2)
+ num_param_groups = len(fp32_flat_groups[0])
+ merged_single_partition_of_fp32_groups = []
+ for i in range(num_param_groups):
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+ avail_numel = sum(
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+ if debug:
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+ # not asserting if there is a mismatch due to possible padding
+ print(f"Have {avail_numel} numels to process.")
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+ # params
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+ # out-of-core computing solution
+ total_numel = 0
+ total_params = 0
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+ offset = 0
+ avail_numel = full_single_fp32_vector.numel()
+ for name, shape in shapes.items():
+
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+ total_numel += unpartitioned_numel
+ total_params += 1
+
+ if debug:
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+ offset += unpartitioned_numel
+
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+ # live optimizer object, so we are checking that the numbers are within the right range
+ align_to = 2 * world_size
+
+ def zero2_align(x):
+ return align_to * math.ceil(x / align_to)
+
+ if debug:
+ print(f"original offset={offset}, avail_numel={avail_numel}")
+
+ offset = zero2_align(offset)
+ avail_numel = zero2_align(avail_numel)
+
+ if debug:
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
+
+ # Sanity check
+ if offset != avail_numel:
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters):
+ state_dict = OrderedDict()
+
+ # buffers
+ buffers = zero_model_states[0].buffers
+ state_dict.update(buffers)
+ if debug:
+ print(f"added {len(buffers)} buffers")
+
+ if not exclude_frozen_parameters:
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+ # recover shared parameters
+ for pair in zero_model_states[0].shared_params:
+ if pair[1] in state_dict:
+ state_dict[pair[0]] = state_dict[pair[1]]
+
+ return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+ remainder = unpartitioned_numel % world_size
+ padding_numel = (world_size - remainder) if remainder else 0
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+ return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+ return
+
+ if debug:
+ for i in range(world_size):
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+ wanted_params = len(frozen_param_shapes)
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+ print(f'Frozen params: Have {avail_numel} numels to process.')
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+ total_params = 0
+ total_numel = 0
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
+ total_params += 1
+ unpartitioned_numel = shape.numel()
+ total_numel += unpartitioned_numel
+
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+ if debug:
+ print(
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+ )
+
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+ param_shapes = zero_model_states[0].param_shapes
+ avail_numel = fp32_flat_groups[0].numel() * world_size
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+ # param, re-consolidating each param, while dealing with padding if any
+
+ # merge list of dicts, preserving order
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+ if debug:
+ for i in range(world_size):
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+ wanted_params = len(param_shapes)
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+ # not asserting if there is a mismatch due to possible padding
+ avail_numel = fp32_flat_groups[0].numel() * world_size
+ print(f"Trainable params: Have {avail_numel} numels to process.")
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+ # params
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+ # out-of-core computing solution
+ offset = 0
+ total_numel = 0
+ total_params = 0
+ for name, shape in param_shapes.items():
+
+ unpartitioned_numel = shape.numel()
+ total_numel += unpartitioned_numel
+ total_params += 1
+
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+ if debug:
+ print(
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+ )
+
+ # XXX: memory usage doubles here
+ state_dict[name] = torch.cat(
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
+ offset += partitioned_numel
+
+ offset *= world_size
+
+ # Sanity check
+ if offset != avail_numel:
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters):
+ state_dict = OrderedDict()
+
+ # buffers
+ buffers = zero_model_states[0].buffers
+ state_dict.update(buffers)
+ if debug:
+ print(f"added {len(buffers)} buffers")
+
+ if not exclude_frozen_parameters:
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+ # recover shared parameters
+ for pair in zero_model_states[0].shared_params:
+ if pair[1] in state_dict:
+ state_dict[pair[0]] = state_dict[pair[1]]
+
+ return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+ """
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+ via a model hub.
+
+ Args:
+ - ``checkpoint_dir``: path to the desired checkpoint folder
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+ - ``exclude_frozen_parameters``: exclude frozen parameters
+
+ Returns:
+ - pytorch ``state_dict``
+
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+ the checkpoint.
+
+ A typical usage might be ::
+
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+ # do the training and checkpoint saving
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+ model = model.cpu() # move to cpu
+ model.load_state_dict(state_dict)
+ # submit to model hub or save the model to share with others
+
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
+ application. i.e. you will need to re-initialize the deepspeed engine, since
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+ """
+ if tag is None:
+ latest_path = os.path.join(checkpoint_dir, 'latest')
+ if os.path.isfile(latest_path):
+ with open(latest_path, 'r') as fd:
+ tag = fd.read().strip()
+ else:
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+ if not os.path.isdir(ds_checkpoint_dir):
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
+ """
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+ Args:
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+ - ``exclude_frozen_parameters``: exclude frozen parameters
+ """
+
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+ print(f"Saving fp32 state dict to {output_file}")
+ torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+ """
+ 1. Put the provided model to cpu
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+ 3. Load it into the provided model
+
+ Args:
+ - ``model``: the model object to update
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+ Returns:
+ - ``model`: modified model
+
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+ conveniently placed for you in the checkpoint folder.
+
+ A typical usage might be ::
+
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+ # submit to model hub or save the model to share with others
+
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+ """
+ logger.info(f"Extracting fp32 weights")
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+ logger.info(f"Overwriting model with fp32 weights")
+ model = model.cpu()
+ model.load_state_dict(state_dict, strict=False)
+
+ return model
+
+
+if __name__ == "__main__":
+
+ parser = argparse.ArgumentParser()
+ parser.add_argument("checkpoint_dir",
+ type=str,
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+ parser.add_argument(
+ "output_file",
+ type=str,
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+ parser.add_argument("-t",
+ "--tag",
+ type=str,
+ default=None,
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+ args = parser.parse_args()
+
+ debug = args.debug
+
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+ args.output_file,
+ tag=args.tag,
+ exclude_frozen_parameters=args.exclude_frozen_parameters)
diff --git a/checkpoint-320/README.md b/checkpoint-320/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..622b1afbf67513c6d5b974cf6a1b6d5ad79c52e7
--- /dev/null
+++ b/checkpoint-320/README.md
@@ -0,0 +1,202 @@
+---
+base_model: liuhaotian/llava-v1.5-13b
+library_name: peft
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.13.2
\ No newline at end of file
diff --git a/checkpoint-320/adapter_config.json b/checkpoint-320/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..31d77354128d962ce655ffa50a52c067d2b8a463
--- /dev/null
+++ b/checkpoint-320/adapter_config.json
@@ -0,0 +1,34 @@
+{
+ "alpha_pattern": {},
+ "auto_mapping": null,
+ "base_model_name_or_path": "liuhaotian/llava-v1.5-13b",
+ "bias": "none",
+ "fan_in_fan_out": false,
+ "inference_mode": true,
+ "init_lora_weights": true,
+ "layer_replication": null,
+ "layers_pattern": null,
+ "layers_to_transform": null,
+ "loftq_config": {},
+ "lora_alpha": 16,
+ "lora_dropout": 0.05,
+ "megatron_config": null,
+ "megatron_core": "megatron.core",
+ "modules_to_save": null,
+ "peft_type": "LORA",
+ "r": 8,
+ "rank_pattern": {},
+ "revision": null,
+ "target_modules": [
+ "up_proj",
+ "k_proj",
+ "v_proj",
+ "gate_proj",
+ "o_proj",
+ "down_proj",
+ "q_proj"
+ ],
+ "task_type": "CAUSAL_LM",
+ "use_dora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-320/adapter_model.safetensors b/checkpoint-320/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5cd5d83c65780ad82efdd0132cce3cfd1f1d64fa
--- /dev/null
+++ b/checkpoint-320/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:623b69502fe51b1eed879b0e0e29485ea633f4af66c5d1cfbc7a6c33c093568a
+size 62660864
diff --git a/checkpoint-320/global_step320/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-320/global_step320/zero_pp_rank_0_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e1ba37b5da2c40a6b1e40a04d67b42e019762623
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_0_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8f38fd19d47d3a7f3f53c6d20cd95926e61f16c114ce7946b1d673f3041f837
+size 593618
diff --git a/checkpoint-320/global_step320/zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-320/global_step320/zero_pp_rank_0_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..089ea71a391360ac72454e41837ab1e71a4698f7
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_0_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a7c8a298ac8838329dd66ff2e8fdac8ffabcefb449e22a3a1d11ee77376209a
+size 188286957
diff --git a/checkpoint-320/global_step320/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-320/global_step320/zero_pp_rank_1_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..717b209308a10e4dbbc3e62b0b703a4fe8c3dcf1
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_1_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91e813727135ac80358f97bf33e38bc7f766a71c143ccf5e7aae24f8dae837e0
+size 593618
diff --git a/checkpoint-320/global_step320/zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-320/global_step320/zero_pp_rank_1_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d6a0e59a6492528dbf6945723440d5112425007b
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_1_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8244a7c9eca3b99273f88fdb9bd50bb1f1bb20c3143e5e6010bbab902181ed11
+size 188286957
diff --git a/checkpoint-320/global_step320/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-320/global_step320/zero_pp_rank_2_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6e8df32bf95f12ee49bf9bbec383192e3fa4835e
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_2_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:439ec584a84b9faf03f1b1df8b8fc29f633f526de043b549f21a36c5c753e21b
+size 593618
diff --git a/checkpoint-320/global_step320/zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-320/global_step320/zero_pp_rank_2_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ebe0efd1b56c584ae43a027597dcd2cb305d4fff
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_2_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:800813a53a6a794d85e69313e2602f05058273df63bfded7fc5a5cbf6f601774
+size 188286957
diff --git a/checkpoint-320/global_step320/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-320/global_step320/zero_pp_rank_3_mp_rank_00_model_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c5cd84c806a36a8ae3e0c631e47f927c91c8e4ed
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_3_mp_rank_00_model_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2203aa85ac0d0ab529317771b382ac283333ea0d10a636bdf695573d5c44cdff
+size 593618
diff --git a/checkpoint-320/global_step320/zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-320/global_step320/zero_pp_rank_3_mp_rank_00_optim_states.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f18b464a1f3a9c81ff0f9acb5812039f65cadbe3
--- /dev/null
+++ b/checkpoint-320/global_step320/zero_pp_rank_3_mp_rank_00_optim_states.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92ba3aa4564dd89fac03a619fe9596391ee3bdbd8a1e7d499e87c5d460ee3ef2
+size 188286957
diff --git a/checkpoint-320/latest b/checkpoint-320/latest
new file mode 100644
index 0000000000000000000000000000000000000000..9d535587efdab3121736d8095481e4143f000213
--- /dev/null
+++ b/checkpoint-320/latest
@@ -0,0 +1 @@
+global_step320
\ No newline at end of file
diff --git a/checkpoint-320/rng_state_0.pth b/checkpoint-320/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..97954d380c745b67032706c585028e5c3c53ee6e
--- /dev/null
+++ b/checkpoint-320/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70e8e40bf1ad96295ca0d42d42672162cccbcbebb089a9837b3df939e99749c9
+size 14960
diff --git a/checkpoint-320/rng_state_1.pth b/checkpoint-320/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..20eb4b1b32177540f1ad5ee2a42627af66c9a859
--- /dev/null
+++ b/checkpoint-320/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63f24b3a2fcd38868dad836e3fd055d6d1cd19f36948ca4f89760988afb3e439
+size 14960
diff --git a/checkpoint-320/rng_state_2.pth b/checkpoint-320/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..6ea4dde9c5048ed402ff120cc177f5e8122fa39f
--- /dev/null
+++ b/checkpoint-320/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a0998e05fda02393da198c76a9d294538ace12d82c157062e1ca1b94450cacc
+size 14960
diff --git a/checkpoint-320/rng_state_3.pth b/checkpoint-320/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..97d51af0909be350a62200b156fd9f94cbd92b29
--- /dev/null
+++ b/checkpoint-320/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86c573687454121dca6cfe001a0e146f7fee70895d84869ebe65004d1bfaa5c5
+size 14960
diff --git a/checkpoint-320/special_tokens_map.json b/checkpoint-320/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..14761dcf1466dc232bd41de9c21d4c617b15755e
--- /dev/null
+++ b/checkpoint-320/special_tokens_map.json
@@ -0,0 +1,24 @@
+{
+ "bos_token": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ "eos_token": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ "pad_token": "",
+ "unk_token": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ }
+}
diff --git a/checkpoint-320/tokenizer.model b/checkpoint-320/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899
--- /dev/null
+++ b/checkpoint-320/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723
diff --git a/checkpoint-320/tokenizer_config.json b/checkpoint-320/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..26c65df1bf794f101c1dd54c908180dc0d880fe3
--- /dev/null
+++ b/checkpoint-320/tokenizer_config.json
@@ -0,0 +1,43 @@
+{
+ "add_bos_token": true,
+ "add_eos_token": false,
+ "add_prefix_space": true,
+ "added_tokens_decoder": {
+ "0": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "1": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "2": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ }
+ },
+ "bos_token": "",
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "",
+ "legacy": false,
+ "model_max_length": 2048,
+ "pad_token": "",
+ "padding_side": "right",
+ "sp_model_kwargs": {},
+ "spaces_between_special_tokens": false,
+ "tokenizer_class": "LlamaTokenizer",
+ "unk_token": "",
+ "use_default_system_prompt": false
+}
diff --git a/checkpoint-320/trainer_state.json b/checkpoint-320/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..51c8d1192accd18fdae72a6e6bb304116d525bb5
--- /dev/null
+++ b/checkpoint-320/trainer_state.json
@@ -0,0 +1,4833 @@
+{
+ "best_metric": 0.6895740032196045,
+ "best_model_checkpoint": "./checkpoints/llava-v1.5-13b/checkpoint-224",
+ "epoch": 10.0,
+ "eval_steps": 1.0,
+ "global_step": 320,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.03125,
+ "grad_norm": 0.2380081706918525,
+ "learning_rate": 0.0,
+ "loss": 1.2458,
+ "step": 1
+ },
+ {
+ "epoch": 0.03125,
+ "eval_loss": 1.3161638975143433,
+ "eval_runtime": 50.8995,
+ "eval_samples_per_second": 3.929,
+ "eval_steps_per_second": 0.255,
+ "step": 1
+ },
+ {
+ "epoch": 0.0625,
+ "grad_norm": 0.20429495268987705,
+ "learning_rate": 8.613531161467863e-06,
+ "loss": 1.2003,
+ "step": 2
+ },
+ {
+ "epoch": 0.0625,
+ "eval_loss": 1.3161638975143433,
+ "eval_runtime": 47.4818,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 2
+ },
+ {
+ "epoch": 0.09375,
+ "grad_norm": 0.20616215800420787,
+ "learning_rate": 1.3652123889719709e-05,
+ "loss": 1.2622,
+ "step": 3
+ },
+ {
+ "epoch": 0.09375,
+ "eval_loss": 1.309991478919983,
+ "eval_runtime": 47.4152,
+ "eval_samples_per_second": 4.218,
+ "eval_steps_per_second": 0.274,
+ "step": 3
+ },
+ {
+ "epoch": 0.125,
+ "grad_norm": 0.20155595022101944,
+ "learning_rate": 1.7227062322935725e-05,
+ "loss": 1.2845,
+ "step": 4
+ },
+ {
+ "epoch": 0.125,
+ "eval_loss": 1.3013781309127808,
+ "eval_runtime": 47.4814,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 4
+ },
+ {
+ "epoch": 0.15625,
+ "grad_norm": 0.21113117474989132,
+ "learning_rate": 2e-05,
+ "loss": 1.246,
+ "step": 5
+ },
+ {
+ "epoch": 0.15625,
+ "eval_loss": 1.2892160415649414,
+ "eval_runtime": 47.7209,
+ "eval_samples_per_second": 4.191,
+ "eval_steps_per_second": 0.272,
+ "step": 5
+ },
+ {
+ "epoch": 0.1875,
+ "grad_norm": 0.21377946631015488,
+ "learning_rate": 2e-05,
+ "loss": 1.2684,
+ "step": 6
+ },
+ {
+ "epoch": 0.1875,
+ "eval_loss": 1.2754532098770142,
+ "eval_runtime": 47.5781,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 6
+ },
+ {
+ "epoch": 0.21875,
+ "grad_norm": 0.2284268997618767,
+ "learning_rate": 2e-05,
+ "loss": 1.2681,
+ "step": 7
+ },
+ {
+ "epoch": 0.21875,
+ "eval_loss": 1.2605774402618408,
+ "eval_runtime": 47.5326,
+ "eval_samples_per_second": 4.208,
+ "eval_steps_per_second": 0.273,
+ "step": 7
+ },
+ {
+ "epoch": 0.25,
+ "grad_norm": 0.23585343568544442,
+ "learning_rate": 2e-05,
+ "loss": 1.2407,
+ "step": 8
+ },
+ {
+ "epoch": 0.25,
+ "eval_loss": 1.244718313217163,
+ "eval_runtime": 47.5001,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 8
+ },
+ {
+ "epoch": 0.28125,
+ "grad_norm": 0.23051191992462533,
+ "learning_rate": 2e-05,
+ "loss": 1.2766,
+ "step": 9
+ },
+ {
+ "epoch": 0.28125,
+ "eval_loss": 1.2285138368606567,
+ "eval_runtime": 47.4631,
+ "eval_samples_per_second": 4.214,
+ "eval_steps_per_second": 0.274,
+ "step": 9
+ },
+ {
+ "epoch": 0.3125,
+ "grad_norm": 0.22726394327484983,
+ "learning_rate": 2e-05,
+ "loss": 1.2024,
+ "step": 10
+ },
+ {
+ "epoch": 0.3125,
+ "eval_loss": 1.2118008136749268,
+ "eval_runtime": 47.4991,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 10
+ },
+ {
+ "epoch": 0.34375,
+ "grad_norm": 0.25404890894461285,
+ "learning_rate": 2e-05,
+ "loss": 1.2742,
+ "step": 11
+ },
+ {
+ "epoch": 0.34375,
+ "eval_loss": 1.1942989826202393,
+ "eval_runtime": 49.2609,
+ "eval_samples_per_second": 4.06,
+ "eval_steps_per_second": 0.264,
+ "step": 11
+ },
+ {
+ "epoch": 0.375,
+ "grad_norm": 0.26336210916526287,
+ "learning_rate": 2e-05,
+ "loss": 1.2258,
+ "step": 12
+ },
+ {
+ "epoch": 0.375,
+ "eval_loss": 1.176426649093628,
+ "eval_runtime": 49.0639,
+ "eval_samples_per_second": 4.076,
+ "eval_steps_per_second": 0.265,
+ "step": 12
+ },
+ {
+ "epoch": 0.40625,
+ "grad_norm": 0.29637148470746666,
+ "learning_rate": 2e-05,
+ "loss": 1.2345,
+ "step": 13
+ },
+ {
+ "epoch": 0.40625,
+ "eval_loss": 1.1577811241149902,
+ "eval_runtime": 49.1352,
+ "eval_samples_per_second": 4.07,
+ "eval_steps_per_second": 0.265,
+ "step": 13
+ },
+ {
+ "epoch": 0.4375,
+ "grad_norm": 0.2841880377627424,
+ "learning_rate": 2e-05,
+ "loss": 1.0765,
+ "step": 14
+ },
+ {
+ "epoch": 0.4375,
+ "eval_loss": 1.1381279230117798,
+ "eval_runtime": 49.25,
+ "eval_samples_per_second": 4.061,
+ "eval_steps_per_second": 0.264,
+ "step": 14
+ },
+ {
+ "epoch": 0.46875,
+ "grad_norm": 0.2773140636191091,
+ "learning_rate": 2e-05,
+ "loss": 1.1812,
+ "step": 15
+ },
+ {
+ "epoch": 0.46875,
+ "eval_loss": 1.1178216934204102,
+ "eval_runtime": 49.0879,
+ "eval_samples_per_second": 4.074,
+ "eval_steps_per_second": 0.265,
+ "step": 15
+ },
+ {
+ "epoch": 0.5,
+ "grad_norm": 0.3568607365552051,
+ "learning_rate": 2e-05,
+ "loss": 1.1327,
+ "step": 16
+ },
+ {
+ "epoch": 0.5,
+ "eval_loss": 1.0954149961471558,
+ "eval_runtime": 48.6546,
+ "eval_samples_per_second": 4.111,
+ "eval_steps_per_second": 0.267,
+ "step": 16
+ },
+ {
+ "epoch": 0.53125,
+ "grad_norm": 0.32574391414112897,
+ "learning_rate": 2e-05,
+ "loss": 1.1162,
+ "step": 17
+ },
+ {
+ "epoch": 0.53125,
+ "eval_loss": 1.071275234222412,
+ "eval_runtime": 48.5618,
+ "eval_samples_per_second": 4.118,
+ "eval_steps_per_second": 0.268,
+ "step": 17
+ },
+ {
+ "epoch": 0.5625,
+ "grad_norm": 0.4256864144638081,
+ "learning_rate": 2e-05,
+ "loss": 1.1138,
+ "step": 18
+ },
+ {
+ "epoch": 0.5625,
+ "eval_loss": 1.0455905199050903,
+ "eval_runtime": 48.4981,
+ "eval_samples_per_second": 4.124,
+ "eval_steps_per_second": 0.268,
+ "step": 18
+ },
+ {
+ "epoch": 0.59375,
+ "grad_norm": 0.31230014132112643,
+ "learning_rate": 2e-05,
+ "loss": 1.0011,
+ "step": 19
+ },
+ {
+ "epoch": 0.59375,
+ "eval_loss": 1.0208789110183716,
+ "eval_runtime": 48.4675,
+ "eval_samples_per_second": 4.126,
+ "eval_steps_per_second": 0.268,
+ "step": 19
+ },
+ {
+ "epoch": 0.625,
+ "grad_norm": 0.3025724039243594,
+ "learning_rate": 2e-05,
+ "loss": 1.109,
+ "step": 20
+ },
+ {
+ "epoch": 0.625,
+ "eval_loss": 1.002480149269104,
+ "eval_runtime": 48.5265,
+ "eval_samples_per_second": 4.121,
+ "eval_steps_per_second": 0.268,
+ "step": 20
+ },
+ {
+ "epoch": 0.65625,
+ "grad_norm": 0.27787879590501874,
+ "learning_rate": 2e-05,
+ "loss": 1.0291,
+ "step": 21
+ },
+ {
+ "epoch": 0.65625,
+ "eval_loss": 0.9933492541313171,
+ "eval_runtime": 50.0369,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 21
+ },
+ {
+ "epoch": 0.6875,
+ "grad_norm": 0.4231294067130801,
+ "learning_rate": 2e-05,
+ "loss": 1.0779,
+ "step": 22
+ },
+ {
+ "epoch": 0.6875,
+ "eval_loss": 0.9850385785102844,
+ "eval_runtime": 50.0062,
+ "eval_samples_per_second": 4.0,
+ "eval_steps_per_second": 0.26,
+ "step": 22
+ },
+ {
+ "epoch": 0.71875,
+ "grad_norm": 0.42130097437373987,
+ "learning_rate": 2e-05,
+ "loss": 1.0897,
+ "step": 23
+ },
+ {
+ "epoch": 0.71875,
+ "eval_loss": 0.9758670330047607,
+ "eval_runtime": 50.1031,
+ "eval_samples_per_second": 3.992,
+ "eval_steps_per_second": 0.259,
+ "step": 23
+ },
+ {
+ "epoch": 0.75,
+ "grad_norm": 0.27711808063263893,
+ "learning_rate": 2e-05,
+ "loss": 1.0739,
+ "step": 24
+ },
+ {
+ "epoch": 0.75,
+ "eval_loss": 0.9674506187438965,
+ "eval_runtime": 50.0337,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 24
+ },
+ {
+ "epoch": 0.78125,
+ "grad_norm": 0.2879649409281791,
+ "learning_rate": 2e-05,
+ "loss": 1.0182,
+ "step": 25
+ },
+ {
+ "epoch": 0.78125,
+ "eval_loss": 0.9592065215110779,
+ "eval_runtime": 50.0709,
+ "eval_samples_per_second": 3.994,
+ "eval_steps_per_second": 0.26,
+ "step": 25
+ },
+ {
+ "epoch": 0.8125,
+ "grad_norm": 0.19327450826076825,
+ "learning_rate": 2e-05,
+ "loss": 1.0413,
+ "step": 26
+ },
+ {
+ "epoch": 0.8125,
+ "eval_loss": 0.9518552422523499,
+ "eval_runtime": 50.0572,
+ "eval_samples_per_second": 3.995,
+ "eval_steps_per_second": 0.26,
+ "step": 26
+ },
+ {
+ "epoch": 0.84375,
+ "grad_norm": 0.19707021382445633,
+ "learning_rate": 2e-05,
+ "loss": 0.9525,
+ "step": 27
+ },
+ {
+ "epoch": 0.84375,
+ "eval_loss": 0.9449941515922546,
+ "eval_runtime": 50.0515,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 27
+ },
+ {
+ "epoch": 0.875,
+ "grad_norm": 0.2420270757641518,
+ "learning_rate": 2e-05,
+ "loss": 0.9658,
+ "step": 28
+ },
+ {
+ "epoch": 0.875,
+ "eval_loss": 0.9378474354743958,
+ "eval_runtime": 49.9299,
+ "eval_samples_per_second": 4.006,
+ "eval_steps_per_second": 0.26,
+ "step": 28
+ },
+ {
+ "epoch": 0.90625,
+ "grad_norm": 0.18074632782127534,
+ "learning_rate": 2e-05,
+ "loss": 0.9866,
+ "step": 29
+ },
+ {
+ "epoch": 0.90625,
+ "eval_loss": 0.93099045753479,
+ "eval_runtime": 50.0096,
+ "eval_samples_per_second": 3.999,
+ "eval_steps_per_second": 0.26,
+ "step": 29
+ },
+ {
+ "epoch": 0.9375,
+ "grad_norm": 0.1936051126921734,
+ "learning_rate": 2e-05,
+ "loss": 1.0128,
+ "step": 30
+ },
+ {
+ "epoch": 0.9375,
+ "eval_loss": 0.9244199991226196,
+ "eval_runtime": 50.2469,
+ "eval_samples_per_second": 3.98,
+ "eval_steps_per_second": 0.259,
+ "step": 30
+ },
+ {
+ "epoch": 0.96875,
+ "grad_norm": 0.26164254459782943,
+ "learning_rate": 2e-05,
+ "loss": 0.88,
+ "step": 31
+ },
+ {
+ "epoch": 0.96875,
+ "eval_loss": 0.9175177216529846,
+ "eval_runtime": 50.1695,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 31
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.18677152741688485,
+ "learning_rate": 2e-05,
+ "loss": 0.9569,
+ "step": 32
+ },
+ {
+ "epoch": 1.0,
+ "eval_loss": 0.9108598828315735,
+ "eval_runtime": 50.0387,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 32
+ },
+ {
+ "epoch": 1.03125,
+ "grad_norm": 0.20486279036126417,
+ "learning_rate": 2e-05,
+ "loss": 1.0208,
+ "step": 33
+ },
+ {
+ "epoch": 1.03125,
+ "eval_loss": 0.9042049646377563,
+ "eval_runtime": 50.1472,
+ "eval_samples_per_second": 3.988,
+ "eval_steps_per_second": 0.259,
+ "step": 33
+ },
+ {
+ "epoch": 1.0625,
+ "grad_norm": 0.2004946169291112,
+ "learning_rate": 2e-05,
+ "loss": 0.9931,
+ "step": 34
+ },
+ {
+ "epoch": 1.0625,
+ "eval_loss": 0.8980298042297363,
+ "eval_runtime": 50.245,
+ "eval_samples_per_second": 3.98,
+ "eval_steps_per_second": 0.259,
+ "step": 34
+ },
+ {
+ "epoch": 1.09375,
+ "grad_norm": 0.1645872432258401,
+ "learning_rate": 2e-05,
+ "loss": 1.0184,
+ "step": 35
+ },
+ {
+ "epoch": 1.09375,
+ "eval_loss": 0.8924428820610046,
+ "eval_runtime": 50.3703,
+ "eval_samples_per_second": 3.971,
+ "eval_steps_per_second": 0.258,
+ "step": 35
+ },
+ {
+ "epoch": 1.125,
+ "grad_norm": 0.18293519304435016,
+ "learning_rate": 2e-05,
+ "loss": 1.0026,
+ "step": 36
+ },
+ {
+ "epoch": 1.125,
+ "eval_loss": 0.8870412707328796,
+ "eval_runtime": 50.0483,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 36
+ },
+ {
+ "epoch": 1.15625,
+ "grad_norm": 0.17712548516246762,
+ "learning_rate": 2e-05,
+ "loss": 0.9387,
+ "step": 37
+ },
+ {
+ "epoch": 1.15625,
+ "eval_loss": 0.881915271282196,
+ "eval_runtime": 49.9751,
+ "eval_samples_per_second": 4.002,
+ "eval_steps_per_second": 0.26,
+ "step": 37
+ },
+ {
+ "epoch": 1.1875,
+ "grad_norm": 0.21472689311609464,
+ "learning_rate": 2e-05,
+ "loss": 0.958,
+ "step": 38
+ },
+ {
+ "epoch": 1.1875,
+ "eval_loss": 0.8768754601478577,
+ "eval_runtime": 50.1204,
+ "eval_samples_per_second": 3.99,
+ "eval_steps_per_second": 0.259,
+ "step": 38
+ },
+ {
+ "epoch": 1.21875,
+ "grad_norm": 0.21117297910005806,
+ "learning_rate": 2e-05,
+ "loss": 0.9922,
+ "step": 39
+ },
+ {
+ "epoch": 1.21875,
+ "eval_loss": 0.8718628883361816,
+ "eval_runtime": 50.1732,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 39
+ },
+ {
+ "epoch": 1.25,
+ "grad_norm": 0.17835587003909165,
+ "learning_rate": 2e-05,
+ "loss": 0.9776,
+ "step": 40
+ },
+ {
+ "epoch": 1.25,
+ "eval_loss": 0.8669865131378174,
+ "eval_runtime": 50.1148,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 40
+ },
+ {
+ "epoch": 1.28125,
+ "grad_norm": 0.2092736372483734,
+ "learning_rate": 2e-05,
+ "loss": 0.9731,
+ "step": 41
+ },
+ {
+ "epoch": 1.28125,
+ "eval_loss": 0.8619834780693054,
+ "eval_runtime": 50.052,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 41
+ },
+ {
+ "epoch": 1.3125,
+ "grad_norm": 0.2338857391910308,
+ "learning_rate": 2e-05,
+ "loss": 0.9319,
+ "step": 42
+ },
+ {
+ "epoch": 1.3125,
+ "eval_loss": 0.8572126030921936,
+ "eval_runtime": 50.1212,
+ "eval_samples_per_second": 3.99,
+ "eval_steps_per_second": 0.259,
+ "step": 42
+ },
+ {
+ "epoch": 1.34375,
+ "grad_norm": 0.19168719284572813,
+ "learning_rate": 2e-05,
+ "loss": 0.9083,
+ "step": 43
+ },
+ {
+ "epoch": 1.34375,
+ "eval_loss": 0.8525611758232117,
+ "eval_runtime": 50.1733,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 43
+ },
+ {
+ "epoch": 1.375,
+ "grad_norm": 0.20004868138433377,
+ "learning_rate": 2e-05,
+ "loss": 0.9118,
+ "step": 44
+ },
+ {
+ "epoch": 1.375,
+ "eval_loss": 0.8483461141586304,
+ "eval_runtime": 50.1083,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 44
+ },
+ {
+ "epoch": 1.40625,
+ "grad_norm": 0.19012965506122342,
+ "learning_rate": 2e-05,
+ "loss": 0.8888,
+ "step": 45
+ },
+ {
+ "epoch": 1.40625,
+ "eval_loss": 0.8446614742279053,
+ "eval_runtime": 50.1171,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 45
+ },
+ {
+ "epoch": 1.4375,
+ "grad_norm": 0.21187005706805245,
+ "learning_rate": 2e-05,
+ "loss": 0.9319,
+ "step": 46
+ },
+ {
+ "epoch": 1.4375,
+ "eval_loss": 0.8412036299705505,
+ "eval_runtime": 50.0918,
+ "eval_samples_per_second": 3.993,
+ "eval_steps_per_second": 0.26,
+ "step": 46
+ },
+ {
+ "epoch": 1.46875,
+ "grad_norm": 0.19673832205926584,
+ "learning_rate": 2e-05,
+ "loss": 0.9359,
+ "step": 47
+ },
+ {
+ "epoch": 1.46875,
+ "eval_loss": 0.8380417823791504,
+ "eval_runtime": 50.2214,
+ "eval_samples_per_second": 3.982,
+ "eval_steps_per_second": 0.259,
+ "step": 47
+ },
+ {
+ "epoch": 1.5,
+ "grad_norm": 0.21712294106174318,
+ "learning_rate": 2e-05,
+ "loss": 0.8511,
+ "step": 48
+ },
+ {
+ "epoch": 1.5,
+ "eval_loss": 0.8353021740913391,
+ "eval_runtime": 50.1617,
+ "eval_samples_per_second": 3.987,
+ "eval_steps_per_second": 0.259,
+ "step": 48
+ },
+ {
+ "epoch": 1.53125,
+ "grad_norm": 0.2138924779700934,
+ "learning_rate": 2e-05,
+ "loss": 0.8695,
+ "step": 49
+ },
+ {
+ "epoch": 1.53125,
+ "eval_loss": 0.8327407836914062,
+ "eval_runtime": 50.1442,
+ "eval_samples_per_second": 3.988,
+ "eval_steps_per_second": 0.259,
+ "step": 49
+ },
+ {
+ "epoch": 1.5625,
+ "grad_norm": 0.22387442384578618,
+ "learning_rate": 2e-05,
+ "loss": 0.8518,
+ "step": 50
+ },
+ {
+ "epoch": 1.5625,
+ "eval_loss": 0.8301742076873779,
+ "eval_runtime": 50.1867,
+ "eval_samples_per_second": 3.985,
+ "eval_steps_per_second": 0.259,
+ "step": 50
+ },
+ {
+ "epoch": 1.59375,
+ "grad_norm": 0.1975577146517192,
+ "learning_rate": 2e-05,
+ "loss": 0.8868,
+ "step": 51
+ },
+ {
+ "epoch": 1.59375,
+ "eval_loss": 0.8275265693664551,
+ "eval_runtime": 51.2257,
+ "eval_samples_per_second": 3.904,
+ "eval_steps_per_second": 0.254,
+ "step": 51
+ },
+ {
+ "epoch": 1.625,
+ "grad_norm": 0.21474817057286624,
+ "learning_rate": 2e-05,
+ "loss": 0.767,
+ "step": 52
+ },
+ {
+ "epoch": 1.625,
+ "eval_loss": 0.824796736240387,
+ "eval_runtime": 51.276,
+ "eval_samples_per_second": 3.9,
+ "eval_steps_per_second": 0.254,
+ "step": 52
+ },
+ {
+ "epoch": 1.65625,
+ "grad_norm": 0.21105651676755652,
+ "learning_rate": 2e-05,
+ "loss": 0.9219,
+ "step": 53
+ },
+ {
+ "epoch": 1.65625,
+ "eval_loss": 0.8221166729927063,
+ "eval_runtime": 51.141,
+ "eval_samples_per_second": 3.911,
+ "eval_steps_per_second": 0.254,
+ "step": 53
+ },
+ {
+ "epoch": 1.6875,
+ "grad_norm": 0.20706475184742085,
+ "learning_rate": 2e-05,
+ "loss": 0.8873,
+ "step": 54
+ },
+ {
+ "epoch": 1.6875,
+ "eval_loss": 0.819589376449585,
+ "eval_runtime": 51.0045,
+ "eval_samples_per_second": 3.921,
+ "eval_steps_per_second": 0.255,
+ "step": 54
+ },
+ {
+ "epoch": 1.71875,
+ "grad_norm": 0.21722220033855957,
+ "learning_rate": 2e-05,
+ "loss": 0.8956,
+ "step": 55
+ },
+ {
+ "epoch": 1.71875,
+ "eval_loss": 0.8176340460777283,
+ "eval_runtime": 51.1941,
+ "eval_samples_per_second": 3.907,
+ "eval_steps_per_second": 0.254,
+ "step": 55
+ },
+ {
+ "epoch": 1.75,
+ "grad_norm": 0.20669001221665667,
+ "learning_rate": 2e-05,
+ "loss": 0.9506,
+ "step": 56
+ },
+ {
+ "epoch": 1.75,
+ "eval_loss": 0.8158826231956482,
+ "eval_runtime": 52.1162,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 56
+ },
+ {
+ "epoch": 1.78125,
+ "grad_norm": 0.22189732090066341,
+ "learning_rate": 2e-05,
+ "loss": 0.8955,
+ "step": 57
+ },
+ {
+ "epoch": 1.78125,
+ "eval_loss": 0.814656674861908,
+ "eval_runtime": 52.1361,
+ "eval_samples_per_second": 3.836,
+ "eval_steps_per_second": 0.249,
+ "step": 57
+ },
+ {
+ "epoch": 1.8125,
+ "grad_norm": 0.2030113892848459,
+ "learning_rate": 2e-05,
+ "loss": 0.9108,
+ "step": 58
+ },
+ {
+ "epoch": 1.8125,
+ "eval_loss": 0.813343346118927,
+ "eval_runtime": 52.2552,
+ "eval_samples_per_second": 3.827,
+ "eval_steps_per_second": 0.249,
+ "step": 58
+ },
+ {
+ "epoch": 1.84375,
+ "grad_norm": 0.2123201057569791,
+ "learning_rate": 2e-05,
+ "loss": 0.8779,
+ "step": 59
+ },
+ {
+ "epoch": 1.84375,
+ "eval_loss": 0.8116877675056458,
+ "eval_runtime": 52.1233,
+ "eval_samples_per_second": 3.837,
+ "eval_steps_per_second": 0.249,
+ "step": 59
+ },
+ {
+ "epoch": 1.875,
+ "grad_norm": 0.211551126937912,
+ "learning_rate": 2e-05,
+ "loss": 0.9294,
+ "step": 60
+ },
+ {
+ "epoch": 1.875,
+ "eval_loss": 0.8098442554473877,
+ "eval_runtime": 52.1091,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 60
+ },
+ {
+ "epoch": 1.90625,
+ "grad_norm": 0.24981344981629752,
+ "learning_rate": 2e-05,
+ "loss": 0.8409,
+ "step": 61
+ },
+ {
+ "epoch": 1.90625,
+ "eval_loss": 0.8070770502090454,
+ "eval_runtime": 53.4187,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 61
+ },
+ {
+ "epoch": 1.9375,
+ "grad_norm": 0.2341550589775159,
+ "learning_rate": 2e-05,
+ "loss": 0.888,
+ "step": 62
+ },
+ {
+ "epoch": 1.9375,
+ "eval_loss": 0.8040286898612976,
+ "eval_runtime": 53.2197,
+ "eval_samples_per_second": 3.758,
+ "eval_steps_per_second": 0.244,
+ "step": 62
+ },
+ {
+ "epoch": 1.96875,
+ "grad_norm": 0.2336241775649256,
+ "learning_rate": 2e-05,
+ "loss": 0.913,
+ "step": 63
+ },
+ {
+ "epoch": 1.96875,
+ "eval_loss": 0.8013430833816528,
+ "eval_runtime": 53.1784,
+ "eval_samples_per_second": 3.761,
+ "eval_steps_per_second": 0.244,
+ "step": 63
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.2414390628081758,
+ "learning_rate": 2e-05,
+ "loss": 0.8754,
+ "step": 64
+ },
+ {
+ "epoch": 2.0,
+ "eval_loss": 0.7985894680023193,
+ "eval_runtime": 53.2454,
+ "eval_samples_per_second": 3.756,
+ "eval_steps_per_second": 0.244,
+ "step": 64
+ },
+ {
+ "epoch": 2.03125,
+ "grad_norm": 0.2484104465653703,
+ "learning_rate": 2e-05,
+ "loss": 0.8497,
+ "step": 65
+ },
+ {
+ "epoch": 2.03125,
+ "eval_loss": 0.7954932451248169,
+ "eval_runtime": 53.3794,
+ "eval_samples_per_second": 3.747,
+ "eval_steps_per_second": 0.244,
+ "step": 65
+ },
+ {
+ "epoch": 2.0625,
+ "grad_norm": 0.23859744120942086,
+ "learning_rate": 2e-05,
+ "loss": 0.8567,
+ "step": 66
+ },
+ {
+ "epoch": 2.0625,
+ "eval_loss": 0.7929843068122864,
+ "eval_runtime": 55.517,
+ "eval_samples_per_second": 3.602,
+ "eval_steps_per_second": 0.234,
+ "step": 66
+ },
+ {
+ "epoch": 2.09375,
+ "grad_norm": 0.24584758647855462,
+ "learning_rate": 2e-05,
+ "loss": 0.8489,
+ "step": 67
+ },
+ {
+ "epoch": 2.09375,
+ "eval_loss": 0.7903321981430054,
+ "eval_runtime": 55.4151,
+ "eval_samples_per_second": 3.609,
+ "eval_steps_per_second": 0.235,
+ "step": 67
+ },
+ {
+ "epoch": 2.125,
+ "grad_norm": 0.2484917818304153,
+ "learning_rate": 2e-05,
+ "loss": 0.9122,
+ "step": 68
+ },
+ {
+ "epoch": 2.125,
+ "eval_loss": 0.7877185344696045,
+ "eval_runtime": 55.4069,
+ "eval_samples_per_second": 3.61,
+ "eval_steps_per_second": 0.235,
+ "step": 68
+ },
+ {
+ "epoch": 2.15625,
+ "grad_norm": 0.2184614083026819,
+ "learning_rate": 2e-05,
+ "loss": 0.8355,
+ "step": 69
+ },
+ {
+ "epoch": 2.15625,
+ "eval_loss": 0.7852210998535156,
+ "eval_runtime": 55.3381,
+ "eval_samples_per_second": 3.614,
+ "eval_steps_per_second": 0.235,
+ "step": 69
+ },
+ {
+ "epoch": 2.1875,
+ "grad_norm": 0.24978410070800153,
+ "learning_rate": 2e-05,
+ "loss": 0.7968,
+ "step": 70
+ },
+ {
+ "epoch": 2.1875,
+ "eval_loss": 0.7827157378196716,
+ "eval_runtime": 55.3708,
+ "eval_samples_per_second": 3.612,
+ "eval_steps_per_second": 0.235,
+ "step": 70
+ },
+ {
+ "epoch": 2.21875,
+ "grad_norm": 0.23059883325890385,
+ "learning_rate": 2e-05,
+ "loss": 0.8783,
+ "step": 71
+ },
+ {
+ "epoch": 2.21875,
+ "eval_loss": 0.7805906534194946,
+ "eval_runtime": 55.6033,
+ "eval_samples_per_second": 3.597,
+ "eval_steps_per_second": 0.234,
+ "step": 71
+ },
+ {
+ "epoch": 2.25,
+ "grad_norm": 0.23261007334915096,
+ "learning_rate": 2e-05,
+ "loss": 0.7956,
+ "step": 72
+ },
+ {
+ "epoch": 2.25,
+ "eval_loss": 0.7786691784858704,
+ "eval_runtime": 55.0913,
+ "eval_samples_per_second": 3.63,
+ "eval_steps_per_second": 0.236,
+ "step": 72
+ },
+ {
+ "epoch": 2.28125,
+ "grad_norm": 0.25779598356574085,
+ "learning_rate": 2e-05,
+ "loss": 0.8426,
+ "step": 73
+ },
+ {
+ "epoch": 2.28125,
+ "eval_loss": 0.7771151661872864,
+ "eval_runtime": 55.0698,
+ "eval_samples_per_second": 3.632,
+ "eval_steps_per_second": 0.236,
+ "step": 73
+ },
+ {
+ "epoch": 2.3125,
+ "grad_norm": 0.2288243335971112,
+ "learning_rate": 2e-05,
+ "loss": 0.8381,
+ "step": 74
+ },
+ {
+ "epoch": 2.3125,
+ "eval_loss": 0.7756838202476501,
+ "eval_runtime": 54.8412,
+ "eval_samples_per_second": 3.647,
+ "eval_steps_per_second": 0.237,
+ "step": 74
+ },
+ {
+ "epoch": 2.34375,
+ "grad_norm": 0.24235644907977733,
+ "learning_rate": 2e-05,
+ "loss": 0.887,
+ "step": 75
+ },
+ {
+ "epoch": 2.34375,
+ "eval_loss": 0.7739972472190857,
+ "eval_runtime": 54.9718,
+ "eval_samples_per_second": 3.638,
+ "eval_steps_per_second": 0.236,
+ "step": 75
+ },
+ {
+ "epoch": 2.375,
+ "grad_norm": 0.23666820017867402,
+ "learning_rate": 2e-05,
+ "loss": 0.8007,
+ "step": 76
+ },
+ {
+ "epoch": 2.375,
+ "eval_loss": 0.7724328637123108,
+ "eval_runtime": 55.0225,
+ "eval_samples_per_second": 3.635,
+ "eval_steps_per_second": 0.236,
+ "step": 76
+ },
+ {
+ "epoch": 2.40625,
+ "grad_norm": 0.22815737396609181,
+ "learning_rate": 2e-05,
+ "loss": 0.8529,
+ "step": 77
+ },
+ {
+ "epoch": 2.40625,
+ "eval_loss": 0.7710004448890686,
+ "eval_runtime": 55.321,
+ "eval_samples_per_second": 3.615,
+ "eval_steps_per_second": 0.235,
+ "step": 77
+ },
+ {
+ "epoch": 2.4375,
+ "grad_norm": 0.2701264871470739,
+ "learning_rate": 2e-05,
+ "loss": 0.8515,
+ "step": 78
+ },
+ {
+ "epoch": 2.4375,
+ "eval_loss": 0.7695322632789612,
+ "eval_runtime": 55.3045,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 78
+ },
+ {
+ "epoch": 2.46875,
+ "grad_norm": 0.24363813951328234,
+ "learning_rate": 2e-05,
+ "loss": 0.8587,
+ "step": 79
+ },
+ {
+ "epoch": 2.46875,
+ "eval_loss": 0.7689024209976196,
+ "eval_runtime": 55.3009,
+ "eval_samples_per_second": 3.617,
+ "eval_steps_per_second": 0.235,
+ "step": 79
+ },
+ {
+ "epoch": 2.5,
+ "grad_norm": 0.30924701355253065,
+ "learning_rate": 2e-05,
+ "loss": 0.9076,
+ "step": 80
+ },
+ {
+ "epoch": 2.5,
+ "eval_loss": 0.7676254510879517,
+ "eval_runtime": 55.2365,
+ "eval_samples_per_second": 3.621,
+ "eval_steps_per_second": 0.235,
+ "step": 80
+ },
+ {
+ "epoch": 2.53125,
+ "grad_norm": 0.2665188280221636,
+ "learning_rate": 2e-05,
+ "loss": 0.8445,
+ "step": 81
+ },
+ {
+ "epoch": 2.53125,
+ "eval_loss": 0.7661146521568298,
+ "eval_runtime": 55.2775,
+ "eval_samples_per_second": 3.618,
+ "eval_steps_per_second": 0.235,
+ "step": 81
+ },
+ {
+ "epoch": 2.5625,
+ "grad_norm": 0.24674191720675534,
+ "learning_rate": 2e-05,
+ "loss": 0.8882,
+ "step": 82
+ },
+ {
+ "epoch": 2.5625,
+ "eval_loss": 0.76513671875,
+ "eval_runtime": 55.0857,
+ "eval_samples_per_second": 3.631,
+ "eval_steps_per_second": 0.236,
+ "step": 82
+ },
+ {
+ "epoch": 2.59375,
+ "grad_norm": 0.2736689405531704,
+ "learning_rate": 2e-05,
+ "loss": 0.8336,
+ "step": 83
+ },
+ {
+ "epoch": 2.59375,
+ "eval_loss": 0.764373779296875,
+ "eval_runtime": 55.2069,
+ "eval_samples_per_second": 3.623,
+ "eval_steps_per_second": 0.235,
+ "step": 83
+ },
+ {
+ "epoch": 2.625,
+ "grad_norm": 0.290841287198557,
+ "learning_rate": 2e-05,
+ "loss": 0.795,
+ "step": 84
+ },
+ {
+ "epoch": 2.625,
+ "eval_loss": 0.7632084488868713,
+ "eval_runtime": 55.1009,
+ "eval_samples_per_second": 3.63,
+ "eval_steps_per_second": 0.236,
+ "step": 84
+ },
+ {
+ "epoch": 2.65625,
+ "grad_norm": 0.2912051076836381,
+ "learning_rate": 2e-05,
+ "loss": 0.772,
+ "step": 85
+ },
+ {
+ "epoch": 2.65625,
+ "eval_loss": 0.7618446350097656,
+ "eval_runtime": 55.3717,
+ "eval_samples_per_second": 3.612,
+ "eval_steps_per_second": 0.235,
+ "step": 85
+ },
+ {
+ "epoch": 2.6875,
+ "grad_norm": 0.3169908538809109,
+ "learning_rate": 2e-05,
+ "loss": 0.8148,
+ "step": 86
+ },
+ {
+ "epoch": 2.6875,
+ "eval_loss": 0.7599577307701111,
+ "eval_runtime": 55.3931,
+ "eval_samples_per_second": 3.611,
+ "eval_steps_per_second": 0.235,
+ "step": 86
+ },
+ {
+ "epoch": 2.71875,
+ "grad_norm": 0.28780549186847426,
+ "learning_rate": 2e-05,
+ "loss": 0.8154,
+ "step": 87
+ },
+ {
+ "epoch": 2.71875,
+ "eval_loss": 0.7583369612693787,
+ "eval_runtime": 55.1679,
+ "eval_samples_per_second": 3.625,
+ "eval_steps_per_second": 0.236,
+ "step": 87
+ },
+ {
+ "epoch": 2.75,
+ "grad_norm": 0.30695250620091474,
+ "learning_rate": 2e-05,
+ "loss": 0.9032,
+ "step": 88
+ },
+ {
+ "epoch": 2.75,
+ "eval_loss": 0.7571613192558289,
+ "eval_runtime": 55.1779,
+ "eval_samples_per_second": 3.625,
+ "eval_steps_per_second": 0.236,
+ "step": 88
+ },
+ {
+ "epoch": 2.78125,
+ "grad_norm": 0.2693887416759828,
+ "learning_rate": 2e-05,
+ "loss": 0.8106,
+ "step": 89
+ },
+ {
+ "epoch": 2.78125,
+ "eval_loss": 0.7566004991531372,
+ "eval_runtime": 55.1107,
+ "eval_samples_per_second": 3.629,
+ "eval_steps_per_second": 0.236,
+ "step": 89
+ },
+ {
+ "epoch": 2.8125,
+ "grad_norm": 0.2887583627563198,
+ "learning_rate": 2e-05,
+ "loss": 0.8518,
+ "step": 90
+ },
+ {
+ "epoch": 2.8125,
+ "eval_loss": 0.7558963298797607,
+ "eval_runtime": 55.2153,
+ "eval_samples_per_second": 3.622,
+ "eval_steps_per_second": 0.235,
+ "step": 90
+ },
+ {
+ "epoch": 2.84375,
+ "grad_norm": 0.3059402168979351,
+ "learning_rate": 2e-05,
+ "loss": 0.7727,
+ "step": 91
+ },
+ {
+ "epoch": 2.84375,
+ "eval_loss": 0.7545350790023804,
+ "eval_runtime": 55.3225,
+ "eval_samples_per_second": 3.615,
+ "eval_steps_per_second": 0.235,
+ "step": 91
+ },
+ {
+ "epoch": 2.875,
+ "grad_norm": 0.3096260477909968,
+ "learning_rate": 2e-05,
+ "loss": 0.8477,
+ "step": 92
+ },
+ {
+ "epoch": 2.875,
+ "eval_loss": 0.7526452541351318,
+ "eval_runtime": 55.4311,
+ "eval_samples_per_second": 3.608,
+ "eval_steps_per_second": 0.235,
+ "step": 92
+ },
+ {
+ "epoch": 2.90625,
+ "grad_norm": 0.31498884686525297,
+ "learning_rate": 2e-05,
+ "loss": 0.7982,
+ "step": 93
+ },
+ {
+ "epoch": 2.90625,
+ "eval_loss": 0.7510760426521301,
+ "eval_runtime": 55.4361,
+ "eval_samples_per_second": 3.608,
+ "eval_steps_per_second": 0.235,
+ "step": 93
+ },
+ {
+ "epoch": 2.9375,
+ "grad_norm": 0.31302830623184313,
+ "learning_rate": 2e-05,
+ "loss": 0.871,
+ "step": 94
+ },
+ {
+ "epoch": 2.9375,
+ "eval_loss": 0.7500898838043213,
+ "eval_runtime": 55.3025,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 94
+ },
+ {
+ "epoch": 2.96875,
+ "grad_norm": 0.3132608568779145,
+ "learning_rate": 2e-05,
+ "loss": 0.8094,
+ "step": 95
+ },
+ {
+ "epoch": 2.96875,
+ "eval_loss": 0.7498895525932312,
+ "eval_runtime": 55.2402,
+ "eval_samples_per_second": 3.621,
+ "eval_steps_per_second": 0.235,
+ "step": 95
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 0.298645350091386,
+ "learning_rate": 2e-05,
+ "loss": 0.7673,
+ "step": 96
+ },
+ {
+ "epoch": 3.0,
+ "eval_loss": 0.7493192553520203,
+ "eval_runtime": 54.8718,
+ "eval_samples_per_second": 3.645,
+ "eval_steps_per_second": 0.237,
+ "step": 96
+ },
+ {
+ "epoch": 3.03125,
+ "grad_norm": 0.34042584783125357,
+ "learning_rate": 2e-05,
+ "loss": 0.7336,
+ "step": 97
+ },
+ {
+ "epoch": 3.03125,
+ "eval_loss": 0.7476670742034912,
+ "eval_runtime": 54.9305,
+ "eval_samples_per_second": 3.641,
+ "eval_steps_per_second": 0.237,
+ "step": 97
+ },
+ {
+ "epoch": 3.0625,
+ "grad_norm": 0.293099043801068,
+ "learning_rate": 2e-05,
+ "loss": 0.8088,
+ "step": 98
+ },
+ {
+ "epoch": 3.0625,
+ "eval_loss": 0.745802640914917,
+ "eval_runtime": 55.2051,
+ "eval_samples_per_second": 3.623,
+ "eval_steps_per_second": 0.235,
+ "step": 98
+ },
+ {
+ "epoch": 3.09375,
+ "grad_norm": 0.3042839507858426,
+ "learning_rate": 2e-05,
+ "loss": 0.787,
+ "step": 99
+ },
+ {
+ "epoch": 3.09375,
+ "eval_loss": 0.7439618110656738,
+ "eval_runtime": 55.0065,
+ "eval_samples_per_second": 3.636,
+ "eval_steps_per_second": 0.236,
+ "step": 99
+ },
+ {
+ "epoch": 3.125,
+ "grad_norm": 0.32992077073227005,
+ "learning_rate": 2e-05,
+ "loss": 0.8296,
+ "step": 100
+ },
+ {
+ "epoch": 3.125,
+ "eval_loss": 0.7424842715263367,
+ "eval_runtime": 55.1254,
+ "eval_samples_per_second": 3.628,
+ "eval_steps_per_second": 0.236,
+ "step": 100
+ },
+ {
+ "epoch": 3.15625,
+ "grad_norm": 0.2798839747424062,
+ "learning_rate": 2e-05,
+ "loss": 0.7642,
+ "step": 101
+ },
+ {
+ "epoch": 3.15625,
+ "eval_loss": 0.7414796948432922,
+ "eval_runtime": 49.183,
+ "eval_samples_per_second": 4.066,
+ "eval_steps_per_second": 0.264,
+ "step": 101
+ },
+ {
+ "epoch": 3.1875,
+ "grad_norm": 0.3046631191964983,
+ "learning_rate": 2e-05,
+ "loss": 0.8203,
+ "step": 102
+ },
+ {
+ "epoch": 3.1875,
+ "eval_loss": 0.7410265207290649,
+ "eval_runtime": 48.1541,
+ "eval_samples_per_second": 4.153,
+ "eval_steps_per_second": 0.27,
+ "step": 102
+ },
+ {
+ "epoch": 3.21875,
+ "grad_norm": 0.3117517214859861,
+ "learning_rate": 2e-05,
+ "loss": 0.8222,
+ "step": 103
+ },
+ {
+ "epoch": 3.21875,
+ "eval_loss": 0.7405675649642944,
+ "eval_runtime": 47.7145,
+ "eval_samples_per_second": 4.192,
+ "eval_steps_per_second": 0.272,
+ "step": 103
+ },
+ {
+ "epoch": 3.25,
+ "grad_norm": 0.3412709249466801,
+ "learning_rate": 2e-05,
+ "loss": 0.7459,
+ "step": 104
+ },
+ {
+ "epoch": 3.25,
+ "eval_loss": 0.7395681738853455,
+ "eval_runtime": 47.5855,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 104
+ },
+ {
+ "epoch": 3.28125,
+ "grad_norm": 0.2917443566507923,
+ "learning_rate": 2e-05,
+ "loss": 0.7849,
+ "step": 105
+ },
+ {
+ "epoch": 3.28125,
+ "eval_loss": 0.7387100458145142,
+ "eval_runtime": 47.6344,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 105
+ },
+ {
+ "epoch": 3.3125,
+ "grad_norm": 0.3054484743574741,
+ "learning_rate": 2e-05,
+ "loss": 0.8354,
+ "step": 106
+ },
+ {
+ "epoch": 3.3125,
+ "eval_loss": 0.7384718060493469,
+ "eval_runtime": 47.8373,
+ "eval_samples_per_second": 4.181,
+ "eval_steps_per_second": 0.272,
+ "step": 106
+ },
+ {
+ "epoch": 3.34375,
+ "grad_norm": 0.34986630381114014,
+ "learning_rate": 2e-05,
+ "loss": 0.7069,
+ "step": 107
+ },
+ {
+ "epoch": 3.34375,
+ "eval_loss": 0.737342357635498,
+ "eval_runtime": 47.5763,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 107
+ },
+ {
+ "epoch": 3.375,
+ "grad_norm": 0.32324403145716496,
+ "learning_rate": 2e-05,
+ "loss": 0.767,
+ "step": 108
+ },
+ {
+ "epoch": 3.375,
+ "eval_loss": 0.7360101938247681,
+ "eval_runtime": 47.5774,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 108
+ },
+ {
+ "epoch": 3.40625,
+ "grad_norm": 0.3795969851258545,
+ "learning_rate": 2e-05,
+ "loss": 0.7556,
+ "step": 109
+ },
+ {
+ "epoch": 3.40625,
+ "eval_loss": 0.7339167594909668,
+ "eval_runtime": 47.5818,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 109
+ },
+ {
+ "epoch": 3.4375,
+ "grad_norm": 0.34401062275458993,
+ "learning_rate": 2e-05,
+ "loss": 0.7494,
+ "step": 110
+ },
+ {
+ "epoch": 3.4375,
+ "eval_loss": 0.7321068644523621,
+ "eval_runtime": 47.7643,
+ "eval_samples_per_second": 4.187,
+ "eval_steps_per_second": 0.272,
+ "step": 110
+ },
+ {
+ "epoch": 3.46875,
+ "grad_norm": 0.3248480010385237,
+ "learning_rate": 2e-05,
+ "loss": 0.8103,
+ "step": 111
+ },
+ {
+ "epoch": 3.46875,
+ "eval_loss": 0.7309197783470154,
+ "eval_runtime": 49.5841,
+ "eval_samples_per_second": 4.034,
+ "eval_steps_per_second": 0.262,
+ "step": 111
+ },
+ {
+ "epoch": 3.5,
+ "grad_norm": 0.3572409124813593,
+ "learning_rate": 2e-05,
+ "loss": 0.7972,
+ "step": 112
+ },
+ {
+ "epoch": 3.5,
+ "eval_loss": 0.7301727533340454,
+ "eval_runtime": 49.3728,
+ "eval_samples_per_second": 4.051,
+ "eval_steps_per_second": 0.263,
+ "step": 112
+ },
+ {
+ "epoch": 3.53125,
+ "grad_norm": 0.37348522775103665,
+ "learning_rate": 2e-05,
+ "loss": 0.88,
+ "step": 113
+ },
+ {
+ "epoch": 3.53125,
+ "eval_loss": 0.7292957305908203,
+ "eval_runtime": 49.2192,
+ "eval_samples_per_second": 4.063,
+ "eval_steps_per_second": 0.264,
+ "step": 113
+ },
+ {
+ "epoch": 3.5625,
+ "grad_norm": 0.37667450960329546,
+ "learning_rate": 2e-05,
+ "loss": 0.7518,
+ "step": 114
+ },
+ {
+ "epoch": 3.5625,
+ "eval_loss": 0.728556215763092,
+ "eval_runtime": 49.0971,
+ "eval_samples_per_second": 4.074,
+ "eval_steps_per_second": 0.265,
+ "step": 114
+ },
+ {
+ "epoch": 3.59375,
+ "grad_norm": 0.3163628607304638,
+ "learning_rate": 2e-05,
+ "loss": 0.7948,
+ "step": 115
+ },
+ {
+ "epoch": 3.59375,
+ "eval_loss": 0.7287828326225281,
+ "eval_runtime": 49.0213,
+ "eval_samples_per_second": 4.08,
+ "eval_steps_per_second": 0.265,
+ "step": 115
+ },
+ {
+ "epoch": 3.625,
+ "grad_norm": 0.3038899302084592,
+ "learning_rate": 2e-05,
+ "loss": 0.7791,
+ "step": 116
+ },
+ {
+ "epoch": 3.625,
+ "eval_loss": 0.7294514179229736,
+ "eval_runtime": 51.9137,
+ "eval_samples_per_second": 3.853,
+ "eval_steps_per_second": 0.25,
+ "step": 116
+ },
+ {
+ "epoch": 3.65625,
+ "grad_norm": 0.3746448663122327,
+ "learning_rate": 2e-05,
+ "loss": 0.7863,
+ "step": 117
+ },
+ {
+ "epoch": 3.65625,
+ "eval_loss": 0.7289304137229919,
+ "eval_runtime": 51.3023,
+ "eval_samples_per_second": 3.898,
+ "eval_steps_per_second": 0.253,
+ "step": 117
+ },
+ {
+ "epoch": 3.6875,
+ "grad_norm": 0.4058937381299434,
+ "learning_rate": 2e-05,
+ "loss": 0.7907,
+ "step": 118
+ },
+ {
+ "epoch": 3.6875,
+ "eval_loss": 0.7281011343002319,
+ "eval_runtime": 50.8635,
+ "eval_samples_per_second": 3.932,
+ "eval_steps_per_second": 0.256,
+ "step": 118
+ },
+ {
+ "epoch": 3.71875,
+ "grad_norm": 0.31608065583227885,
+ "learning_rate": 2e-05,
+ "loss": 0.8348,
+ "step": 119
+ },
+ {
+ "epoch": 3.71875,
+ "eval_loss": 0.7280247211456299,
+ "eval_runtime": 50.4903,
+ "eval_samples_per_second": 3.961,
+ "eval_steps_per_second": 0.257,
+ "step": 119
+ },
+ {
+ "epoch": 3.75,
+ "grad_norm": 0.3375768031046084,
+ "learning_rate": 2e-05,
+ "loss": 0.7783,
+ "step": 120
+ },
+ {
+ "epoch": 3.75,
+ "eval_loss": 0.7281913757324219,
+ "eval_runtime": 50.5906,
+ "eval_samples_per_second": 3.953,
+ "eval_steps_per_second": 0.257,
+ "step": 120
+ },
+ {
+ "epoch": 3.78125,
+ "grad_norm": 0.36047493494859845,
+ "learning_rate": 2e-05,
+ "loss": 0.765,
+ "step": 121
+ },
+ {
+ "epoch": 3.78125,
+ "eval_loss": 0.7269737124443054,
+ "eval_runtime": 53.4722,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 121
+ },
+ {
+ "epoch": 3.8125,
+ "grad_norm": 0.389743860171921,
+ "learning_rate": 2e-05,
+ "loss": 0.8269,
+ "step": 122
+ },
+ {
+ "epoch": 3.8125,
+ "eval_loss": 0.7251996397972107,
+ "eval_runtime": 53.4986,
+ "eval_samples_per_second": 3.738,
+ "eval_steps_per_second": 0.243,
+ "step": 122
+ },
+ {
+ "epoch": 3.84375,
+ "grad_norm": 0.33850935145960215,
+ "learning_rate": 2e-05,
+ "loss": 0.7497,
+ "step": 123
+ },
+ {
+ "epoch": 3.84375,
+ "eval_loss": 0.723595142364502,
+ "eval_runtime": 53.4196,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 123
+ },
+ {
+ "epoch": 3.875,
+ "grad_norm": 0.3166770012114478,
+ "learning_rate": 2e-05,
+ "loss": 0.7648,
+ "step": 124
+ },
+ {
+ "epoch": 3.875,
+ "eval_loss": 0.7223578095436096,
+ "eval_runtime": 52.6143,
+ "eval_samples_per_second": 3.801,
+ "eval_steps_per_second": 0.247,
+ "step": 124
+ },
+ {
+ "epoch": 3.90625,
+ "grad_norm": 0.41948670305268276,
+ "learning_rate": 2e-05,
+ "loss": 0.8306,
+ "step": 125
+ },
+ {
+ "epoch": 3.90625,
+ "eval_loss": 0.7206680774688721,
+ "eval_runtime": 52.3885,
+ "eval_samples_per_second": 3.818,
+ "eval_steps_per_second": 0.248,
+ "step": 125
+ },
+ {
+ "epoch": 3.9375,
+ "grad_norm": 0.35580041105853477,
+ "learning_rate": 2e-05,
+ "loss": 0.7945,
+ "step": 126
+ },
+ {
+ "epoch": 3.9375,
+ "eval_loss": 0.7196171283721924,
+ "eval_runtime": 55.1225,
+ "eval_samples_per_second": 3.628,
+ "eval_steps_per_second": 0.236,
+ "step": 126
+ },
+ {
+ "epoch": 3.96875,
+ "grad_norm": 0.38411890663257114,
+ "learning_rate": 2e-05,
+ "loss": 0.7466,
+ "step": 127
+ },
+ {
+ "epoch": 3.96875,
+ "eval_loss": 0.7188088297843933,
+ "eval_runtime": 55.3068,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 127
+ },
+ {
+ "epoch": 4.0,
+ "grad_norm": 0.3682220575203032,
+ "learning_rate": 2e-05,
+ "loss": 0.6752,
+ "step": 128
+ },
+ {
+ "epoch": 4.0,
+ "eval_loss": 0.7181470990180969,
+ "eval_runtime": 53.9116,
+ "eval_samples_per_second": 3.71,
+ "eval_steps_per_second": 0.241,
+ "step": 128
+ },
+ {
+ "epoch": 4.03125,
+ "grad_norm": 0.34160763542661665,
+ "learning_rate": 2e-05,
+ "loss": 0.7788,
+ "step": 129
+ },
+ {
+ "epoch": 4.03125,
+ "eval_loss": 0.717949390411377,
+ "eval_runtime": 53.8446,
+ "eval_samples_per_second": 3.714,
+ "eval_steps_per_second": 0.241,
+ "step": 129
+ },
+ {
+ "epoch": 4.0625,
+ "grad_norm": 0.35709301353799944,
+ "learning_rate": 2e-05,
+ "loss": 0.8002,
+ "step": 130
+ },
+ {
+ "epoch": 4.0625,
+ "eval_loss": 0.7179380655288696,
+ "eval_runtime": 53.9299,
+ "eval_samples_per_second": 3.709,
+ "eval_steps_per_second": 0.241,
+ "step": 130
+ },
+ {
+ "epoch": 4.09375,
+ "grad_norm": 0.3503147340749238,
+ "learning_rate": 2e-05,
+ "loss": 0.7789,
+ "step": 131
+ },
+ {
+ "epoch": 4.09375,
+ "eval_loss": 0.7180312871932983,
+ "eval_runtime": 53.4091,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 131
+ },
+ {
+ "epoch": 4.125,
+ "grad_norm": 0.3931715546229069,
+ "learning_rate": 2e-05,
+ "loss": 0.762,
+ "step": 132
+ },
+ {
+ "epoch": 4.125,
+ "eval_loss": 0.717825710773468,
+ "eval_runtime": 53.6366,
+ "eval_samples_per_second": 3.729,
+ "eval_steps_per_second": 0.242,
+ "step": 132
+ },
+ {
+ "epoch": 4.15625,
+ "grad_norm": 0.36864033862644363,
+ "learning_rate": 2e-05,
+ "loss": 0.829,
+ "step": 133
+ },
+ {
+ "epoch": 4.15625,
+ "eval_loss": 0.7178698182106018,
+ "eval_runtime": 53.4891,
+ "eval_samples_per_second": 3.739,
+ "eval_steps_per_second": 0.243,
+ "step": 133
+ },
+ {
+ "epoch": 4.1875,
+ "grad_norm": 0.41393587587462155,
+ "learning_rate": 2e-05,
+ "loss": 0.7624,
+ "step": 134
+ },
+ {
+ "epoch": 4.1875,
+ "eval_loss": 0.7181968092918396,
+ "eval_runtime": 53.5395,
+ "eval_samples_per_second": 3.736,
+ "eval_steps_per_second": 0.243,
+ "step": 134
+ },
+ {
+ "epoch": 4.21875,
+ "grad_norm": 0.36727603900023204,
+ "learning_rate": 2e-05,
+ "loss": 0.7572,
+ "step": 135
+ },
+ {
+ "epoch": 4.21875,
+ "eval_loss": 0.7187527418136597,
+ "eval_runtime": 53.4818,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 135
+ },
+ {
+ "epoch": 4.25,
+ "grad_norm": 0.3684078795455007,
+ "learning_rate": 2e-05,
+ "loss": 0.7352,
+ "step": 136
+ },
+ {
+ "epoch": 4.25,
+ "eval_loss": 0.7194793820381165,
+ "eval_runtime": 53.4694,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 136
+ },
+ {
+ "epoch": 4.28125,
+ "grad_norm": 0.42414766562621153,
+ "learning_rate": 2e-05,
+ "loss": 0.7433,
+ "step": 137
+ },
+ {
+ "epoch": 4.28125,
+ "eval_loss": 0.7189603447914124,
+ "eval_runtime": 53.8049,
+ "eval_samples_per_second": 3.717,
+ "eval_steps_per_second": 0.242,
+ "step": 137
+ },
+ {
+ "epoch": 4.3125,
+ "grad_norm": 0.40420796619211563,
+ "learning_rate": 2e-05,
+ "loss": 0.7466,
+ "step": 138
+ },
+ {
+ "epoch": 4.3125,
+ "eval_loss": 0.7173956036567688,
+ "eval_runtime": 53.4014,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 138
+ },
+ {
+ "epoch": 4.34375,
+ "grad_norm": 0.36419740641344456,
+ "learning_rate": 2e-05,
+ "loss": 0.7045,
+ "step": 139
+ },
+ {
+ "epoch": 4.34375,
+ "eval_loss": 0.7153105139732361,
+ "eval_runtime": 53.285,
+ "eval_samples_per_second": 3.753,
+ "eval_steps_per_second": 0.244,
+ "step": 139
+ },
+ {
+ "epoch": 4.375,
+ "grad_norm": 0.384927357409491,
+ "learning_rate": 2e-05,
+ "loss": 0.7437,
+ "step": 140
+ },
+ {
+ "epoch": 4.375,
+ "eval_loss": 0.7135314345359802,
+ "eval_runtime": 53.4056,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 140
+ },
+ {
+ "epoch": 4.40625,
+ "grad_norm": 0.37218579680263697,
+ "learning_rate": 2e-05,
+ "loss": 0.7693,
+ "step": 141
+ },
+ {
+ "epoch": 4.40625,
+ "eval_loss": 0.7120725512504578,
+ "eval_runtime": 53.5467,
+ "eval_samples_per_second": 3.735,
+ "eval_steps_per_second": 0.243,
+ "step": 141
+ },
+ {
+ "epoch": 4.4375,
+ "grad_norm": 0.38541382926033946,
+ "learning_rate": 2e-05,
+ "loss": 0.708,
+ "step": 142
+ },
+ {
+ "epoch": 4.4375,
+ "eval_loss": 0.7110380530357361,
+ "eval_runtime": 53.4119,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 142
+ },
+ {
+ "epoch": 4.46875,
+ "grad_norm": 0.4028726453247759,
+ "learning_rate": 2e-05,
+ "loss": 0.7263,
+ "step": 143
+ },
+ {
+ "epoch": 4.46875,
+ "eval_loss": 0.7100683450698853,
+ "eval_runtime": 53.4337,
+ "eval_samples_per_second": 3.743,
+ "eval_steps_per_second": 0.243,
+ "step": 143
+ },
+ {
+ "epoch": 4.5,
+ "grad_norm": 0.3736204162232246,
+ "learning_rate": 2e-05,
+ "loss": 0.698,
+ "step": 144
+ },
+ {
+ "epoch": 4.5,
+ "eval_loss": 0.7093971371650696,
+ "eval_runtime": 53.4582,
+ "eval_samples_per_second": 3.741,
+ "eval_steps_per_second": 0.243,
+ "step": 144
+ },
+ {
+ "epoch": 4.53125,
+ "grad_norm": 0.4179284798304916,
+ "learning_rate": 2e-05,
+ "loss": 0.7611,
+ "step": 145
+ },
+ {
+ "epoch": 4.53125,
+ "eval_loss": 0.7089446783065796,
+ "eval_runtime": 53.4752,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 145
+ },
+ {
+ "epoch": 4.5625,
+ "grad_norm": 0.4038858950888911,
+ "learning_rate": 2e-05,
+ "loss": 0.6652,
+ "step": 146
+ },
+ {
+ "epoch": 4.5625,
+ "eval_loss": 0.7089542150497437,
+ "eval_runtime": 53.4741,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 146
+ },
+ {
+ "epoch": 4.59375,
+ "grad_norm": 0.41740068710674544,
+ "learning_rate": 2e-05,
+ "loss": 0.7319,
+ "step": 147
+ },
+ {
+ "epoch": 4.59375,
+ "eval_loss": 0.7090431451797485,
+ "eval_runtime": 53.2419,
+ "eval_samples_per_second": 3.756,
+ "eval_steps_per_second": 0.244,
+ "step": 147
+ },
+ {
+ "epoch": 4.625,
+ "grad_norm": 0.4288335811568808,
+ "learning_rate": 2e-05,
+ "loss": 0.6837,
+ "step": 148
+ },
+ {
+ "epoch": 4.625,
+ "eval_loss": 0.7088204026222229,
+ "eval_runtime": 53.3614,
+ "eval_samples_per_second": 3.748,
+ "eval_steps_per_second": 0.244,
+ "step": 148
+ },
+ {
+ "epoch": 4.65625,
+ "grad_norm": 0.399955010119186,
+ "learning_rate": 2e-05,
+ "loss": 0.7989,
+ "step": 149
+ },
+ {
+ "epoch": 4.65625,
+ "eval_loss": 0.7084855437278748,
+ "eval_runtime": 53.4923,
+ "eval_samples_per_second": 3.739,
+ "eval_steps_per_second": 0.243,
+ "step": 149
+ },
+ {
+ "epoch": 4.6875,
+ "grad_norm": 0.41794643164255846,
+ "learning_rate": 2e-05,
+ "loss": 0.7194,
+ "step": 150
+ },
+ {
+ "epoch": 4.6875,
+ "eval_loss": 0.7080708146095276,
+ "eval_runtime": 53.639,
+ "eval_samples_per_second": 3.729,
+ "eval_steps_per_second": 0.242,
+ "step": 150
+ },
+ {
+ "epoch": 4.71875,
+ "grad_norm": 0.40953367303148197,
+ "learning_rate": 2e-05,
+ "loss": 0.7354,
+ "step": 151
+ },
+ {
+ "epoch": 4.71875,
+ "eval_loss": 0.7077429890632629,
+ "eval_runtime": 53.3837,
+ "eval_samples_per_second": 3.746,
+ "eval_steps_per_second": 0.244,
+ "step": 151
+ },
+ {
+ "epoch": 4.75,
+ "grad_norm": 0.5012282841513718,
+ "learning_rate": 2e-05,
+ "loss": 0.7662,
+ "step": 152
+ },
+ {
+ "epoch": 4.75,
+ "eval_loss": 0.7064151167869568,
+ "eval_runtime": 53.3549,
+ "eval_samples_per_second": 3.748,
+ "eval_steps_per_second": 0.244,
+ "step": 152
+ },
+ {
+ "epoch": 4.78125,
+ "grad_norm": 0.4210784420989087,
+ "learning_rate": 2e-05,
+ "loss": 0.7133,
+ "step": 153
+ },
+ {
+ "epoch": 4.78125,
+ "eval_loss": 0.7052726745605469,
+ "eval_runtime": 53.5059,
+ "eval_samples_per_second": 3.738,
+ "eval_steps_per_second": 0.243,
+ "step": 153
+ },
+ {
+ "epoch": 4.8125,
+ "grad_norm": 0.43520348530514996,
+ "learning_rate": 2e-05,
+ "loss": 0.729,
+ "step": 154
+ },
+ {
+ "epoch": 4.8125,
+ "eval_loss": 0.7045274972915649,
+ "eval_runtime": 53.8352,
+ "eval_samples_per_second": 3.715,
+ "eval_steps_per_second": 0.241,
+ "step": 154
+ },
+ {
+ "epoch": 4.84375,
+ "grad_norm": 0.4287647569802656,
+ "learning_rate": 2e-05,
+ "loss": 0.6727,
+ "step": 155
+ },
+ {
+ "epoch": 4.84375,
+ "eval_loss": 0.7041358947753906,
+ "eval_runtime": 53.7435,
+ "eval_samples_per_second": 3.721,
+ "eval_steps_per_second": 0.242,
+ "step": 155
+ },
+ {
+ "epoch": 4.875,
+ "grad_norm": 0.41883715320456333,
+ "learning_rate": 2e-05,
+ "loss": 0.7755,
+ "step": 156
+ },
+ {
+ "epoch": 4.875,
+ "eval_loss": 0.7037128210067749,
+ "eval_runtime": 53.8035,
+ "eval_samples_per_second": 3.717,
+ "eval_steps_per_second": 0.242,
+ "step": 156
+ },
+ {
+ "epoch": 4.90625,
+ "grad_norm": 0.40617584505395354,
+ "learning_rate": 2e-05,
+ "loss": 0.7776,
+ "step": 157
+ },
+ {
+ "epoch": 4.90625,
+ "eval_loss": 0.703965425491333,
+ "eval_runtime": 53.8731,
+ "eval_samples_per_second": 3.712,
+ "eval_steps_per_second": 0.241,
+ "step": 157
+ },
+ {
+ "epoch": 4.9375,
+ "grad_norm": 0.4085802225532245,
+ "learning_rate": 2e-05,
+ "loss": 0.7628,
+ "step": 158
+ },
+ {
+ "epoch": 4.9375,
+ "eval_loss": 0.7040860056877136,
+ "eval_runtime": 53.9059,
+ "eval_samples_per_second": 3.71,
+ "eval_steps_per_second": 0.241,
+ "step": 158
+ },
+ {
+ "epoch": 4.96875,
+ "grad_norm": 0.418039298119887,
+ "learning_rate": 2e-05,
+ "loss": 0.7221,
+ "step": 159
+ },
+ {
+ "epoch": 4.96875,
+ "eval_loss": 0.7039948105812073,
+ "eval_runtime": 53.7323,
+ "eval_samples_per_second": 3.722,
+ "eval_steps_per_second": 0.242,
+ "step": 159
+ },
+ {
+ "epoch": 5.0,
+ "grad_norm": 0.46118870048713073,
+ "learning_rate": 2e-05,
+ "loss": 0.7029,
+ "step": 160
+ },
+ {
+ "epoch": 5.0,
+ "eval_loss": 0.703814685344696,
+ "eval_runtime": 53.8975,
+ "eval_samples_per_second": 3.711,
+ "eval_steps_per_second": 0.241,
+ "step": 160
+ },
+ {
+ "epoch": 5.03125,
+ "grad_norm": 0.431474386110294,
+ "learning_rate": 2e-05,
+ "loss": 0.6772,
+ "step": 161
+ },
+ {
+ "epoch": 5.03125,
+ "eval_loss": 0.7034456133842468,
+ "eval_runtime": 51.1105,
+ "eval_samples_per_second": 3.913,
+ "eval_steps_per_second": 0.254,
+ "step": 161
+ },
+ {
+ "epoch": 5.0625,
+ "grad_norm": 0.39618929325750435,
+ "learning_rate": 2e-05,
+ "loss": 0.8219,
+ "step": 162
+ },
+ {
+ "epoch": 5.0625,
+ "eval_loss": 0.7042189240455627,
+ "eval_runtime": 47.2927,
+ "eval_samples_per_second": 4.229,
+ "eval_steps_per_second": 0.275,
+ "step": 162
+ },
+ {
+ "epoch": 5.09375,
+ "grad_norm": 0.4489132713249424,
+ "learning_rate": 2e-05,
+ "loss": 0.6387,
+ "step": 163
+ },
+ {
+ "epoch": 5.09375,
+ "eval_loss": 0.7061256170272827,
+ "eval_runtime": 47.387,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 163
+ },
+ {
+ "epoch": 5.125,
+ "grad_norm": 0.5100329637159183,
+ "learning_rate": 2e-05,
+ "loss": 0.7677,
+ "step": 164
+ },
+ {
+ "epoch": 5.125,
+ "eval_loss": 0.708121657371521,
+ "eval_runtime": 47.3311,
+ "eval_samples_per_second": 4.226,
+ "eval_steps_per_second": 0.275,
+ "step": 164
+ },
+ {
+ "epoch": 5.15625,
+ "grad_norm": 0.525511631981176,
+ "learning_rate": 2e-05,
+ "loss": 0.5956,
+ "step": 165
+ },
+ {
+ "epoch": 5.15625,
+ "eval_loss": 0.7091134786605835,
+ "eval_runtime": 47.2978,
+ "eval_samples_per_second": 4.229,
+ "eval_steps_per_second": 0.275,
+ "step": 165
+ },
+ {
+ "epoch": 5.1875,
+ "grad_norm": 0.534675354231597,
+ "learning_rate": 2e-05,
+ "loss": 0.7097,
+ "step": 166
+ },
+ {
+ "epoch": 5.1875,
+ "eval_loss": 0.7097848653793335,
+ "eval_runtime": 47.4095,
+ "eval_samples_per_second": 4.219,
+ "eval_steps_per_second": 0.274,
+ "step": 166
+ },
+ {
+ "epoch": 5.21875,
+ "grad_norm": 0.47286903698857446,
+ "learning_rate": 2e-05,
+ "loss": 0.7371,
+ "step": 167
+ },
+ {
+ "epoch": 5.21875,
+ "eval_loss": 0.7090296745300293,
+ "eval_runtime": 47.4487,
+ "eval_samples_per_second": 4.215,
+ "eval_steps_per_second": 0.274,
+ "step": 167
+ },
+ {
+ "epoch": 5.25,
+ "grad_norm": 0.4734705066820788,
+ "learning_rate": 2e-05,
+ "loss": 0.7652,
+ "step": 168
+ },
+ {
+ "epoch": 5.25,
+ "eval_loss": 0.7079525589942932,
+ "eval_runtime": 47.4101,
+ "eval_samples_per_second": 4.219,
+ "eval_steps_per_second": 0.274,
+ "step": 168
+ },
+ {
+ "epoch": 5.28125,
+ "grad_norm": 0.46209764763985184,
+ "learning_rate": 2e-05,
+ "loss": 0.6852,
+ "step": 169
+ },
+ {
+ "epoch": 5.28125,
+ "eval_loss": 0.7072803974151611,
+ "eval_runtime": 47.3704,
+ "eval_samples_per_second": 4.222,
+ "eval_steps_per_second": 0.274,
+ "step": 169
+ },
+ {
+ "epoch": 5.3125,
+ "grad_norm": 0.4828284708486433,
+ "learning_rate": 2e-05,
+ "loss": 0.6609,
+ "step": 170
+ },
+ {
+ "epoch": 5.3125,
+ "eval_loss": 0.7068901062011719,
+ "eval_runtime": 47.425,
+ "eval_samples_per_second": 4.217,
+ "eval_steps_per_second": 0.274,
+ "step": 170
+ },
+ {
+ "epoch": 5.34375,
+ "grad_norm": 0.5230116179180577,
+ "learning_rate": 2e-05,
+ "loss": 0.6872,
+ "step": 171
+ },
+ {
+ "epoch": 5.34375,
+ "eval_loss": 0.7058187127113342,
+ "eval_runtime": 47.5711,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 171
+ },
+ {
+ "epoch": 5.375,
+ "grad_norm": 0.48081340678536255,
+ "learning_rate": 2e-05,
+ "loss": 0.7694,
+ "step": 172
+ },
+ {
+ "epoch": 5.375,
+ "eval_loss": 0.7044984698295593,
+ "eval_runtime": 47.4233,
+ "eval_samples_per_second": 4.217,
+ "eval_steps_per_second": 0.274,
+ "step": 172
+ },
+ {
+ "epoch": 5.40625,
+ "grad_norm": 0.4787525602476421,
+ "learning_rate": 2e-05,
+ "loss": 0.7342,
+ "step": 173
+ },
+ {
+ "epoch": 5.40625,
+ "eval_loss": 0.7032212018966675,
+ "eval_runtime": 47.3534,
+ "eval_samples_per_second": 4.224,
+ "eval_steps_per_second": 0.275,
+ "step": 173
+ },
+ {
+ "epoch": 5.4375,
+ "grad_norm": 0.4871847582306217,
+ "learning_rate": 2e-05,
+ "loss": 0.7562,
+ "step": 174
+ },
+ {
+ "epoch": 5.4375,
+ "eval_loss": 0.7019696235656738,
+ "eval_runtime": 47.382,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 174
+ },
+ {
+ "epoch": 5.46875,
+ "grad_norm": 0.47999745025553603,
+ "learning_rate": 2e-05,
+ "loss": 0.7534,
+ "step": 175
+ },
+ {
+ "epoch": 5.46875,
+ "eval_loss": 0.7014529705047607,
+ "eval_runtime": 47.4435,
+ "eval_samples_per_second": 4.216,
+ "eval_steps_per_second": 0.274,
+ "step": 175
+ },
+ {
+ "epoch": 5.5,
+ "grad_norm": 0.5168030891996357,
+ "learning_rate": 2e-05,
+ "loss": 0.707,
+ "step": 176
+ },
+ {
+ "epoch": 5.5,
+ "eval_loss": 0.6993884444236755,
+ "eval_runtime": 47.4943,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 176
+ },
+ {
+ "epoch": 5.53125,
+ "grad_norm": 0.536450206978984,
+ "learning_rate": 2e-05,
+ "loss": 0.7318,
+ "step": 177
+ },
+ {
+ "epoch": 5.53125,
+ "eval_loss": 0.6971662640571594,
+ "eval_runtime": 47.4193,
+ "eval_samples_per_second": 4.218,
+ "eval_steps_per_second": 0.274,
+ "step": 177
+ },
+ {
+ "epoch": 5.5625,
+ "grad_norm": 0.45352543205020696,
+ "learning_rate": 2e-05,
+ "loss": 0.7421,
+ "step": 178
+ },
+ {
+ "epoch": 5.5625,
+ "eval_loss": 0.6962605118751526,
+ "eval_runtime": 47.3798,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 178
+ },
+ {
+ "epoch": 5.59375,
+ "grad_norm": 0.5054883443109318,
+ "learning_rate": 2e-05,
+ "loss": 0.6668,
+ "step": 179
+ },
+ {
+ "epoch": 5.59375,
+ "eval_loss": 0.6970357298851013,
+ "eval_runtime": 47.3311,
+ "eval_samples_per_second": 4.226,
+ "eval_steps_per_second": 0.275,
+ "step": 179
+ },
+ {
+ "epoch": 5.625,
+ "grad_norm": 0.49584660418833293,
+ "learning_rate": 2e-05,
+ "loss": 0.6548,
+ "step": 180
+ },
+ {
+ "epoch": 5.625,
+ "eval_loss": 0.6980059146881104,
+ "eval_runtime": 47.299,
+ "eval_samples_per_second": 4.228,
+ "eval_steps_per_second": 0.275,
+ "step": 180
+ },
+ {
+ "epoch": 5.65625,
+ "grad_norm": 0.5114381326491793,
+ "learning_rate": 2e-05,
+ "loss": 0.6691,
+ "step": 181
+ },
+ {
+ "epoch": 5.65625,
+ "eval_loss": 0.6995040774345398,
+ "eval_runtime": 47.3887,
+ "eval_samples_per_second": 4.22,
+ "eval_steps_per_second": 0.274,
+ "step": 181
+ },
+ {
+ "epoch": 5.6875,
+ "grad_norm": 0.48550125668870825,
+ "learning_rate": 2e-05,
+ "loss": 0.6525,
+ "step": 182
+ },
+ {
+ "epoch": 5.6875,
+ "eval_loss": 0.7020326256752014,
+ "eval_runtime": 47.3838,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 182
+ },
+ {
+ "epoch": 5.71875,
+ "grad_norm": 0.5860847796671736,
+ "learning_rate": 2e-05,
+ "loss": 0.674,
+ "step": 183
+ },
+ {
+ "epoch": 5.71875,
+ "eval_loss": 0.7027825713157654,
+ "eval_runtime": 47.3875,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 183
+ },
+ {
+ "epoch": 5.75,
+ "grad_norm": 0.5535582209035479,
+ "learning_rate": 2e-05,
+ "loss": 0.6643,
+ "step": 184
+ },
+ {
+ "epoch": 5.75,
+ "eval_loss": 0.7025408148765564,
+ "eval_runtime": 47.5534,
+ "eval_samples_per_second": 4.206,
+ "eval_steps_per_second": 0.273,
+ "step": 184
+ },
+ {
+ "epoch": 5.78125,
+ "grad_norm": 0.5443574176405931,
+ "learning_rate": 2e-05,
+ "loss": 0.709,
+ "step": 185
+ },
+ {
+ "epoch": 5.78125,
+ "eval_loss": 0.7007840871810913,
+ "eval_runtime": 47.4469,
+ "eval_samples_per_second": 4.215,
+ "eval_steps_per_second": 0.274,
+ "step": 185
+ },
+ {
+ "epoch": 5.8125,
+ "grad_norm": 0.563830259704143,
+ "learning_rate": 2e-05,
+ "loss": 0.6884,
+ "step": 186
+ },
+ {
+ "epoch": 5.8125,
+ "eval_loss": 0.6979361176490784,
+ "eval_runtime": 49.1203,
+ "eval_samples_per_second": 4.072,
+ "eval_steps_per_second": 0.265,
+ "step": 186
+ },
+ {
+ "epoch": 5.84375,
+ "grad_norm": 0.5094956892765212,
+ "learning_rate": 2e-05,
+ "loss": 0.7318,
+ "step": 187
+ },
+ {
+ "epoch": 5.84375,
+ "eval_loss": 0.6962587237358093,
+ "eval_runtime": 49.1831,
+ "eval_samples_per_second": 4.066,
+ "eval_steps_per_second": 0.264,
+ "step": 187
+ },
+ {
+ "epoch": 5.875,
+ "grad_norm": 0.5264819980742595,
+ "learning_rate": 2e-05,
+ "loss": 0.6746,
+ "step": 188
+ },
+ {
+ "epoch": 5.875,
+ "eval_loss": 0.694776713848114,
+ "eval_runtime": 49.1994,
+ "eval_samples_per_second": 4.065,
+ "eval_steps_per_second": 0.264,
+ "step": 188
+ },
+ {
+ "epoch": 5.90625,
+ "grad_norm": 0.4737429304023209,
+ "learning_rate": 2e-05,
+ "loss": 0.664,
+ "step": 189
+ },
+ {
+ "epoch": 5.90625,
+ "eval_loss": 0.6939517855644226,
+ "eval_runtime": 49.2438,
+ "eval_samples_per_second": 4.061,
+ "eval_steps_per_second": 0.264,
+ "step": 189
+ },
+ {
+ "epoch": 5.9375,
+ "grad_norm": 0.494163934813738,
+ "learning_rate": 2e-05,
+ "loss": 0.6978,
+ "step": 190
+ },
+ {
+ "epoch": 5.9375,
+ "eval_loss": 0.6933834552764893,
+ "eval_runtime": 49.3494,
+ "eval_samples_per_second": 4.053,
+ "eval_steps_per_second": 0.263,
+ "step": 190
+ },
+ {
+ "epoch": 5.96875,
+ "grad_norm": 0.4945972278087299,
+ "learning_rate": 2e-05,
+ "loss": 0.6909,
+ "step": 191
+ },
+ {
+ "epoch": 5.96875,
+ "eval_loss": 0.6924250721931458,
+ "eval_runtime": 50.3255,
+ "eval_samples_per_second": 3.974,
+ "eval_steps_per_second": 0.258,
+ "step": 191
+ },
+ {
+ "epoch": 6.0,
+ "grad_norm": 0.48872556688745233,
+ "learning_rate": 2e-05,
+ "loss": 0.6622,
+ "step": 192
+ },
+ {
+ "epoch": 6.0,
+ "eval_loss": 0.6922193765640259,
+ "eval_runtime": 50.4561,
+ "eval_samples_per_second": 3.964,
+ "eval_steps_per_second": 0.258,
+ "step": 192
+ },
+ {
+ "epoch": 6.03125,
+ "grad_norm": 0.5013452255378538,
+ "learning_rate": 2e-05,
+ "loss": 0.7458,
+ "step": 193
+ },
+ {
+ "epoch": 6.03125,
+ "eval_loss": 0.6931161284446716,
+ "eval_runtime": 50.5049,
+ "eval_samples_per_second": 3.96,
+ "eval_steps_per_second": 0.257,
+ "step": 193
+ },
+ {
+ "epoch": 6.0625,
+ "grad_norm": 0.48271161232093784,
+ "learning_rate": 2e-05,
+ "loss": 0.7171,
+ "step": 194
+ },
+ {
+ "epoch": 6.0625,
+ "eval_loss": 0.6959040760993958,
+ "eval_runtime": 50.2441,
+ "eval_samples_per_second": 3.981,
+ "eval_steps_per_second": 0.259,
+ "step": 194
+ },
+ {
+ "epoch": 6.09375,
+ "grad_norm": 0.5414562703154852,
+ "learning_rate": 2e-05,
+ "loss": 0.6419,
+ "step": 195
+ },
+ {
+ "epoch": 6.09375,
+ "eval_loss": 0.7000604271888733,
+ "eval_runtime": 50.4261,
+ "eval_samples_per_second": 3.966,
+ "eval_steps_per_second": 0.258,
+ "step": 195
+ },
+ {
+ "epoch": 6.125,
+ "grad_norm": 0.5074661247335385,
+ "learning_rate": 2e-05,
+ "loss": 0.6881,
+ "step": 196
+ },
+ {
+ "epoch": 6.125,
+ "eval_loss": 0.7039622664451599,
+ "eval_runtime": 51.5214,
+ "eval_samples_per_second": 3.882,
+ "eval_steps_per_second": 0.252,
+ "step": 196
+ },
+ {
+ "epoch": 6.15625,
+ "grad_norm": 0.5603468534764365,
+ "learning_rate": 2e-05,
+ "loss": 0.7085,
+ "step": 197
+ },
+ {
+ "epoch": 6.15625,
+ "eval_loss": 0.7055023312568665,
+ "eval_runtime": 51.7102,
+ "eval_samples_per_second": 3.868,
+ "eval_steps_per_second": 0.251,
+ "step": 197
+ },
+ {
+ "epoch": 6.1875,
+ "grad_norm": 0.5992190802422799,
+ "learning_rate": 2e-05,
+ "loss": 0.7614,
+ "step": 198
+ },
+ {
+ "epoch": 6.1875,
+ "eval_loss": 0.7046856880187988,
+ "eval_runtime": 51.5464,
+ "eval_samples_per_second": 3.88,
+ "eval_steps_per_second": 0.252,
+ "step": 198
+ },
+ {
+ "epoch": 6.21875,
+ "grad_norm": 0.6293684167527106,
+ "learning_rate": 2e-05,
+ "loss": 0.6435,
+ "step": 199
+ },
+ {
+ "epoch": 6.21875,
+ "eval_loss": 0.7021151781082153,
+ "eval_runtime": 51.5328,
+ "eval_samples_per_second": 3.881,
+ "eval_steps_per_second": 0.252,
+ "step": 199
+ },
+ {
+ "epoch": 6.25,
+ "grad_norm": 0.591265449241434,
+ "learning_rate": 2e-05,
+ "loss": 0.688,
+ "step": 200
+ },
+ {
+ "epoch": 6.25,
+ "eval_loss": 0.7002359628677368,
+ "eval_runtime": 51.5812,
+ "eval_samples_per_second": 3.877,
+ "eval_steps_per_second": 0.252,
+ "step": 200
+ },
+ {
+ "epoch": 6.28125,
+ "grad_norm": 0.543141536526749,
+ "learning_rate": 2e-05,
+ "loss": 0.7027,
+ "step": 201
+ },
+ {
+ "epoch": 6.28125,
+ "eval_loss": 0.6986366510391235,
+ "eval_runtime": 52.6956,
+ "eval_samples_per_second": 3.795,
+ "eval_steps_per_second": 0.247,
+ "step": 201
+ },
+ {
+ "epoch": 6.3125,
+ "grad_norm": 0.5679656300203245,
+ "learning_rate": 2e-05,
+ "loss": 0.625,
+ "step": 202
+ },
+ {
+ "epoch": 6.3125,
+ "eval_loss": 0.698679506778717,
+ "eval_runtime": 52.5102,
+ "eval_samples_per_second": 3.809,
+ "eval_steps_per_second": 0.248,
+ "step": 202
+ },
+ {
+ "epoch": 6.34375,
+ "grad_norm": 0.5285839896523021,
+ "learning_rate": 2e-05,
+ "loss": 0.7687,
+ "step": 203
+ },
+ {
+ "epoch": 6.34375,
+ "eval_loss": 0.7005956768989563,
+ "eval_runtime": 52.6067,
+ "eval_samples_per_second": 3.802,
+ "eval_steps_per_second": 0.247,
+ "step": 203
+ },
+ {
+ "epoch": 6.375,
+ "grad_norm": 0.6512964945211068,
+ "learning_rate": 2e-05,
+ "loss": 0.623,
+ "step": 204
+ },
+ {
+ "epoch": 6.375,
+ "eval_loss": 0.7013595104217529,
+ "eval_runtime": 52.5428,
+ "eval_samples_per_second": 3.806,
+ "eval_steps_per_second": 0.247,
+ "step": 204
+ },
+ {
+ "epoch": 6.40625,
+ "grad_norm": 0.5295248631519638,
+ "learning_rate": 2e-05,
+ "loss": 0.5941,
+ "step": 205
+ },
+ {
+ "epoch": 6.40625,
+ "eval_loss": 0.7016547322273254,
+ "eval_runtime": 52.6142,
+ "eval_samples_per_second": 3.801,
+ "eval_steps_per_second": 0.247,
+ "step": 205
+ },
+ {
+ "epoch": 6.4375,
+ "grad_norm": 0.6134157701434021,
+ "learning_rate": 2e-05,
+ "loss": 0.6506,
+ "step": 206
+ },
+ {
+ "epoch": 6.4375,
+ "eval_loss": 0.7009623646736145,
+ "eval_runtime": 52.1942,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 206
+ },
+ {
+ "epoch": 6.46875,
+ "grad_norm": 0.57886797614996,
+ "learning_rate": 2e-05,
+ "loss": 0.6983,
+ "step": 207
+ },
+ {
+ "epoch": 6.46875,
+ "eval_loss": 0.6988092064857483,
+ "eval_runtime": 52.2577,
+ "eval_samples_per_second": 3.827,
+ "eval_steps_per_second": 0.249,
+ "step": 207
+ },
+ {
+ "epoch": 6.5,
+ "grad_norm": 0.5593482836944472,
+ "learning_rate": 2e-05,
+ "loss": 0.6348,
+ "step": 208
+ },
+ {
+ "epoch": 6.5,
+ "eval_loss": 0.698823094367981,
+ "eval_runtime": 52.2296,
+ "eval_samples_per_second": 3.829,
+ "eval_steps_per_second": 0.249,
+ "step": 208
+ },
+ {
+ "epoch": 6.53125,
+ "grad_norm": 0.662802162179718,
+ "learning_rate": 2e-05,
+ "loss": 0.6206,
+ "step": 209
+ },
+ {
+ "epoch": 6.53125,
+ "eval_loss": 0.6990167498588562,
+ "eval_runtime": 52.4316,
+ "eval_samples_per_second": 3.814,
+ "eval_steps_per_second": 0.248,
+ "step": 209
+ },
+ {
+ "epoch": 6.5625,
+ "grad_norm": 0.6874374231122908,
+ "learning_rate": 2e-05,
+ "loss": 0.6033,
+ "step": 210
+ },
+ {
+ "epoch": 6.5625,
+ "eval_loss": 0.699796736240387,
+ "eval_runtime": 52.3193,
+ "eval_samples_per_second": 3.823,
+ "eval_steps_per_second": 0.248,
+ "step": 210
+ },
+ {
+ "epoch": 6.59375,
+ "grad_norm": 0.6625766736772473,
+ "learning_rate": 2e-05,
+ "loss": 0.6398,
+ "step": 211
+ },
+ {
+ "epoch": 6.59375,
+ "eval_loss": 0.6989737153053284,
+ "eval_runtime": 52.1885,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 211
+ },
+ {
+ "epoch": 6.625,
+ "grad_norm": 0.6563419096027812,
+ "learning_rate": 2e-05,
+ "loss": 0.6119,
+ "step": 212
+ },
+ {
+ "epoch": 6.625,
+ "eval_loss": 0.6973609924316406,
+ "eval_runtime": 52.1628,
+ "eval_samples_per_second": 3.834,
+ "eval_steps_per_second": 0.249,
+ "step": 212
+ },
+ {
+ "epoch": 6.65625,
+ "grad_norm": 0.5796353226697397,
+ "learning_rate": 2e-05,
+ "loss": 0.7041,
+ "step": 213
+ },
+ {
+ "epoch": 6.65625,
+ "eval_loss": 0.6957942247390747,
+ "eval_runtime": 52.2028,
+ "eval_samples_per_second": 3.831,
+ "eval_steps_per_second": 0.249,
+ "step": 213
+ },
+ {
+ "epoch": 6.6875,
+ "grad_norm": 0.5711947110504899,
+ "learning_rate": 2e-05,
+ "loss": 0.6465,
+ "step": 214
+ },
+ {
+ "epoch": 6.6875,
+ "eval_loss": 0.696739673614502,
+ "eval_runtime": 52.1849,
+ "eval_samples_per_second": 3.833,
+ "eval_steps_per_second": 0.249,
+ "step": 214
+ },
+ {
+ "epoch": 6.71875,
+ "grad_norm": 0.6619502413653232,
+ "learning_rate": 2e-05,
+ "loss": 0.6563,
+ "step": 215
+ },
+ {
+ "epoch": 6.71875,
+ "eval_loss": 0.6960940361022949,
+ "eval_runtime": 52.0996,
+ "eval_samples_per_second": 3.839,
+ "eval_steps_per_second": 0.25,
+ "step": 215
+ },
+ {
+ "epoch": 6.75,
+ "grad_norm": 0.6587126256919645,
+ "learning_rate": 2e-05,
+ "loss": 0.6505,
+ "step": 216
+ },
+ {
+ "epoch": 6.75,
+ "eval_loss": 0.6959022283554077,
+ "eval_runtime": 52.1062,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 216
+ },
+ {
+ "epoch": 6.78125,
+ "grad_norm": 0.648164277941964,
+ "learning_rate": 2e-05,
+ "loss": 0.5969,
+ "step": 217
+ },
+ {
+ "epoch": 6.78125,
+ "eval_loss": 0.6999121308326721,
+ "eval_runtime": 51.9356,
+ "eval_samples_per_second": 3.851,
+ "eval_steps_per_second": 0.25,
+ "step": 217
+ },
+ {
+ "epoch": 6.8125,
+ "grad_norm": 0.6595860789738482,
+ "learning_rate": 2e-05,
+ "loss": 0.5945,
+ "step": 218
+ },
+ {
+ "epoch": 6.8125,
+ "eval_loss": 0.7028067111968994,
+ "eval_runtime": 52.2232,
+ "eval_samples_per_second": 3.83,
+ "eval_steps_per_second": 0.249,
+ "step": 218
+ },
+ {
+ "epoch": 6.84375,
+ "grad_norm": 0.7116894779822719,
+ "learning_rate": 2e-05,
+ "loss": 0.7027,
+ "step": 219
+ },
+ {
+ "epoch": 6.84375,
+ "eval_loss": 0.7035638689994812,
+ "eval_runtime": 52.1471,
+ "eval_samples_per_second": 3.835,
+ "eval_steps_per_second": 0.249,
+ "step": 219
+ },
+ {
+ "epoch": 6.875,
+ "grad_norm": 0.7581142336087988,
+ "learning_rate": 2e-05,
+ "loss": 0.7171,
+ "step": 220
+ },
+ {
+ "epoch": 6.875,
+ "eval_loss": 0.6981176733970642,
+ "eval_runtime": 52.1366,
+ "eval_samples_per_second": 3.836,
+ "eval_steps_per_second": 0.249,
+ "step": 220
+ },
+ {
+ "epoch": 6.90625,
+ "grad_norm": 0.6261292745909233,
+ "learning_rate": 2e-05,
+ "loss": 0.658,
+ "step": 221
+ },
+ {
+ "epoch": 6.90625,
+ "eval_loss": 0.6939045786857605,
+ "eval_runtime": 52.2211,
+ "eval_samples_per_second": 3.83,
+ "eval_steps_per_second": 0.249,
+ "step": 221
+ },
+ {
+ "epoch": 6.9375,
+ "grad_norm": 0.7256427809370966,
+ "learning_rate": 2e-05,
+ "loss": 0.6576,
+ "step": 222
+ },
+ {
+ "epoch": 6.9375,
+ "eval_loss": 0.6904327273368835,
+ "eval_runtime": 52.1829,
+ "eval_samples_per_second": 3.833,
+ "eval_steps_per_second": 0.249,
+ "step": 222
+ },
+ {
+ "epoch": 6.96875,
+ "grad_norm": 0.6653711103404113,
+ "learning_rate": 2e-05,
+ "loss": 0.6938,
+ "step": 223
+ },
+ {
+ "epoch": 6.96875,
+ "eval_loss": 0.6893274188041687,
+ "eval_runtime": 51.899,
+ "eval_samples_per_second": 3.854,
+ "eval_steps_per_second": 0.25,
+ "step": 223
+ },
+ {
+ "epoch": 7.0,
+ "grad_norm": 0.6730688267524797,
+ "learning_rate": 2e-05,
+ "loss": 0.7397,
+ "step": 224
+ },
+ {
+ "epoch": 7.0,
+ "eval_loss": 0.6895740032196045,
+ "eval_runtime": 52.1977,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 224
+ },
+ {
+ "epoch": 7.03125,
+ "grad_norm": 0.5832904533111831,
+ "learning_rate": 2e-05,
+ "loss": 0.6366,
+ "step": 225
+ },
+ {
+ "epoch": 7.03125,
+ "eval_loss": 0.690305769443512,
+ "eval_runtime": 51.0898,
+ "eval_samples_per_second": 3.915,
+ "eval_steps_per_second": 0.254,
+ "step": 225
+ },
+ {
+ "epoch": 7.0625,
+ "grad_norm": 0.7244416322910332,
+ "learning_rate": 2e-05,
+ "loss": 0.5756,
+ "step": 226
+ },
+ {
+ "epoch": 7.0625,
+ "eval_loss": 0.6943302154541016,
+ "eval_runtime": 47.5876,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 226
+ },
+ {
+ "epoch": 7.09375,
+ "grad_norm": 0.6507055762944723,
+ "learning_rate": 2e-05,
+ "loss": 0.622,
+ "step": 227
+ },
+ {
+ "epoch": 7.09375,
+ "eval_loss": 0.7073258757591248,
+ "eval_runtime": 47.5809,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 227
+ },
+ {
+ "epoch": 7.125,
+ "grad_norm": 0.7122561204700196,
+ "learning_rate": 2e-05,
+ "loss": 0.5908,
+ "step": 228
+ },
+ {
+ "epoch": 7.125,
+ "eval_loss": 0.7263233065605164,
+ "eval_runtime": 47.544,
+ "eval_samples_per_second": 4.207,
+ "eval_steps_per_second": 0.273,
+ "step": 228
+ },
+ {
+ "epoch": 7.15625,
+ "grad_norm": 1.053512823308346,
+ "learning_rate": 2e-05,
+ "loss": 0.6193,
+ "step": 229
+ },
+ {
+ "epoch": 7.15625,
+ "eval_loss": 0.7283624410629272,
+ "eval_runtime": 47.5998,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 229
+ },
+ {
+ "epoch": 7.1875,
+ "grad_norm": 1.0167138351900848,
+ "learning_rate": 2e-05,
+ "loss": 0.5942,
+ "step": 230
+ },
+ {
+ "epoch": 7.1875,
+ "eval_loss": 0.7136476039886475,
+ "eval_runtime": 47.5738,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 230
+ },
+ {
+ "epoch": 7.21875,
+ "grad_norm": 0.7388726343392281,
+ "learning_rate": 2e-05,
+ "loss": 0.6898,
+ "step": 231
+ },
+ {
+ "epoch": 7.21875,
+ "eval_loss": 0.7017656564712524,
+ "eval_runtime": 47.5857,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 231
+ },
+ {
+ "epoch": 7.25,
+ "grad_norm": 0.6255681554939039,
+ "learning_rate": 2e-05,
+ "loss": 0.669,
+ "step": 232
+ },
+ {
+ "epoch": 7.25,
+ "eval_loss": 0.6967242956161499,
+ "eval_runtime": 47.7483,
+ "eval_samples_per_second": 4.189,
+ "eval_steps_per_second": 0.272,
+ "step": 232
+ },
+ {
+ "epoch": 7.28125,
+ "grad_norm": 0.7000438574267057,
+ "learning_rate": 2e-05,
+ "loss": 0.6143,
+ "step": 233
+ },
+ {
+ "epoch": 7.28125,
+ "eval_loss": 0.694460391998291,
+ "eval_runtime": 47.7828,
+ "eval_samples_per_second": 4.186,
+ "eval_steps_per_second": 0.272,
+ "step": 233
+ },
+ {
+ "epoch": 7.3125,
+ "grad_norm": 0.6658391411050186,
+ "learning_rate": 2e-05,
+ "loss": 0.6737,
+ "step": 234
+ },
+ {
+ "epoch": 7.3125,
+ "eval_loss": 0.6925583481788635,
+ "eval_runtime": 47.7913,
+ "eval_samples_per_second": 4.185,
+ "eval_steps_per_second": 0.272,
+ "step": 234
+ },
+ {
+ "epoch": 7.34375,
+ "grad_norm": 0.6473191970636399,
+ "learning_rate": 2e-05,
+ "loss": 0.6347,
+ "step": 235
+ },
+ {
+ "epoch": 7.34375,
+ "eval_loss": 0.6907203793525696,
+ "eval_runtime": 47.6866,
+ "eval_samples_per_second": 4.194,
+ "eval_steps_per_second": 0.273,
+ "step": 235
+ },
+ {
+ "epoch": 7.375,
+ "grad_norm": 0.703409963718735,
+ "learning_rate": 2e-05,
+ "loss": 0.5991,
+ "step": 236
+ },
+ {
+ "epoch": 7.375,
+ "eval_loss": 0.6898574829101562,
+ "eval_runtime": 47.6481,
+ "eval_samples_per_second": 4.197,
+ "eval_steps_per_second": 0.273,
+ "step": 236
+ },
+ {
+ "epoch": 7.40625,
+ "grad_norm": 0.6957469611517898,
+ "learning_rate": 2e-05,
+ "loss": 0.6428,
+ "step": 237
+ },
+ {
+ "epoch": 7.40625,
+ "eval_loss": 0.6922276020050049,
+ "eval_runtime": 47.7072,
+ "eval_samples_per_second": 4.192,
+ "eval_steps_per_second": 0.272,
+ "step": 237
+ },
+ {
+ "epoch": 7.4375,
+ "grad_norm": 0.7383281551578481,
+ "learning_rate": 2e-05,
+ "loss": 0.6272,
+ "step": 238
+ },
+ {
+ "epoch": 7.4375,
+ "eval_loss": 0.6988270282745361,
+ "eval_runtime": 47.5925,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 238
+ },
+ {
+ "epoch": 7.46875,
+ "grad_norm": 0.7113722006702997,
+ "learning_rate": 2e-05,
+ "loss": 0.6594,
+ "step": 239
+ },
+ {
+ "epoch": 7.46875,
+ "eval_loss": 0.7074680328369141,
+ "eval_runtime": 47.7257,
+ "eval_samples_per_second": 4.191,
+ "eval_steps_per_second": 0.272,
+ "step": 239
+ },
+ {
+ "epoch": 7.5,
+ "grad_norm": 0.7233836456752487,
+ "learning_rate": 2e-05,
+ "loss": 0.6003,
+ "step": 240
+ },
+ {
+ "epoch": 7.5,
+ "eval_loss": 0.7172031402587891,
+ "eval_runtime": 47.7463,
+ "eval_samples_per_second": 4.189,
+ "eval_steps_per_second": 0.272,
+ "step": 240
+ },
+ {
+ "epoch": 7.53125,
+ "grad_norm": 0.7452166529670862,
+ "learning_rate": 2e-05,
+ "loss": 0.6463,
+ "step": 241
+ },
+ {
+ "epoch": 7.53125,
+ "eval_loss": 0.7228195071220398,
+ "eval_runtime": 47.6283,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 241
+ },
+ {
+ "epoch": 7.5625,
+ "grad_norm": 0.88949489838851,
+ "learning_rate": 2e-05,
+ "loss": 0.6463,
+ "step": 242
+ },
+ {
+ "epoch": 7.5625,
+ "eval_loss": 0.7194420099258423,
+ "eval_runtime": 47.6221,
+ "eval_samples_per_second": 4.2,
+ "eval_steps_per_second": 0.273,
+ "step": 242
+ },
+ {
+ "epoch": 7.59375,
+ "grad_norm": 0.7592408002786533,
+ "learning_rate": 2e-05,
+ "loss": 0.6301,
+ "step": 243
+ },
+ {
+ "epoch": 7.59375,
+ "eval_loss": 0.7122278809547424,
+ "eval_runtime": 47.7549,
+ "eval_samples_per_second": 4.188,
+ "eval_steps_per_second": 0.272,
+ "step": 243
+ },
+ {
+ "epoch": 7.625,
+ "grad_norm": 0.910753798896517,
+ "learning_rate": 2e-05,
+ "loss": 0.7016,
+ "step": 244
+ },
+ {
+ "epoch": 7.625,
+ "eval_loss": 0.7019688487052917,
+ "eval_runtime": 47.5592,
+ "eval_samples_per_second": 4.205,
+ "eval_steps_per_second": 0.273,
+ "step": 244
+ },
+ {
+ "epoch": 7.65625,
+ "grad_norm": 0.7861795541835009,
+ "learning_rate": 2e-05,
+ "loss": 0.6107,
+ "step": 245
+ },
+ {
+ "epoch": 7.65625,
+ "eval_loss": 0.6964650750160217,
+ "eval_runtime": 47.5842,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 245
+ },
+ {
+ "epoch": 7.6875,
+ "grad_norm": 0.7162378610377871,
+ "learning_rate": 2e-05,
+ "loss": 0.6474,
+ "step": 246
+ },
+ {
+ "epoch": 7.6875,
+ "eval_loss": 0.6934291124343872,
+ "eval_runtime": 47.4792,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 246
+ },
+ {
+ "epoch": 7.71875,
+ "grad_norm": 0.7261823254305776,
+ "learning_rate": 2e-05,
+ "loss": 0.636,
+ "step": 247
+ },
+ {
+ "epoch": 7.71875,
+ "eval_loss": 0.6925876140594482,
+ "eval_runtime": 47.6623,
+ "eval_samples_per_second": 4.196,
+ "eval_steps_per_second": 0.273,
+ "step": 247
+ },
+ {
+ "epoch": 7.75,
+ "grad_norm": 0.6757318335309442,
+ "learning_rate": 2e-05,
+ "loss": 0.6249,
+ "step": 248
+ },
+ {
+ "epoch": 7.75,
+ "eval_loss": 0.6934402585029602,
+ "eval_runtime": 47.5464,
+ "eval_samples_per_second": 4.206,
+ "eval_steps_per_second": 0.273,
+ "step": 248
+ },
+ {
+ "epoch": 7.78125,
+ "grad_norm": 0.7182105984315053,
+ "learning_rate": 2e-05,
+ "loss": 0.6676,
+ "step": 249
+ },
+ {
+ "epoch": 7.78125,
+ "eval_loss": 0.6956924200057983,
+ "eval_runtime": 47.6014,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 249
+ },
+ {
+ "epoch": 7.8125,
+ "grad_norm": 0.7231439954921842,
+ "learning_rate": 2e-05,
+ "loss": 0.6719,
+ "step": 250
+ },
+ {
+ "epoch": 7.8125,
+ "eval_loss": 0.6987011432647705,
+ "eval_runtime": 47.64,
+ "eval_samples_per_second": 4.198,
+ "eval_steps_per_second": 0.273,
+ "step": 250
+ },
+ {
+ "epoch": 7.84375,
+ "grad_norm": 0.7938681326839265,
+ "learning_rate": 2e-05,
+ "loss": 0.584,
+ "step": 251
+ },
+ {
+ "epoch": 7.84375,
+ "eval_loss": 0.7026040554046631,
+ "eval_runtime": 47.6391,
+ "eval_samples_per_second": 4.198,
+ "eval_steps_per_second": 0.273,
+ "step": 251
+ },
+ {
+ "epoch": 7.875,
+ "grad_norm": 0.8011657536057513,
+ "learning_rate": 2e-05,
+ "loss": 0.594,
+ "step": 252
+ },
+ {
+ "epoch": 7.875,
+ "eval_loss": 0.7068576216697693,
+ "eval_runtime": 47.635,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 252
+ },
+ {
+ "epoch": 7.90625,
+ "grad_norm": 0.819763617578999,
+ "learning_rate": 2e-05,
+ "loss": 0.6758,
+ "step": 253
+ },
+ {
+ "epoch": 7.90625,
+ "eval_loss": 0.7079121470451355,
+ "eval_runtime": 47.6352,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 253
+ },
+ {
+ "epoch": 7.9375,
+ "grad_norm": 0.7697343122686975,
+ "learning_rate": 2e-05,
+ "loss": 0.6224,
+ "step": 254
+ },
+ {
+ "epoch": 7.9375,
+ "eval_loss": 0.7092974781990051,
+ "eval_runtime": 47.5993,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 254
+ },
+ {
+ "epoch": 7.96875,
+ "grad_norm": 0.8148531217392738,
+ "learning_rate": 2e-05,
+ "loss": 0.5579,
+ "step": 255
+ },
+ {
+ "epoch": 7.96875,
+ "eval_loss": 0.7090660333633423,
+ "eval_runtime": 47.5602,
+ "eval_samples_per_second": 4.205,
+ "eval_steps_per_second": 0.273,
+ "step": 255
+ },
+ {
+ "epoch": 8.0,
+ "grad_norm": 0.7576748044477204,
+ "learning_rate": 2e-05,
+ "loss": 0.609,
+ "step": 256
+ },
+ {
+ "epoch": 8.0,
+ "eval_loss": 0.7068901062011719,
+ "eval_runtime": 47.5944,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 256
+ },
+ {
+ "epoch": 8.03125,
+ "grad_norm": 0.814119412415159,
+ "learning_rate": 2e-05,
+ "loss": 0.5816,
+ "step": 257
+ },
+ {
+ "epoch": 8.03125,
+ "eval_loss": 0.7052778005599976,
+ "eval_runtime": 50.9012,
+ "eval_samples_per_second": 3.929,
+ "eval_steps_per_second": 0.255,
+ "step": 257
+ },
+ {
+ "epoch": 8.0625,
+ "grad_norm": 0.7940502590060119,
+ "learning_rate": 2e-05,
+ "loss": 0.5974,
+ "step": 258
+ },
+ {
+ "epoch": 8.0625,
+ "eval_loss": 0.7055818438529968,
+ "eval_runtime": 47.5726,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 258
+ },
+ {
+ "epoch": 8.09375,
+ "grad_norm": 0.7373690747574106,
+ "learning_rate": 2e-05,
+ "loss": 0.6267,
+ "step": 259
+ },
+ {
+ "epoch": 8.09375,
+ "eval_loss": 0.7084596753120422,
+ "eval_runtime": 47.5924,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 259
+ },
+ {
+ "epoch": 8.125,
+ "grad_norm": 0.8486372724795598,
+ "learning_rate": 2e-05,
+ "loss": 0.6349,
+ "step": 260
+ },
+ {
+ "epoch": 8.125,
+ "eval_loss": 0.7118301391601562,
+ "eval_runtime": 47.9994,
+ "eval_samples_per_second": 4.167,
+ "eval_steps_per_second": 0.271,
+ "step": 260
+ },
+ {
+ "epoch": 8.15625,
+ "grad_norm": 0.8391397763830329,
+ "learning_rate": 2e-05,
+ "loss": 0.5575,
+ "step": 261
+ },
+ {
+ "epoch": 8.15625,
+ "eval_loss": 0.7155640125274658,
+ "eval_runtime": 47.6071,
+ "eval_samples_per_second": 4.201,
+ "eval_steps_per_second": 0.273,
+ "step": 261
+ },
+ {
+ "epoch": 8.1875,
+ "grad_norm": 0.7928693737279656,
+ "learning_rate": 2e-05,
+ "loss": 0.6777,
+ "step": 262
+ },
+ {
+ "epoch": 8.1875,
+ "eval_loss": 0.7209051251411438,
+ "eval_runtime": 47.6324,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 262
+ },
+ {
+ "epoch": 8.21875,
+ "grad_norm": 0.9171124624201488,
+ "learning_rate": 2e-05,
+ "loss": 0.5582,
+ "step": 263
+ },
+ {
+ "epoch": 8.21875,
+ "eval_loss": 0.7233929634094238,
+ "eval_runtime": 47.7509,
+ "eval_samples_per_second": 4.188,
+ "eval_steps_per_second": 0.272,
+ "step": 263
+ },
+ {
+ "epoch": 8.25,
+ "grad_norm": 0.9128766641132847,
+ "learning_rate": 2e-05,
+ "loss": 0.597,
+ "step": 264
+ },
+ {
+ "epoch": 8.25,
+ "eval_loss": 0.7227862477302551,
+ "eval_runtime": 47.5667,
+ "eval_samples_per_second": 4.205,
+ "eval_steps_per_second": 0.273,
+ "step": 264
+ },
+ {
+ "epoch": 8.28125,
+ "grad_norm": 1.0298171058788395,
+ "learning_rate": 2e-05,
+ "loss": 0.6262,
+ "step": 265
+ },
+ {
+ "epoch": 8.28125,
+ "eval_loss": 0.7159123420715332,
+ "eval_runtime": 47.6441,
+ "eval_samples_per_second": 4.198,
+ "eval_steps_per_second": 0.273,
+ "step": 265
+ },
+ {
+ "epoch": 8.3125,
+ "grad_norm": 0.8345277253579861,
+ "learning_rate": 2e-05,
+ "loss": 0.5973,
+ "step": 266
+ },
+ {
+ "epoch": 8.3125,
+ "eval_loss": 0.7099489569664001,
+ "eval_runtime": 49.5358,
+ "eval_samples_per_second": 4.037,
+ "eval_steps_per_second": 0.262,
+ "step": 266
+ },
+ {
+ "epoch": 8.34375,
+ "grad_norm": 0.8270640865043484,
+ "learning_rate": 2e-05,
+ "loss": 0.5418,
+ "step": 267
+ },
+ {
+ "epoch": 8.34375,
+ "eval_loss": 0.7083099484443665,
+ "eval_runtime": 49.7373,
+ "eval_samples_per_second": 4.021,
+ "eval_steps_per_second": 0.261,
+ "step": 267
+ },
+ {
+ "epoch": 8.375,
+ "grad_norm": 0.8670483383004401,
+ "learning_rate": 2e-05,
+ "loss": 0.5935,
+ "step": 268
+ },
+ {
+ "epoch": 8.375,
+ "eval_loss": 0.7091077566146851,
+ "eval_runtime": 49.6764,
+ "eval_samples_per_second": 4.026,
+ "eval_steps_per_second": 0.262,
+ "step": 268
+ },
+ {
+ "epoch": 8.40625,
+ "grad_norm": 0.8373742279582174,
+ "learning_rate": 2e-05,
+ "loss": 0.5947,
+ "step": 269
+ },
+ {
+ "epoch": 8.40625,
+ "eval_loss": 0.709764301776886,
+ "eval_runtime": 49.5613,
+ "eval_samples_per_second": 4.035,
+ "eval_steps_per_second": 0.262,
+ "step": 269
+ },
+ {
+ "epoch": 8.4375,
+ "grad_norm": 0.9406584622840672,
+ "learning_rate": 2e-05,
+ "loss": 0.6079,
+ "step": 270
+ },
+ {
+ "epoch": 8.4375,
+ "eval_loss": 0.7089658379554749,
+ "eval_runtime": 49.6241,
+ "eval_samples_per_second": 4.03,
+ "eval_steps_per_second": 0.262,
+ "step": 270
+ },
+ {
+ "epoch": 8.46875,
+ "grad_norm": 0.9394463996884406,
+ "learning_rate": 2e-05,
+ "loss": 0.5102,
+ "step": 271
+ },
+ {
+ "epoch": 8.46875,
+ "eval_loss": 0.7126440405845642,
+ "eval_runtime": 50.6997,
+ "eval_samples_per_second": 3.945,
+ "eval_steps_per_second": 0.256,
+ "step": 271
+ },
+ {
+ "epoch": 8.5,
+ "grad_norm": 0.8618711805362732,
+ "learning_rate": 2e-05,
+ "loss": 0.5883,
+ "step": 272
+ },
+ {
+ "epoch": 8.5,
+ "eval_loss": 0.7210386395454407,
+ "eval_runtime": 47.7127,
+ "eval_samples_per_second": 4.192,
+ "eval_steps_per_second": 0.272,
+ "step": 272
+ },
+ {
+ "epoch": 8.53125,
+ "grad_norm": 0.9598465596200918,
+ "learning_rate": 2e-05,
+ "loss": 0.5958,
+ "step": 273
+ },
+ {
+ "epoch": 8.53125,
+ "eval_loss": 0.7250240445137024,
+ "eval_runtime": 47.5731,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 273
+ },
+ {
+ "epoch": 8.5625,
+ "grad_norm": 0.9512065591304456,
+ "learning_rate": 2e-05,
+ "loss": 0.5701,
+ "step": 274
+ },
+ {
+ "epoch": 8.5625,
+ "eval_loss": 0.7265011072158813,
+ "eval_runtime": 47.611,
+ "eval_samples_per_second": 4.201,
+ "eval_steps_per_second": 0.273,
+ "step": 274
+ },
+ {
+ "epoch": 8.59375,
+ "grad_norm": 1.0268459491950561,
+ "learning_rate": 2e-05,
+ "loss": 0.6169,
+ "step": 275
+ },
+ {
+ "epoch": 8.59375,
+ "eval_loss": 0.723859965801239,
+ "eval_runtime": 47.5959,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 275
+ },
+ {
+ "epoch": 8.625,
+ "grad_norm": 0.9424594037649877,
+ "learning_rate": 2e-05,
+ "loss": 0.6084,
+ "step": 276
+ },
+ {
+ "epoch": 8.625,
+ "eval_loss": 0.7198401093482971,
+ "eval_runtime": 49.4929,
+ "eval_samples_per_second": 4.041,
+ "eval_steps_per_second": 0.263,
+ "step": 276
+ },
+ {
+ "epoch": 8.65625,
+ "grad_norm": 0.9035217720347092,
+ "learning_rate": 2e-05,
+ "loss": 0.5512,
+ "step": 277
+ },
+ {
+ "epoch": 8.65625,
+ "eval_loss": 0.7168082594871521,
+ "eval_runtime": 49.6613,
+ "eval_samples_per_second": 4.027,
+ "eval_steps_per_second": 0.262,
+ "step": 277
+ },
+ {
+ "epoch": 8.6875,
+ "grad_norm": 0.8659031266239389,
+ "learning_rate": 2e-05,
+ "loss": 0.5863,
+ "step": 278
+ },
+ {
+ "epoch": 8.6875,
+ "eval_loss": 0.7159530520439148,
+ "eval_runtime": 49.5693,
+ "eval_samples_per_second": 4.035,
+ "eval_steps_per_second": 0.262,
+ "step": 278
+ },
+ {
+ "epoch": 8.71875,
+ "grad_norm": 0.8740167542953284,
+ "learning_rate": 2e-05,
+ "loss": 0.5667,
+ "step": 279
+ },
+ {
+ "epoch": 8.71875,
+ "eval_loss": 0.7145251631736755,
+ "eval_runtime": 49.4465,
+ "eval_samples_per_second": 4.045,
+ "eval_steps_per_second": 0.263,
+ "step": 279
+ },
+ {
+ "epoch": 8.75,
+ "grad_norm": 0.9263844516793406,
+ "learning_rate": 2e-05,
+ "loss": 0.6124,
+ "step": 280
+ },
+ {
+ "epoch": 8.75,
+ "eval_loss": 0.7149668335914612,
+ "eval_runtime": 49.6649,
+ "eval_samples_per_second": 4.027,
+ "eval_steps_per_second": 0.262,
+ "step": 280
+ },
+ {
+ "epoch": 8.78125,
+ "grad_norm": 0.8604543323600852,
+ "learning_rate": 2e-05,
+ "loss": 0.5688,
+ "step": 281
+ },
+ {
+ "epoch": 8.78125,
+ "eval_loss": 0.7160521149635315,
+ "eval_runtime": 50.6672,
+ "eval_samples_per_second": 3.947,
+ "eval_steps_per_second": 0.257,
+ "step": 281
+ },
+ {
+ "epoch": 8.8125,
+ "grad_norm": 0.9357009474127106,
+ "learning_rate": 2e-05,
+ "loss": 0.5463,
+ "step": 282
+ },
+ {
+ "epoch": 8.8125,
+ "eval_loss": 0.7187457084655762,
+ "eval_runtime": 50.6875,
+ "eval_samples_per_second": 3.946,
+ "eval_steps_per_second": 0.256,
+ "step": 282
+ },
+ {
+ "epoch": 8.84375,
+ "grad_norm": 0.8237087244624672,
+ "learning_rate": 2e-05,
+ "loss": 0.5393,
+ "step": 283
+ },
+ {
+ "epoch": 8.84375,
+ "eval_loss": 0.7205131649971008,
+ "eval_runtime": 50.5794,
+ "eval_samples_per_second": 3.954,
+ "eval_steps_per_second": 0.257,
+ "step": 283
+ },
+ {
+ "epoch": 8.875,
+ "grad_norm": 0.8962206816300475,
+ "learning_rate": 2e-05,
+ "loss": 0.484,
+ "step": 284
+ },
+ {
+ "epoch": 8.875,
+ "eval_loss": 0.7228506207466125,
+ "eval_runtime": 50.5953,
+ "eval_samples_per_second": 3.953,
+ "eval_steps_per_second": 0.257,
+ "step": 284
+ },
+ {
+ "epoch": 8.90625,
+ "grad_norm": 0.9983325109069782,
+ "learning_rate": 2e-05,
+ "loss": 0.5592,
+ "step": 285
+ },
+ {
+ "epoch": 8.90625,
+ "eval_loss": 0.7194100022315979,
+ "eval_runtime": 50.8657,
+ "eval_samples_per_second": 3.932,
+ "eval_steps_per_second": 0.256,
+ "step": 285
+ },
+ {
+ "epoch": 8.9375,
+ "grad_norm": 0.8875985843008509,
+ "learning_rate": 2e-05,
+ "loss": 0.6679,
+ "step": 286
+ },
+ {
+ "epoch": 8.9375,
+ "eval_loss": 0.7146596312522888,
+ "eval_runtime": 51.9576,
+ "eval_samples_per_second": 3.849,
+ "eval_steps_per_second": 0.25,
+ "step": 286
+ },
+ {
+ "epoch": 8.96875,
+ "grad_norm": 0.8611052694088349,
+ "learning_rate": 2e-05,
+ "loss": 0.5812,
+ "step": 287
+ },
+ {
+ "epoch": 8.96875,
+ "eval_loss": 0.710852861404419,
+ "eval_runtime": 51.9658,
+ "eval_samples_per_second": 3.849,
+ "eval_steps_per_second": 0.25,
+ "step": 287
+ },
+ {
+ "epoch": 9.0,
+ "grad_norm": 0.8497210900533776,
+ "learning_rate": 2e-05,
+ "loss": 0.5212,
+ "step": 288
+ },
+ {
+ "epoch": 9.0,
+ "eval_loss": 0.7121503353118896,
+ "eval_runtime": 51.6828,
+ "eval_samples_per_second": 3.87,
+ "eval_steps_per_second": 0.252,
+ "step": 288
+ },
+ {
+ "epoch": 9.03125,
+ "grad_norm": 0.8921157674462687,
+ "learning_rate": 2e-05,
+ "loss": 0.5437,
+ "step": 289
+ },
+ {
+ "epoch": 9.03125,
+ "eval_loss": 0.7179412841796875,
+ "eval_runtime": 51.9759,
+ "eval_samples_per_second": 3.848,
+ "eval_steps_per_second": 0.25,
+ "step": 289
+ },
+ {
+ "epoch": 9.0625,
+ "grad_norm": 0.9291292967074066,
+ "learning_rate": 2e-05,
+ "loss": 0.5679,
+ "step": 290
+ },
+ {
+ "epoch": 9.0625,
+ "eval_loss": 0.7306573390960693,
+ "eval_runtime": 51.603,
+ "eval_samples_per_second": 3.876,
+ "eval_steps_per_second": 0.252,
+ "step": 290
+ },
+ {
+ "epoch": 9.09375,
+ "grad_norm": 0.9871115113489229,
+ "learning_rate": 2e-05,
+ "loss": 0.5744,
+ "step": 291
+ },
+ {
+ "epoch": 9.09375,
+ "eval_loss": 0.74213707447052,
+ "eval_runtime": 51.5255,
+ "eval_samples_per_second": 3.882,
+ "eval_steps_per_second": 0.252,
+ "step": 291
+ },
+ {
+ "epoch": 9.125,
+ "grad_norm": 1.1662734879135015,
+ "learning_rate": 2e-05,
+ "loss": 0.5274,
+ "step": 292
+ },
+ {
+ "epoch": 9.125,
+ "eval_loss": 0.7484179139137268,
+ "eval_runtime": 51.3131,
+ "eval_samples_per_second": 3.898,
+ "eval_steps_per_second": 0.253,
+ "step": 292
+ },
+ {
+ "epoch": 9.15625,
+ "grad_norm": 1.096240777006249,
+ "learning_rate": 2e-05,
+ "loss": 0.5864,
+ "step": 293
+ },
+ {
+ "epoch": 9.15625,
+ "eval_loss": 0.745439887046814,
+ "eval_runtime": 51.1121,
+ "eval_samples_per_second": 3.913,
+ "eval_steps_per_second": 0.254,
+ "step": 293
+ },
+ {
+ "epoch": 9.1875,
+ "grad_norm": 0.944903135330694,
+ "learning_rate": 2e-05,
+ "loss": 0.5131,
+ "step": 294
+ },
+ {
+ "epoch": 9.1875,
+ "eval_loss": 0.7430945038795471,
+ "eval_runtime": 51.307,
+ "eval_samples_per_second": 3.898,
+ "eval_steps_per_second": 0.253,
+ "step": 294
+ },
+ {
+ "epoch": 9.21875,
+ "grad_norm": 1.0736115005040638,
+ "learning_rate": 2e-05,
+ "loss": 0.4866,
+ "step": 295
+ },
+ {
+ "epoch": 9.21875,
+ "eval_loss": 0.7417933940887451,
+ "eval_runtime": 51.2372,
+ "eval_samples_per_second": 3.903,
+ "eval_steps_per_second": 0.254,
+ "step": 295
+ },
+ {
+ "epoch": 9.25,
+ "grad_norm": 1.0688144195951634,
+ "learning_rate": 2e-05,
+ "loss": 0.509,
+ "step": 296
+ },
+ {
+ "epoch": 9.25,
+ "eval_loss": 0.7381229996681213,
+ "eval_runtime": 51.1494,
+ "eval_samples_per_second": 3.91,
+ "eval_steps_per_second": 0.254,
+ "step": 296
+ },
+ {
+ "epoch": 9.28125,
+ "grad_norm": 1.0276146013155785,
+ "learning_rate": 2e-05,
+ "loss": 0.5708,
+ "step": 297
+ },
+ {
+ "epoch": 9.28125,
+ "eval_loss": 0.7391738891601562,
+ "eval_runtime": 51.6779,
+ "eval_samples_per_second": 3.87,
+ "eval_steps_per_second": 0.252,
+ "step": 297
+ },
+ {
+ "epoch": 9.3125,
+ "grad_norm": 1.1618114955183,
+ "learning_rate": 2e-05,
+ "loss": 0.5337,
+ "step": 298
+ },
+ {
+ "epoch": 9.3125,
+ "eval_loss": 0.7411096096038818,
+ "eval_runtime": 51.5937,
+ "eval_samples_per_second": 3.876,
+ "eval_steps_per_second": 0.252,
+ "step": 298
+ },
+ {
+ "epoch": 9.34375,
+ "grad_norm": 1.08837375836462,
+ "learning_rate": 2e-05,
+ "loss": 0.5241,
+ "step": 299
+ },
+ {
+ "epoch": 9.34375,
+ "eval_loss": 0.7420552968978882,
+ "eval_runtime": 51.5437,
+ "eval_samples_per_second": 3.88,
+ "eval_steps_per_second": 0.252,
+ "step": 299
+ },
+ {
+ "epoch": 9.375,
+ "grad_norm": 1.0106379800787466,
+ "learning_rate": 2e-05,
+ "loss": 0.5198,
+ "step": 300
+ },
+ {
+ "epoch": 9.375,
+ "eval_loss": 0.7437419295310974,
+ "eval_runtime": 51.3565,
+ "eval_samples_per_second": 3.894,
+ "eval_steps_per_second": 0.253,
+ "step": 300
+ },
+ {
+ "epoch": 9.40625,
+ "grad_norm": 1.0700897207702011,
+ "learning_rate": 2e-05,
+ "loss": 0.5107,
+ "step": 301
+ },
+ {
+ "epoch": 9.40625,
+ "eval_loss": 0.7382708787918091,
+ "eval_runtime": 51.4533,
+ "eval_samples_per_second": 3.887,
+ "eval_steps_per_second": 0.253,
+ "step": 301
+ },
+ {
+ "epoch": 9.4375,
+ "grad_norm": 1.1021606769115393,
+ "learning_rate": 2e-05,
+ "loss": 0.5679,
+ "step": 302
+ },
+ {
+ "epoch": 9.4375,
+ "eval_loss": 0.7324429154396057,
+ "eval_runtime": 51.4117,
+ "eval_samples_per_second": 3.89,
+ "eval_steps_per_second": 0.253,
+ "step": 302
+ },
+ {
+ "epoch": 9.46875,
+ "grad_norm": 0.9792628984982289,
+ "learning_rate": 2e-05,
+ "loss": 0.5509,
+ "step": 303
+ },
+ {
+ "epoch": 9.46875,
+ "eval_loss": 0.7311490774154663,
+ "eval_runtime": 51.8022,
+ "eval_samples_per_second": 3.861,
+ "eval_steps_per_second": 0.251,
+ "step": 303
+ },
+ {
+ "epoch": 9.5,
+ "grad_norm": 0.9256898215171215,
+ "learning_rate": 2e-05,
+ "loss": 0.5824,
+ "step": 304
+ },
+ {
+ "epoch": 9.5,
+ "eval_loss": 0.736283540725708,
+ "eval_runtime": 51.7678,
+ "eval_samples_per_second": 3.863,
+ "eval_steps_per_second": 0.251,
+ "step": 304
+ },
+ {
+ "epoch": 9.53125,
+ "grad_norm": 0.993495109546069,
+ "learning_rate": 2e-05,
+ "loss": 0.5452,
+ "step": 305
+ },
+ {
+ "epoch": 9.53125,
+ "eval_loss": 0.7425567507743835,
+ "eval_runtime": 51.6022,
+ "eval_samples_per_second": 3.876,
+ "eval_steps_per_second": 0.252,
+ "step": 305
+ },
+ {
+ "epoch": 9.5625,
+ "grad_norm": 1.096995253097988,
+ "learning_rate": 2e-05,
+ "loss": 0.5359,
+ "step": 306
+ },
+ {
+ "epoch": 9.5625,
+ "eval_loss": 0.7483149766921997,
+ "eval_runtime": 51.5727,
+ "eval_samples_per_second": 3.878,
+ "eval_steps_per_second": 0.252,
+ "step": 306
+ },
+ {
+ "epoch": 9.59375,
+ "grad_norm": 1.1542996117677211,
+ "learning_rate": 2e-05,
+ "loss": 0.5229,
+ "step": 307
+ },
+ {
+ "epoch": 9.59375,
+ "eval_loss": 0.7505038380622864,
+ "eval_runtime": 51.846,
+ "eval_samples_per_second": 3.858,
+ "eval_steps_per_second": 0.251,
+ "step": 307
+ },
+ {
+ "epoch": 9.625,
+ "grad_norm": 1.1044494998416634,
+ "learning_rate": 2e-05,
+ "loss": 0.5718,
+ "step": 308
+ },
+ {
+ "epoch": 9.625,
+ "eval_loss": 0.7511885166168213,
+ "eval_runtime": 51.613,
+ "eval_samples_per_second": 3.875,
+ "eval_steps_per_second": 0.252,
+ "step": 308
+ },
+ {
+ "epoch": 9.65625,
+ "grad_norm": 1.0517094139644794,
+ "learning_rate": 2e-05,
+ "loss": 0.5395,
+ "step": 309
+ },
+ {
+ "epoch": 9.65625,
+ "eval_loss": 0.750588059425354,
+ "eval_runtime": 51.9083,
+ "eval_samples_per_second": 3.853,
+ "eval_steps_per_second": 0.25,
+ "step": 309
+ },
+ {
+ "epoch": 9.6875,
+ "grad_norm": 1.2320471917997522,
+ "learning_rate": 2e-05,
+ "loss": 0.5266,
+ "step": 310
+ },
+ {
+ "epoch": 9.6875,
+ "eval_loss": 0.7492180466651917,
+ "eval_runtime": 51.3612,
+ "eval_samples_per_second": 3.894,
+ "eval_steps_per_second": 0.253,
+ "step": 310
+ },
+ {
+ "epoch": 9.71875,
+ "grad_norm": 1.189122697506972,
+ "learning_rate": 2e-05,
+ "loss": 0.4893,
+ "step": 311
+ },
+ {
+ "epoch": 9.71875,
+ "eval_loss": 0.7448427081108093,
+ "eval_runtime": 51.8761,
+ "eval_samples_per_second": 3.855,
+ "eval_steps_per_second": 0.251,
+ "step": 311
+ },
+ {
+ "epoch": 9.75,
+ "grad_norm": 1.1250245833360049,
+ "learning_rate": 2e-05,
+ "loss": 0.5434,
+ "step": 312
+ },
+ {
+ "epoch": 9.75,
+ "eval_loss": 0.742850661277771,
+ "eval_runtime": 51.4442,
+ "eval_samples_per_second": 3.888,
+ "eval_steps_per_second": 0.253,
+ "step": 312
+ },
+ {
+ "epoch": 9.78125,
+ "grad_norm": 1.0320917220089818,
+ "learning_rate": 2e-05,
+ "loss": 0.539,
+ "step": 313
+ },
+ {
+ "epoch": 9.78125,
+ "eval_loss": 0.7389761209487915,
+ "eval_runtime": 51.609,
+ "eval_samples_per_second": 3.875,
+ "eval_steps_per_second": 0.252,
+ "step": 313
+ },
+ {
+ "epoch": 9.8125,
+ "grad_norm": 1.1419373892040323,
+ "learning_rate": 2e-05,
+ "loss": 0.5077,
+ "step": 314
+ },
+ {
+ "epoch": 9.8125,
+ "eval_loss": 0.7384924292564392,
+ "eval_runtime": 51.6937,
+ "eval_samples_per_second": 3.869,
+ "eval_steps_per_second": 0.251,
+ "step": 314
+ },
+ {
+ "epoch": 9.84375,
+ "grad_norm": 1.0260401820964369,
+ "learning_rate": 2e-05,
+ "loss": 0.534,
+ "step": 315
+ },
+ {
+ "epoch": 9.84375,
+ "eval_loss": 0.738023579120636,
+ "eval_runtime": 51.5428,
+ "eval_samples_per_second": 3.88,
+ "eval_steps_per_second": 0.252,
+ "step": 315
+ },
+ {
+ "epoch": 9.875,
+ "grad_norm": 1.0164514553564235,
+ "learning_rate": 2e-05,
+ "loss": 0.5514,
+ "step": 316
+ },
+ {
+ "epoch": 9.875,
+ "eval_loss": 0.7399526834487915,
+ "eval_runtime": 51.6232,
+ "eval_samples_per_second": 3.874,
+ "eval_steps_per_second": 0.252,
+ "step": 316
+ },
+ {
+ "epoch": 9.90625,
+ "grad_norm": 1.1847056085947891,
+ "learning_rate": 2e-05,
+ "loss": 0.5216,
+ "step": 317
+ },
+ {
+ "epoch": 9.90625,
+ "eval_loss": 0.7401251196861267,
+ "eval_runtime": 51.7617,
+ "eval_samples_per_second": 3.864,
+ "eval_steps_per_second": 0.251,
+ "step": 317
+ },
+ {
+ "epoch": 9.9375,
+ "grad_norm": 1.075888871715244,
+ "learning_rate": 2e-05,
+ "loss": 0.511,
+ "step": 318
+ },
+ {
+ "epoch": 9.9375,
+ "eval_loss": 0.739520788192749,
+ "eval_runtime": 51.7458,
+ "eval_samples_per_second": 3.865,
+ "eval_steps_per_second": 0.251,
+ "step": 318
+ },
+ {
+ "epoch": 9.96875,
+ "grad_norm": 1.16238118046427,
+ "learning_rate": 2e-05,
+ "loss": 0.546,
+ "step": 319
+ },
+ {
+ "epoch": 9.96875,
+ "eval_loss": 0.7371450662612915,
+ "eval_runtime": 51.4519,
+ "eval_samples_per_second": 3.887,
+ "eval_steps_per_second": 0.253,
+ "step": 319
+ },
+ {
+ "epoch": 10.0,
+ "grad_norm": 1.109611378591182,
+ "learning_rate": 2e-05,
+ "loss": 0.4855,
+ "step": 320
+ },
+ {
+ "epoch": 10.0,
+ "eval_loss": 0.7406165599822998,
+ "eval_runtime": 51.6984,
+ "eval_samples_per_second": 3.869,
+ "eval_steps_per_second": 0.251,
+ "step": 320
+ }
+ ],
+ "logging_steps": 1.0,
+ "max_steps": 320,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 10,
+ "save_steps": 5,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 461377729855488.0,
+ "train_batch_size": 8,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-320/training_args.bin b/checkpoint-320/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..197f502ac6603d740d3dd433a661ce8fd5d89125
--- /dev/null
+++ b/checkpoint-320/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01c288b68aca2db1424771fefc72a3d0ade725c40fa055d7766bbc2e9001652d
+size 8248
diff --git a/checkpoint-320/zero_to_fp32.py b/checkpoint-320/zero_to_fp32.py
new file mode 100755
index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8
--- /dev/null
+++ b/checkpoint-320/zero_to_fp32.py
@@ -0,0 +1,604 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+ buffers: dict()
+ param_shapes: dict()
+ shared_params: list
+ ds_version: int
+ frozen_param_shapes: dict()
+ frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+ return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+ '''
+ alist.sort(key=natural_keys) sorts in human order
+ http://nedbatchelder.com/blog/200712/human_sorting.html
+ (See Toothy's implementation in the comments)
+ '''
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+ if not os.path.isdir(checkpoint_dir):
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+ # there should be only one file
+ if zero_stage <= 2:
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+ elif zero_stage == 3:
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+ if not os.path.exists(file):
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+ return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+ # XXX: need to test that this simple glob rule works for multi-node setup too
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+ if len(ckpt_files) == 0:
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+ return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+ zero_model_states = []
+ for file in files:
+ state_dict = torch.load(file, map_location=device)
+
+ if BUFFER_NAMES not in state_dict:
+ raise ValueError(f"{file} is not a model state checkpoint")
+ buffer_names = state_dict[BUFFER_NAMES]
+ if debug:
+ print("Found buffers:", buffer_names)
+
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+ param_shapes = state_dict[PARAM_SHAPES]
+
+ # collect parameters that are included in param_shapes
+ param_names = []
+ for s in param_shapes:
+ for name in s.keys():
+ param_names.append(name)
+
+ # update with frozen parameters
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+ if frozen_param_shapes is not None:
+ if debug:
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+ param_names += list(frozen_param_shapes.keys())
+
+ # handle shared params
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+ ds_version = state_dict.get(DS_VERSION, None)
+
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+ z_model_state = zero_model_state(buffers=buffers,
+ param_shapes=param_shapes,
+ shared_params=shared_params,
+ ds_version=ds_version,
+ frozen_param_shapes=frozen_param_shapes,
+ frozen_param_fragments=frozen_param_fragments)
+ zero_model_states.append(z_model_state)
+
+ return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+ total_files = len(files)
+ state_dicts = []
+ for f in files:
+ state_dict = torch.load(f, map_location=device)
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+ # and also handle the case where it was already removed by another helper script
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+ state_dicts.append(state_dict)
+
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
+ # use the max of the partition_count to get the dp world_size.
+
+ if type(world_size) is list:
+ world_size = max(world_size)
+
+ if world_size != total_files:
+ raise ValueError(
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+ )
+
+ # the groups are named differently in each stage
+ if zero_stage <= 2:
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+ elif zero_stage == 3:
+ fp32_groups_key = FP32_FLAT_GROUPS
+ else:
+ raise ValueError(f"unknown zero stage {zero_stage}")
+
+ if zero_stage <= 2:
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+ elif zero_stage == 3:
+ # if there is more than one param group, there will be multiple flattened tensors - one
+ # flattened tensor per group - for simplicity merge them into a single tensor
+ #
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+ fp32_flat_groups = [
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+ ]
+
+ return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+ """
+ Returns fp32 state_dict reconstructed from ds checkpoint
+
+ Args:
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+ """
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+ optim_files = get_optim_files(ds_checkpoint_dir)
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+ model_files = get_model_state_files(ds_checkpoint_dir)
+
+ zero_model_states = parse_model_states(model_files)
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+ if zero_stage <= 2:
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters)
+ elif zero_stage == 3:
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+ return
+
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+ if debug:
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+ wanted_params = len(frozen_param_shapes)
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+ print(f'Frozen params: Have {avail_numel} numels to process.')
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+ total_params = 0
+ total_numel = 0
+ for name, shape in frozen_param_shapes.items():
+ total_params += 1
+ unpartitioned_numel = shape.numel()
+ total_numel += unpartitioned_numel
+
+ state_dict[name] = frozen_param_fragments[name]
+
+ if debug:
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+ attr = getattr(obj, fn, None)
+ return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+ param_shapes = zero_model_states[0].param_shapes
+
+ # Reconstruction protocol:
+ #
+ # XXX: document this
+
+ if debug:
+ for i in range(world_size):
+ for j in range(len(fp32_flat_groups[0])):
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+ # XXX: memory usage doubles here (zero2)
+ num_param_groups = len(fp32_flat_groups[0])
+ merged_single_partition_of_fp32_groups = []
+ for i in range(num_param_groups):
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+ avail_numel = sum(
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+ if debug:
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+ # not asserting if there is a mismatch due to possible padding
+ print(f"Have {avail_numel} numels to process.")
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+ # params
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+ # out-of-core computing solution
+ total_numel = 0
+ total_params = 0
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+ offset = 0
+ avail_numel = full_single_fp32_vector.numel()
+ for name, shape in shapes.items():
+
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+ total_numel += unpartitioned_numel
+ total_params += 1
+
+ if debug:
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+ offset += unpartitioned_numel
+
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+ # live optimizer object, so we are checking that the numbers are within the right range
+ align_to = 2 * world_size
+
+ def zero2_align(x):
+ return align_to * math.ceil(x / align_to)
+
+ if debug:
+ print(f"original offset={offset}, avail_numel={avail_numel}")
+
+ offset = zero2_align(offset)
+ avail_numel = zero2_align(avail_numel)
+
+ if debug:
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
+
+ # Sanity check
+ if offset != avail_numel:
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters):
+ state_dict = OrderedDict()
+
+ # buffers
+ buffers = zero_model_states[0].buffers
+ state_dict.update(buffers)
+ if debug:
+ print(f"added {len(buffers)} buffers")
+
+ if not exclude_frozen_parameters:
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+ # recover shared parameters
+ for pair in zero_model_states[0].shared_params:
+ if pair[1] in state_dict:
+ state_dict[pair[0]] = state_dict[pair[1]]
+
+ return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+ remainder = unpartitioned_numel % world_size
+ padding_numel = (world_size - remainder) if remainder else 0
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+ return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+ return
+
+ if debug:
+ for i in range(world_size):
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+ wanted_params = len(frozen_param_shapes)
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+ print(f'Frozen params: Have {avail_numel} numels to process.')
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+ total_params = 0
+ total_numel = 0
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
+ total_params += 1
+ unpartitioned_numel = shape.numel()
+ total_numel += unpartitioned_numel
+
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+ if debug:
+ print(
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+ )
+
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+ param_shapes = zero_model_states[0].param_shapes
+ avail_numel = fp32_flat_groups[0].numel() * world_size
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+ # param, re-consolidating each param, while dealing with padding if any
+
+ # merge list of dicts, preserving order
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+ if debug:
+ for i in range(world_size):
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+ wanted_params = len(param_shapes)
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+ # not asserting if there is a mismatch due to possible padding
+ avail_numel = fp32_flat_groups[0].numel() * world_size
+ print(f"Trainable params: Have {avail_numel} numels to process.")
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+ # params
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+ # out-of-core computing solution
+ offset = 0
+ total_numel = 0
+ total_params = 0
+ for name, shape in param_shapes.items():
+
+ unpartitioned_numel = shape.numel()
+ total_numel += unpartitioned_numel
+ total_params += 1
+
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+ if debug:
+ print(
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+ )
+
+ # XXX: memory usage doubles here
+ state_dict[name] = torch.cat(
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
+ offset += partitioned_numel
+
+ offset *= world_size
+
+ # Sanity check
+ if offset != avail_numel:
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+ exclude_frozen_parameters):
+ state_dict = OrderedDict()
+
+ # buffers
+ buffers = zero_model_states[0].buffers
+ state_dict.update(buffers)
+ if debug:
+ print(f"added {len(buffers)} buffers")
+
+ if not exclude_frozen_parameters:
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+ # recover shared parameters
+ for pair in zero_model_states[0].shared_params:
+ if pair[1] in state_dict:
+ state_dict[pair[0]] = state_dict[pair[1]]
+
+ return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+ """
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+ via a model hub.
+
+ Args:
+ - ``checkpoint_dir``: path to the desired checkpoint folder
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+ - ``exclude_frozen_parameters``: exclude frozen parameters
+
+ Returns:
+ - pytorch ``state_dict``
+
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+ the checkpoint.
+
+ A typical usage might be ::
+
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+ # do the training and checkpoint saving
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+ model = model.cpu() # move to cpu
+ model.load_state_dict(state_dict)
+ # submit to model hub or save the model to share with others
+
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
+ application. i.e. you will need to re-initialize the deepspeed engine, since
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+ """
+ if tag is None:
+ latest_path = os.path.join(checkpoint_dir, 'latest')
+ if os.path.isfile(latest_path):
+ with open(latest_path, 'r') as fd:
+ tag = fd.read().strip()
+ else:
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+ if not os.path.isdir(ds_checkpoint_dir):
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
+ """
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+ Args:
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+ - ``exclude_frozen_parameters``: exclude frozen parameters
+ """
+
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+ print(f"Saving fp32 state dict to {output_file}")
+ torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+ """
+ 1. Put the provided model to cpu
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+ 3. Load it into the provided model
+
+ Args:
+ - ``model``: the model object to update
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+ Returns:
+ - ``model`: modified model
+
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+ conveniently placed for you in the checkpoint folder.
+
+ A typical usage might be ::
+
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+ # submit to model hub or save the model to share with others
+
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+ """
+ logger.info(f"Extracting fp32 weights")
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+ logger.info(f"Overwriting model with fp32 weights")
+ model = model.cpu()
+ model.load_state_dict(state_dict, strict=False)
+
+ return model
+
+
+if __name__ == "__main__":
+
+ parser = argparse.ArgumentParser()
+ parser.add_argument("checkpoint_dir",
+ type=str,
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+ parser.add_argument(
+ "output_file",
+ type=str,
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+ parser.add_argument("-t",
+ "--tag",
+ type=str,
+ default=None,
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+ args = parser.parse_args()
+
+ debug = args.debug
+
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+ args.output_file,
+ tag=args.tag,
+ exclude_frozen_parameters=args.exclude_frozen_parameters)
diff --git a/config.json b/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9461d19c5605a107241c0300f3cffdeb257c667b
--- /dev/null
+++ b/config.json
@@ -0,0 +1,52 @@
+{
+ "_attn_implementation_autoset": true,
+ "_name_or_path": "liuhaotian/llava-v1.5-13b",
+ "architectures": [
+ "LlavaLlamaForCausalLM"
+ ],
+ "attention_bias": false,
+ "attention_dropout": 0.0,
+ "bos_token_id": 1,
+ "eos_token_id": 2,
+ "freeze_mm_mlp_adapter": false,
+ "freeze_mm_vision_resampler": false,
+ "head_dim": 128,
+ "hidden_act": "silu",
+ "hidden_size": 5120,
+ "image_aspect_ratio": "pad",
+ "initializer_range": 0.02,
+ "intermediate_size": 13824,
+ "max_length": 4096,
+ "max_position_embeddings": 4096,
+ "mlp_bias": false,
+ "mm_hidden_size": 1024,
+ "mm_patch_merge_type": "flat",
+ "mm_projector_lr": 2e-05,
+ "mm_projector_type": "mlp2x_gelu",
+ "mm_resampler_type": null,
+ "mm_use_im_patch_token": false,
+ "mm_use_im_start_end": false,
+ "mm_vision_select_feature": "patch",
+ "mm_vision_select_layer": -2,
+ "mm_vision_tower": "openai/clip-vit-large-patch14-336",
+ "model_type": "llava_llama",
+ "num_attention_heads": 40,
+ "num_hidden_layers": 40,
+ "num_key_value_heads": 40,
+ "pad_token_id": 0,
+ "pretraining_tp": 1,
+ "rms_norm_eps": 1e-05,
+ "rope_scaling": null,
+ "rope_theta": 10000.0,
+ "tie_word_embeddings": false,
+ "tokenizer_model_max_length": 2048,
+ "tokenizer_padding_side": "right",
+ "torch_dtype": "float16",
+ "transformers_version": "4.46.0",
+ "tune_mm_mlp_adapter": false,
+ "tune_mm_vision_resampler": false,
+ "unfreeze_mm_vision_tower": false,
+ "use_cache": true,
+ "use_mm_proj": true,
+ "vocab_size": 32000
+}
diff --git a/non_lora_trainables.bin b/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..7623c2a7a2c4965d4dd6b5a3bceee18c6b46d9fe
--- /dev/null
+++ b/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6df2742e15c72ec9864fc1273e3cf58fe5d383be2bd94bd9c4c3f9b0c636c0c
+size 62937264
diff --git a/optimizer.pt b/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5a18c9211933e4ef997cd70a90b82212516100ae
--- /dev/null
+++ b/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f3e9883e556f25ad08adc0128c738f5af230da2703707d2fefab59aab93ff82
+size 188285474
diff --git a/trainer_state.json b/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..82f1775ce4e3e5265a4e46fe41059957f9118bfa
--- /dev/null
+++ b/trainer_state.json
@@ -0,0 +1,4842 @@
+{
+ "best_metric": 0.6895740032196045,
+ "best_model_checkpoint": "./checkpoints/llava-v1.5-13b/checkpoint-224",
+ "epoch": 10.0,
+ "eval_steps": 1.0,
+ "global_step": 320,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.03125,
+ "grad_norm": 0.2380081706918525,
+ "learning_rate": 0.0,
+ "loss": 1.2458,
+ "step": 1
+ },
+ {
+ "epoch": 0.03125,
+ "eval_loss": 1.3161638975143433,
+ "eval_runtime": 50.8995,
+ "eval_samples_per_second": 3.929,
+ "eval_steps_per_second": 0.255,
+ "step": 1
+ },
+ {
+ "epoch": 0.0625,
+ "grad_norm": 0.20429495268987705,
+ "learning_rate": 8.613531161467863e-06,
+ "loss": 1.2003,
+ "step": 2
+ },
+ {
+ "epoch": 0.0625,
+ "eval_loss": 1.3161638975143433,
+ "eval_runtime": 47.4818,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 2
+ },
+ {
+ "epoch": 0.09375,
+ "grad_norm": 0.20616215800420787,
+ "learning_rate": 1.3652123889719709e-05,
+ "loss": 1.2622,
+ "step": 3
+ },
+ {
+ "epoch": 0.09375,
+ "eval_loss": 1.309991478919983,
+ "eval_runtime": 47.4152,
+ "eval_samples_per_second": 4.218,
+ "eval_steps_per_second": 0.274,
+ "step": 3
+ },
+ {
+ "epoch": 0.125,
+ "grad_norm": 0.20155595022101944,
+ "learning_rate": 1.7227062322935725e-05,
+ "loss": 1.2845,
+ "step": 4
+ },
+ {
+ "epoch": 0.125,
+ "eval_loss": 1.3013781309127808,
+ "eval_runtime": 47.4814,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 4
+ },
+ {
+ "epoch": 0.15625,
+ "grad_norm": 0.21113117474989132,
+ "learning_rate": 2e-05,
+ "loss": 1.246,
+ "step": 5
+ },
+ {
+ "epoch": 0.15625,
+ "eval_loss": 1.2892160415649414,
+ "eval_runtime": 47.7209,
+ "eval_samples_per_second": 4.191,
+ "eval_steps_per_second": 0.272,
+ "step": 5
+ },
+ {
+ "epoch": 0.1875,
+ "grad_norm": 0.21377946631015488,
+ "learning_rate": 2e-05,
+ "loss": 1.2684,
+ "step": 6
+ },
+ {
+ "epoch": 0.1875,
+ "eval_loss": 1.2754532098770142,
+ "eval_runtime": 47.5781,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 6
+ },
+ {
+ "epoch": 0.21875,
+ "grad_norm": 0.2284268997618767,
+ "learning_rate": 2e-05,
+ "loss": 1.2681,
+ "step": 7
+ },
+ {
+ "epoch": 0.21875,
+ "eval_loss": 1.2605774402618408,
+ "eval_runtime": 47.5326,
+ "eval_samples_per_second": 4.208,
+ "eval_steps_per_second": 0.273,
+ "step": 7
+ },
+ {
+ "epoch": 0.25,
+ "grad_norm": 0.23585343568544442,
+ "learning_rate": 2e-05,
+ "loss": 1.2407,
+ "step": 8
+ },
+ {
+ "epoch": 0.25,
+ "eval_loss": 1.244718313217163,
+ "eval_runtime": 47.5001,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 8
+ },
+ {
+ "epoch": 0.28125,
+ "grad_norm": 0.23051191992462533,
+ "learning_rate": 2e-05,
+ "loss": 1.2766,
+ "step": 9
+ },
+ {
+ "epoch": 0.28125,
+ "eval_loss": 1.2285138368606567,
+ "eval_runtime": 47.4631,
+ "eval_samples_per_second": 4.214,
+ "eval_steps_per_second": 0.274,
+ "step": 9
+ },
+ {
+ "epoch": 0.3125,
+ "grad_norm": 0.22726394327484983,
+ "learning_rate": 2e-05,
+ "loss": 1.2024,
+ "step": 10
+ },
+ {
+ "epoch": 0.3125,
+ "eval_loss": 1.2118008136749268,
+ "eval_runtime": 47.4991,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 10
+ },
+ {
+ "epoch": 0.34375,
+ "grad_norm": 0.25404890894461285,
+ "learning_rate": 2e-05,
+ "loss": 1.2742,
+ "step": 11
+ },
+ {
+ "epoch": 0.34375,
+ "eval_loss": 1.1942989826202393,
+ "eval_runtime": 49.2609,
+ "eval_samples_per_second": 4.06,
+ "eval_steps_per_second": 0.264,
+ "step": 11
+ },
+ {
+ "epoch": 0.375,
+ "grad_norm": 0.26336210916526287,
+ "learning_rate": 2e-05,
+ "loss": 1.2258,
+ "step": 12
+ },
+ {
+ "epoch": 0.375,
+ "eval_loss": 1.176426649093628,
+ "eval_runtime": 49.0639,
+ "eval_samples_per_second": 4.076,
+ "eval_steps_per_second": 0.265,
+ "step": 12
+ },
+ {
+ "epoch": 0.40625,
+ "grad_norm": 0.29637148470746666,
+ "learning_rate": 2e-05,
+ "loss": 1.2345,
+ "step": 13
+ },
+ {
+ "epoch": 0.40625,
+ "eval_loss": 1.1577811241149902,
+ "eval_runtime": 49.1352,
+ "eval_samples_per_second": 4.07,
+ "eval_steps_per_second": 0.265,
+ "step": 13
+ },
+ {
+ "epoch": 0.4375,
+ "grad_norm": 0.2841880377627424,
+ "learning_rate": 2e-05,
+ "loss": 1.0765,
+ "step": 14
+ },
+ {
+ "epoch": 0.4375,
+ "eval_loss": 1.1381279230117798,
+ "eval_runtime": 49.25,
+ "eval_samples_per_second": 4.061,
+ "eval_steps_per_second": 0.264,
+ "step": 14
+ },
+ {
+ "epoch": 0.46875,
+ "grad_norm": 0.2773140636191091,
+ "learning_rate": 2e-05,
+ "loss": 1.1812,
+ "step": 15
+ },
+ {
+ "epoch": 0.46875,
+ "eval_loss": 1.1178216934204102,
+ "eval_runtime": 49.0879,
+ "eval_samples_per_second": 4.074,
+ "eval_steps_per_second": 0.265,
+ "step": 15
+ },
+ {
+ "epoch": 0.5,
+ "grad_norm": 0.3568607365552051,
+ "learning_rate": 2e-05,
+ "loss": 1.1327,
+ "step": 16
+ },
+ {
+ "epoch": 0.5,
+ "eval_loss": 1.0954149961471558,
+ "eval_runtime": 48.6546,
+ "eval_samples_per_second": 4.111,
+ "eval_steps_per_second": 0.267,
+ "step": 16
+ },
+ {
+ "epoch": 0.53125,
+ "grad_norm": 0.32574391414112897,
+ "learning_rate": 2e-05,
+ "loss": 1.1162,
+ "step": 17
+ },
+ {
+ "epoch": 0.53125,
+ "eval_loss": 1.071275234222412,
+ "eval_runtime": 48.5618,
+ "eval_samples_per_second": 4.118,
+ "eval_steps_per_second": 0.268,
+ "step": 17
+ },
+ {
+ "epoch": 0.5625,
+ "grad_norm": 0.4256864144638081,
+ "learning_rate": 2e-05,
+ "loss": 1.1138,
+ "step": 18
+ },
+ {
+ "epoch": 0.5625,
+ "eval_loss": 1.0455905199050903,
+ "eval_runtime": 48.4981,
+ "eval_samples_per_second": 4.124,
+ "eval_steps_per_second": 0.268,
+ "step": 18
+ },
+ {
+ "epoch": 0.59375,
+ "grad_norm": 0.31230014132112643,
+ "learning_rate": 2e-05,
+ "loss": 1.0011,
+ "step": 19
+ },
+ {
+ "epoch": 0.59375,
+ "eval_loss": 1.0208789110183716,
+ "eval_runtime": 48.4675,
+ "eval_samples_per_second": 4.126,
+ "eval_steps_per_second": 0.268,
+ "step": 19
+ },
+ {
+ "epoch": 0.625,
+ "grad_norm": 0.3025724039243594,
+ "learning_rate": 2e-05,
+ "loss": 1.109,
+ "step": 20
+ },
+ {
+ "epoch": 0.625,
+ "eval_loss": 1.002480149269104,
+ "eval_runtime": 48.5265,
+ "eval_samples_per_second": 4.121,
+ "eval_steps_per_second": 0.268,
+ "step": 20
+ },
+ {
+ "epoch": 0.65625,
+ "grad_norm": 0.27787879590501874,
+ "learning_rate": 2e-05,
+ "loss": 1.0291,
+ "step": 21
+ },
+ {
+ "epoch": 0.65625,
+ "eval_loss": 0.9933492541313171,
+ "eval_runtime": 50.0369,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 21
+ },
+ {
+ "epoch": 0.6875,
+ "grad_norm": 0.4231294067130801,
+ "learning_rate": 2e-05,
+ "loss": 1.0779,
+ "step": 22
+ },
+ {
+ "epoch": 0.6875,
+ "eval_loss": 0.9850385785102844,
+ "eval_runtime": 50.0062,
+ "eval_samples_per_second": 4.0,
+ "eval_steps_per_second": 0.26,
+ "step": 22
+ },
+ {
+ "epoch": 0.71875,
+ "grad_norm": 0.42130097437373987,
+ "learning_rate": 2e-05,
+ "loss": 1.0897,
+ "step": 23
+ },
+ {
+ "epoch": 0.71875,
+ "eval_loss": 0.9758670330047607,
+ "eval_runtime": 50.1031,
+ "eval_samples_per_second": 3.992,
+ "eval_steps_per_second": 0.259,
+ "step": 23
+ },
+ {
+ "epoch": 0.75,
+ "grad_norm": 0.27711808063263893,
+ "learning_rate": 2e-05,
+ "loss": 1.0739,
+ "step": 24
+ },
+ {
+ "epoch": 0.75,
+ "eval_loss": 0.9674506187438965,
+ "eval_runtime": 50.0337,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 24
+ },
+ {
+ "epoch": 0.78125,
+ "grad_norm": 0.2879649409281791,
+ "learning_rate": 2e-05,
+ "loss": 1.0182,
+ "step": 25
+ },
+ {
+ "epoch": 0.78125,
+ "eval_loss": 0.9592065215110779,
+ "eval_runtime": 50.0709,
+ "eval_samples_per_second": 3.994,
+ "eval_steps_per_second": 0.26,
+ "step": 25
+ },
+ {
+ "epoch": 0.8125,
+ "grad_norm": 0.19327450826076825,
+ "learning_rate": 2e-05,
+ "loss": 1.0413,
+ "step": 26
+ },
+ {
+ "epoch": 0.8125,
+ "eval_loss": 0.9518552422523499,
+ "eval_runtime": 50.0572,
+ "eval_samples_per_second": 3.995,
+ "eval_steps_per_second": 0.26,
+ "step": 26
+ },
+ {
+ "epoch": 0.84375,
+ "grad_norm": 0.19707021382445633,
+ "learning_rate": 2e-05,
+ "loss": 0.9525,
+ "step": 27
+ },
+ {
+ "epoch": 0.84375,
+ "eval_loss": 0.9449941515922546,
+ "eval_runtime": 50.0515,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 27
+ },
+ {
+ "epoch": 0.875,
+ "grad_norm": 0.2420270757641518,
+ "learning_rate": 2e-05,
+ "loss": 0.9658,
+ "step": 28
+ },
+ {
+ "epoch": 0.875,
+ "eval_loss": 0.9378474354743958,
+ "eval_runtime": 49.9299,
+ "eval_samples_per_second": 4.006,
+ "eval_steps_per_second": 0.26,
+ "step": 28
+ },
+ {
+ "epoch": 0.90625,
+ "grad_norm": 0.18074632782127534,
+ "learning_rate": 2e-05,
+ "loss": 0.9866,
+ "step": 29
+ },
+ {
+ "epoch": 0.90625,
+ "eval_loss": 0.93099045753479,
+ "eval_runtime": 50.0096,
+ "eval_samples_per_second": 3.999,
+ "eval_steps_per_second": 0.26,
+ "step": 29
+ },
+ {
+ "epoch": 0.9375,
+ "grad_norm": 0.1936051126921734,
+ "learning_rate": 2e-05,
+ "loss": 1.0128,
+ "step": 30
+ },
+ {
+ "epoch": 0.9375,
+ "eval_loss": 0.9244199991226196,
+ "eval_runtime": 50.2469,
+ "eval_samples_per_second": 3.98,
+ "eval_steps_per_second": 0.259,
+ "step": 30
+ },
+ {
+ "epoch": 0.96875,
+ "grad_norm": 0.26164254459782943,
+ "learning_rate": 2e-05,
+ "loss": 0.88,
+ "step": 31
+ },
+ {
+ "epoch": 0.96875,
+ "eval_loss": 0.9175177216529846,
+ "eval_runtime": 50.1695,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 31
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.18677152741688485,
+ "learning_rate": 2e-05,
+ "loss": 0.9569,
+ "step": 32
+ },
+ {
+ "epoch": 1.0,
+ "eval_loss": 0.9108598828315735,
+ "eval_runtime": 50.0387,
+ "eval_samples_per_second": 3.997,
+ "eval_steps_per_second": 0.26,
+ "step": 32
+ },
+ {
+ "epoch": 1.03125,
+ "grad_norm": 0.20486279036126417,
+ "learning_rate": 2e-05,
+ "loss": 1.0208,
+ "step": 33
+ },
+ {
+ "epoch": 1.03125,
+ "eval_loss": 0.9042049646377563,
+ "eval_runtime": 50.1472,
+ "eval_samples_per_second": 3.988,
+ "eval_steps_per_second": 0.259,
+ "step": 33
+ },
+ {
+ "epoch": 1.0625,
+ "grad_norm": 0.2004946169291112,
+ "learning_rate": 2e-05,
+ "loss": 0.9931,
+ "step": 34
+ },
+ {
+ "epoch": 1.0625,
+ "eval_loss": 0.8980298042297363,
+ "eval_runtime": 50.245,
+ "eval_samples_per_second": 3.98,
+ "eval_steps_per_second": 0.259,
+ "step": 34
+ },
+ {
+ "epoch": 1.09375,
+ "grad_norm": 0.1645872432258401,
+ "learning_rate": 2e-05,
+ "loss": 1.0184,
+ "step": 35
+ },
+ {
+ "epoch": 1.09375,
+ "eval_loss": 0.8924428820610046,
+ "eval_runtime": 50.3703,
+ "eval_samples_per_second": 3.971,
+ "eval_steps_per_second": 0.258,
+ "step": 35
+ },
+ {
+ "epoch": 1.125,
+ "grad_norm": 0.18293519304435016,
+ "learning_rate": 2e-05,
+ "loss": 1.0026,
+ "step": 36
+ },
+ {
+ "epoch": 1.125,
+ "eval_loss": 0.8870412707328796,
+ "eval_runtime": 50.0483,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 36
+ },
+ {
+ "epoch": 1.15625,
+ "grad_norm": 0.17712548516246762,
+ "learning_rate": 2e-05,
+ "loss": 0.9387,
+ "step": 37
+ },
+ {
+ "epoch": 1.15625,
+ "eval_loss": 0.881915271282196,
+ "eval_runtime": 49.9751,
+ "eval_samples_per_second": 4.002,
+ "eval_steps_per_second": 0.26,
+ "step": 37
+ },
+ {
+ "epoch": 1.1875,
+ "grad_norm": 0.21472689311609464,
+ "learning_rate": 2e-05,
+ "loss": 0.958,
+ "step": 38
+ },
+ {
+ "epoch": 1.1875,
+ "eval_loss": 0.8768754601478577,
+ "eval_runtime": 50.1204,
+ "eval_samples_per_second": 3.99,
+ "eval_steps_per_second": 0.259,
+ "step": 38
+ },
+ {
+ "epoch": 1.21875,
+ "grad_norm": 0.21117297910005806,
+ "learning_rate": 2e-05,
+ "loss": 0.9922,
+ "step": 39
+ },
+ {
+ "epoch": 1.21875,
+ "eval_loss": 0.8718628883361816,
+ "eval_runtime": 50.1732,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 39
+ },
+ {
+ "epoch": 1.25,
+ "grad_norm": 0.17835587003909165,
+ "learning_rate": 2e-05,
+ "loss": 0.9776,
+ "step": 40
+ },
+ {
+ "epoch": 1.25,
+ "eval_loss": 0.8669865131378174,
+ "eval_runtime": 50.1148,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 40
+ },
+ {
+ "epoch": 1.28125,
+ "grad_norm": 0.2092736372483734,
+ "learning_rate": 2e-05,
+ "loss": 0.9731,
+ "step": 41
+ },
+ {
+ "epoch": 1.28125,
+ "eval_loss": 0.8619834780693054,
+ "eval_runtime": 50.052,
+ "eval_samples_per_second": 3.996,
+ "eval_steps_per_second": 0.26,
+ "step": 41
+ },
+ {
+ "epoch": 1.3125,
+ "grad_norm": 0.2338857391910308,
+ "learning_rate": 2e-05,
+ "loss": 0.9319,
+ "step": 42
+ },
+ {
+ "epoch": 1.3125,
+ "eval_loss": 0.8572126030921936,
+ "eval_runtime": 50.1212,
+ "eval_samples_per_second": 3.99,
+ "eval_steps_per_second": 0.259,
+ "step": 42
+ },
+ {
+ "epoch": 1.34375,
+ "grad_norm": 0.19168719284572813,
+ "learning_rate": 2e-05,
+ "loss": 0.9083,
+ "step": 43
+ },
+ {
+ "epoch": 1.34375,
+ "eval_loss": 0.8525611758232117,
+ "eval_runtime": 50.1733,
+ "eval_samples_per_second": 3.986,
+ "eval_steps_per_second": 0.259,
+ "step": 43
+ },
+ {
+ "epoch": 1.375,
+ "grad_norm": 0.20004868138433377,
+ "learning_rate": 2e-05,
+ "loss": 0.9118,
+ "step": 44
+ },
+ {
+ "epoch": 1.375,
+ "eval_loss": 0.8483461141586304,
+ "eval_runtime": 50.1083,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 44
+ },
+ {
+ "epoch": 1.40625,
+ "grad_norm": 0.19012965506122342,
+ "learning_rate": 2e-05,
+ "loss": 0.8888,
+ "step": 45
+ },
+ {
+ "epoch": 1.40625,
+ "eval_loss": 0.8446614742279053,
+ "eval_runtime": 50.1171,
+ "eval_samples_per_second": 3.991,
+ "eval_steps_per_second": 0.259,
+ "step": 45
+ },
+ {
+ "epoch": 1.4375,
+ "grad_norm": 0.21187005706805245,
+ "learning_rate": 2e-05,
+ "loss": 0.9319,
+ "step": 46
+ },
+ {
+ "epoch": 1.4375,
+ "eval_loss": 0.8412036299705505,
+ "eval_runtime": 50.0918,
+ "eval_samples_per_second": 3.993,
+ "eval_steps_per_second": 0.26,
+ "step": 46
+ },
+ {
+ "epoch": 1.46875,
+ "grad_norm": 0.19673832205926584,
+ "learning_rate": 2e-05,
+ "loss": 0.9359,
+ "step": 47
+ },
+ {
+ "epoch": 1.46875,
+ "eval_loss": 0.8380417823791504,
+ "eval_runtime": 50.2214,
+ "eval_samples_per_second": 3.982,
+ "eval_steps_per_second": 0.259,
+ "step": 47
+ },
+ {
+ "epoch": 1.5,
+ "grad_norm": 0.21712294106174318,
+ "learning_rate": 2e-05,
+ "loss": 0.8511,
+ "step": 48
+ },
+ {
+ "epoch": 1.5,
+ "eval_loss": 0.8353021740913391,
+ "eval_runtime": 50.1617,
+ "eval_samples_per_second": 3.987,
+ "eval_steps_per_second": 0.259,
+ "step": 48
+ },
+ {
+ "epoch": 1.53125,
+ "grad_norm": 0.2138924779700934,
+ "learning_rate": 2e-05,
+ "loss": 0.8695,
+ "step": 49
+ },
+ {
+ "epoch": 1.53125,
+ "eval_loss": 0.8327407836914062,
+ "eval_runtime": 50.1442,
+ "eval_samples_per_second": 3.988,
+ "eval_steps_per_second": 0.259,
+ "step": 49
+ },
+ {
+ "epoch": 1.5625,
+ "grad_norm": 0.22387442384578618,
+ "learning_rate": 2e-05,
+ "loss": 0.8518,
+ "step": 50
+ },
+ {
+ "epoch": 1.5625,
+ "eval_loss": 0.8301742076873779,
+ "eval_runtime": 50.1867,
+ "eval_samples_per_second": 3.985,
+ "eval_steps_per_second": 0.259,
+ "step": 50
+ },
+ {
+ "epoch": 1.59375,
+ "grad_norm": 0.1975577146517192,
+ "learning_rate": 2e-05,
+ "loss": 0.8868,
+ "step": 51
+ },
+ {
+ "epoch": 1.59375,
+ "eval_loss": 0.8275265693664551,
+ "eval_runtime": 51.2257,
+ "eval_samples_per_second": 3.904,
+ "eval_steps_per_second": 0.254,
+ "step": 51
+ },
+ {
+ "epoch": 1.625,
+ "grad_norm": 0.21474817057286624,
+ "learning_rate": 2e-05,
+ "loss": 0.767,
+ "step": 52
+ },
+ {
+ "epoch": 1.625,
+ "eval_loss": 0.824796736240387,
+ "eval_runtime": 51.276,
+ "eval_samples_per_second": 3.9,
+ "eval_steps_per_second": 0.254,
+ "step": 52
+ },
+ {
+ "epoch": 1.65625,
+ "grad_norm": 0.21105651676755652,
+ "learning_rate": 2e-05,
+ "loss": 0.9219,
+ "step": 53
+ },
+ {
+ "epoch": 1.65625,
+ "eval_loss": 0.8221166729927063,
+ "eval_runtime": 51.141,
+ "eval_samples_per_second": 3.911,
+ "eval_steps_per_second": 0.254,
+ "step": 53
+ },
+ {
+ "epoch": 1.6875,
+ "grad_norm": 0.20706475184742085,
+ "learning_rate": 2e-05,
+ "loss": 0.8873,
+ "step": 54
+ },
+ {
+ "epoch": 1.6875,
+ "eval_loss": 0.819589376449585,
+ "eval_runtime": 51.0045,
+ "eval_samples_per_second": 3.921,
+ "eval_steps_per_second": 0.255,
+ "step": 54
+ },
+ {
+ "epoch": 1.71875,
+ "grad_norm": 0.21722220033855957,
+ "learning_rate": 2e-05,
+ "loss": 0.8956,
+ "step": 55
+ },
+ {
+ "epoch": 1.71875,
+ "eval_loss": 0.8176340460777283,
+ "eval_runtime": 51.1941,
+ "eval_samples_per_second": 3.907,
+ "eval_steps_per_second": 0.254,
+ "step": 55
+ },
+ {
+ "epoch": 1.75,
+ "grad_norm": 0.20669001221665667,
+ "learning_rate": 2e-05,
+ "loss": 0.9506,
+ "step": 56
+ },
+ {
+ "epoch": 1.75,
+ "eval_loss": 0.8158826231956482,
+ "eval_runtime": 52.1162,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 56
+ },
+ {
+ "epoch": 1.78125,
+ "grad_norm": 0.22189732090066341,
+ "learning_rate": 2e-05,
+ "loss": 0.8955,
+ "step": 57
+ },
+ {
+ "epoch": 1.78125,
+ "eval_loss": 0.814656674861908,
+ "eval_runtime": 52.1361,
+ "eval_samples_per_second": 3.836,
+ "eval_steps_per_second": 0.249,
+ "step": 57
+ },
+ {
+ "epoch": 1.8125,
+ "grad_norm": 0.2030113892848459,
+ "learning_rate": 2e-05,
+ "loss": 0.9108,
+ "step": 58
+ },
+ {
+ "epoch": 1.8125,
+ "eval_loss": 0.813343346118927,
+ "eval_runtime": 52.2552,
+ "eval_samples_per_second": 3.827,
+ "eval_steps_per_second": 0.249,
+ "step": 58
+ },
+ {
+ "epoch": 1.84375,
+ "grad_norm": 0.2123201057569791,
+ "learning_rate": 2e-05,
+ "loss": 0.8779,
+ "step": 59
+ },
+ {
+ "epoch": 1.84375,
+ "eval_loss": 0.8116877675056458,
+ "eval_runtime": 52.1233,
+ "eval_samples_per_second": 3.837,
+ "eval_steps_per_second": 0.249,
+ "step": 59
+ },
+ {
+ "epoch": 1.875,
+ "grad_norm": 0.211551126937912,
+ "learning_rate": 2e-05,
+ "loss": 0.9294,
+ "step": 60
+ },
+ {
+ "epoch": 1.875,
+ "eval_loss": 0.8098442554473877,
+ "eval_runtime": 52.1091,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 60
+ },
+ {
+ "epoch": 1.90625,
+ "grad_norm": 0.24981344981629752,
+ "learning_rate": 2e-05,
+ "loss": 0.8409,
+ "step": 61
+ },
+ {
+ "epoch": 1.90625,
+ "eval_loss": 0.8070770502090454,
+ "eval_runtime": 53.4187,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 61
+ },
+ {
+ "epoch": 1.9375,
+ "grad_norm": 0.2341550589775159,
+ "learning_rate": 2e-05,
+ "loss": 0.888,
+ "step": 62
+ },
+ {
+ "epoch": 1.9375,
+ "eval_loss": 0.8040286898612976,
+ "eval_runtime": 53.2197,
+ "eval_samples_per_second": 3.758,
+ "eval_steps_per_second": 0.244,
+ "step": 62
+ },
+ {
+ "epoch": 1.96875,
+ "grad_norm": 0.2336241775649256,
+ "learning_rate": 2e-05,
+ "loss": 0.913,
+ "step": 63
+ },
+ {
+ "epoch": 1.96875,
+ "eval_loss": 0.8013430833816528,
+ "eval_runtime": 53.1784,
+ "eval_samples_per_second": 3.761,
+ "eval_steps_per_second": 0.244,
+ "step": 63
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.2414390628081758,
+ "learning_rate": 2e-05,
+ "loss": 0.8754,
+ "step": 64
+ },
+ {
+ "epoch": 2.0,
+ "eval_loss": 0.7985894680023193,
+ "eval_runtime": 53.2454,
+ "eval_samples_per_second": 3.756,
+ "eval_steps_per_second": 0.244,
+ "step": 64
+ },
+ {
+ "epoch": 2.03125,
+ "grad_norm": 0.2484104465653703,
+ "learning_rate": 2e-05,
+ "loss": 0.8497,
+ "step": 65
+ },
+ {
+ "epoch": 2.03125,
+ "eval_loss": 0.7954932451248169,
+ "eval_runtime": 53.3794,
+ "eval_samples_per_second": 3.747,
+ "eval_steps_per_second": 0.244,
+ "step": 65
+ },
+ {
+ "epoch": 2.0625,
+ "grad_norm": 0.23859744120942086,
+ "learning_rate": 2e-05,
+ "loss": 0.8567,
+ "step": 66
+ },
+ {
+ "epoch": 2.0625,
+ "eval_loss": 0.7929843068122864,
+ "eval_runtime": 55.517,
+ "eval_samples_per_second": 3.602,
+ "eval_steps_per_second": 0.234,
+ "step": 66
+ },
+ {
+ "epoch": 2.09375,
+ "grad_norm": 0.24584758647855462,
+ "learning_rate": 2e-05,
+ "loss": 0.8489,
+ "step": 67
+ },
+ {
+ "epoch": 2.09375,
+ "eval_loss": 0.7903321981430054,
+ "eval_runtime": 55.4151,
+ "eval_samples_per_second": 3.609,
+ "eval_steps_per_second": 0.235,
+ "step": 67
+ },
+ {
+ "epoch": 2.125,
+ "grad_norm": 0.2484917818304153,
+ "learning_rate": 2e-05,
+ "loss": 0.9122,
+ "step": 68
+ },
+ {
+ "epoch": 2.125,
+ "eval_loss": 0.7877185344696045,
+ "eval_runtime": 55.4069,
+ "eval_samples_per_second": 3.61,
+ "eval_steps_per_second": 0.235,
+ "step": 68
+ },
+ {
+ "epoch": 2.15625,
+ "grad_norm": 0.2184614083026819,
+ "learning_rate": 2e-05,
+ "loss": 0.8355,
+ "step": 69
+ },
+ {
+ "epoch": 2.15625,
+ "eval_loss": 0.7852210998535156,
+ "eval_runtime": 55.3381,
+ "eval_samples_per_second": 3.614,
+ "eval_steps_per_second": 0.235,
+ "step": 69
+ },
+ {
+ "epoch": 2.1875,
+ "grad_norm": 0.24978410070800153,
+ "learning_rate": 2e-05,
+ "loss": 0.7968,
+ "step": 70
+ },
+ {
+ "epoch": 2.1875,
+ "eval_loss": 0.7827157378196716,
+ "eval_runtime": 55.3708,
+ "eval_samples_per_second": 3.612,
+ "eval_steps_per_second": 0.235,
+ "step": 70
+ },
+ {
+ "epoch": 2.21875,
+ "grad_norm": 0.23059883325890385,
+ "learning_rate": 2e-05,
+ "loss": 0.8783,
+ "step": 71
+ },
+ {
+ "epoch": 2.21875,
+ "eval_loss": 0.7805906534194946,
+ "eval_runtime": 55.6033,
+ "eval_samples_per_second": 3.597,
+ "eval_steps_per_second": 0.234,
+ "step": 71
+ },
+ {
+ "epoch": 2.25,
+ "grad_norm": 0.23261007334915096,
+ "learning_rate": 2e-05,
+ "loss": 0.7956,
+ "step": 72
+ },
+ {
+ "epoch": 2.25,
+ "eval_loss": 0.7786691784858704,
+ "eval_runtime": 55.0913,
+ "eval_samples_per_second": 3.63,
+ "eval_steps_per_second": 0.236,
+ "step": 72
+ },
+ {
+ "epoch": 2.28125,
+ "grad_norm": 0.25779598356574085,
+ "learning_rate": 2e-05,
+ "loss": 0.8426,
+ "step": 73
+ },
+ {
+ "epoch": 2.28125,
+ "eval_loss": 0.7771151661872864,
+ "eval_runtime": 55.0698,
+ "eval_samples_per_second": 3.632,
+ "eval_steps_per_second": 0.236,
+ "step": 73
+ },
+ {
+ "epoch": 2.3125,
+ "grad_norm": 0.2288243335971112,
+ "learning_rate": 2e-05,
+ "loss": 0.8381,
+ "step": 74
+ },
+ {
+ "epoch": 2.3125,
+ "eval_loss": 0.7756838202476501,
+ "eval_runtime": 54.8412,
+ "eval_samples_per_second": 3.647,
+ "eval_steps_per_second": 0.237,
+ "step": 74
+ },
+ {
+ "epoch": 2.34375,
+ "grad_norm": 0.24235644907977733,
+ "learning_rate": 2e-05,
+ "loss": 0.887,
+ "step": 75
+ },
+ {
+ "epoch": 2.34375,
+ "eval_loss": 0.7739972472190857,
+ "eval_runtime": 54.9718,
+ "eval_samples_per_second": 3.638,
+ "eval_steps_per_second": 0.236,
+ "step": 75
+ },
+ {
+ "epoch": 2.375,
+ "grad_norm": 0.23666820017867402,
+ "learning_rate": 2e-05,
+ "loss": 0.8007,
+ "step": 76
+ },
+ {
+ "epoch": 2.375,
+ "eval_loss": 0.7724328637123108,
+ "eval_runtime": 55.0225,
+ "eval_samples_per_second": 3.635,
+ "eval_steps_per_second": 0.236,
+ "step": 76
+ },
+ {
+ "epoch": 2.40625,
+ "grad_norm": 0.22815737396609181,
+ "learning_rate": 2e-05,
+ "loss": 0.8529,
+ "step": 77
+ },
+ {
+ "epoch": 2.40625,
+ "eval_loss": 0.7710004448890686,
+ "eval_runtime": 55.321,
+ "eval_samples_per_second": 3.615,
+ "eval_steps_per_second": 0.235,
+ "step": 77
+ },
+ {
+ "epoch": 2.4375,
+ "grad_norm": 0.2701264871470739,
+ "learning_rate": 2e-05,
+ "loss": 0.8515,
+ "step": 78
+ },
+ {
+ "epoch": 2.4375,
+ "eval_loss": 0.7695322632789612,
+ "eval_runtime": 55.3045,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 78
+ },
+ {
+ "epoch": 2.46875,
+ "grad_norm": 0.24363813951328234,
+ "learning_rate": 2e-05,
+ "loss": 0.8587,
+ "step": 79
+ },
+ {
+ "epoch": 2.46875,
+ "eval_loss": 0.7689024209976196,
+ "eval_runtime": 55.3009,
+ "eval_samples_per_second": 3.617,
+ "eval_steps_per_second": 0.235,
+ "step": 79
+ },
+ {
+ "epoch": 2.5,
+ "grad_norm": 0.30924701355253065,
+ "learning_rate": 2e-05,
+ "loss": 0.9076,
+ "step": 80
+ },
+ {
+ "epoch": 2.5,
+ "eval_loss": 0.7676254510879517,
+ "eval_runtime": 55.2365,
+ "eval_samples_per_second": 3.621,
+ "eval_steps_per_second": 0.235,
+ "step": 80
+ },
+ {
+ "epoch": 2.53125,
+ "grad_norm": 0.2665188280221636,
+ "learning_rate": 2e-05,
+ "loss": 0.8445,
+ "step": 81
+ },
+ {
+ "epoch": 2.53125,
+ "eval_loss": 0.7661146521568298,
+ "eval_runtime": 55.2775,
+ "eval_samples_per_second": 3.618,
+ "eval_steps_per_second": 0.235,
+ "step": 81
+ },
+ {
+ "epoch": 2.5625,
+ "grad_norm": 0.24674191720675534,
+ "learning_rate": 2e-05,
+ "loss": 0.8882,
+ "step": 82
+ },
+ {
+ "epoch": 2.5625,
+ "eval_loss": 0.76513671875,
+ "eval_runtime": 55.0857,
+ "eval_samples_per_second": 3.631,
+ "eval_steps_per_second": 0.236,
+ "step": 82
+ },
+ {
+ "epoch": 2.59375,
+ "grad_norm": 0.2736689405531704,
+ "learning_rate": 2e-05,
+ "loss": 0.8336,
+ "step": 83
+ },
+ {
+ "epoch": 2.59375,
+ "eval_loss": 0.764373779296875,
+ "eval_runtime": 55.2069,
+ "eval_samples_per_second": 3.623,
+ "eval_steps_per_second": 0.235,
+ "step": 83
+ },
+ {
+ "epoch": 2.625,
+ "grad_norm": 0.290841287198557,
+ "learning_rate": 2e-05,
+ "loss": 0.795,
+ "step": 84
+ },
+ {
+ "epoch": 2.625,
+ "eval_loss": 0.7632084488868713,
+ "eval_runtime": 55.1009,
+ "eval_samples_per_second": 3.63,
+ "eval_steps_per_second": 0.236,
+ "step": 84
+ },
+ {
+ "epoch": 2.65625,
+ "grad_norm": 0.2912051076836381,
+ "learning_rate": 2e-05,
+ "loss": 0.772,
+ "step": 85
+ },
+ {
+ "epoch": 2.65625,
+ "eval_loss": 0.7618446350097656,
+ "eval_runtime": 55.3717,
+ "eval_samples_per_second": 3.612,
+ "eval_steps_per_second": 0.235,
+ "step": 85
+ },
+ {
+ "epoch": 2.6875,
+ "grad_norm": 0.3169908538809109,
+ "learning_rate": 2e-05,
+ "loss": 0.8148,
+ "step": 86
+ },
+ {
+ "epoch": 2.6875,
+ "eval_loss": 0.7599577307701111,
+ "eval_runtime": 55.3931,
+ "eval_samples_per_second": 3.611,
+ "eval_steps_per_second": 0.235,
+ "step": 86
+ },
+ {
+ "epoch": 2.71875,
+ "grad_norm": 0.28780549186847426,
+ "learning_rate": 2e-05,
+ "loss": 0.8154,
+ "step": 87
+ },
+ {
+ "epoch": 2.71875,
+ "eval_loss": 0.7583369612693787,
+ "eval_runtime": 55.1679,
+ "eval_samples_per_second": 3.625,
+ "eval_steps_per_second": 0.236,
+ "step": 87
+ },
+ {
+ "epoch": 2.75,
+ "grad_norm": 0.30695250620091474,
+ "learning_rate": 2e-05,
+ "loss": 0.9032,
+ "step": 88
+ },
+ {
+ "epoch": 2.75,
+ "eval_loss": 0.7571613192558289,
+ "eval_runtime": 55.1779,
+ "eval_samples_per_second": 3.625,
+ "eval_steps_per_second": 0.236,
+ "step": 88
+ },
+ {
+ "epoch": 2.78125,
+ "grad_norm": 0.2693887416759828,
+ "learning_rate": 2e-05,
+ "loss": 0.8106,
+ "step": 89
+ },
+ {
+ "epoch": 2.78125,
+ "eval_loss": 0.7566004991531372,
+ "eval_runtime": 55.1107,
+ "eval_samples_per_second": 3.629,
+ "eval_steps_per_second": 0.236,
+ "step": 89
+ },
+ {
+ "epoch": 2.8125,
+ "grad_norm": 0.2887583627563198,
+ "learning_rate": 2e-05,
+ "loss": 0.8518,
+ "step": 90
+ },
+ {
+ "epoch": 2.8125,
+ "eval_loss": 0.7558963298797607,
+ "eval_runtime": 55.2153,
+ "eval_samples_per_second": 3.622,
+ "eval_steps_per_second": 0.235,
+ "step": 90
+ },
+ {
+ "epoch": 2.84375,
+ "grad_norm": 0.3059402168979351,
+ "learning_rate": 2e-05,
+ "loss": 0.7727,
+ "step": 91
+ },
+ {
+ "epoch": 2.84375,
+ "eval_loss": 0.7545350790023804,
+ "eval_runtime": 55.3225,
+ "eval_samples_per_second": 3.615,
+ "eval_steps_per_second": 0.235,
+ "step": 91
+ },
+ {
+ "epoch": 2.875,
+ "grad_norm": 0.3096260477909968,
+ "learning_rate": 2e-05,
+ "loss": 0.8477,
+ "step": 92
+ },
+ {
+ "epoch": 2.875,
+ "eval_loss": 0.7526452541351318,
+ "eval_runtime": 55.4311,
+ "eval_samples_per_second": 3.608,
+ "eval_steps_per_second": 0.235,
+ "step": 92
+ },
+ {
+ "epoch": 2.90625,
+ "grad_norm": 0.31498884686525297,
+ "learning_rate": 2e-05,
+ "loss": 0.7982,
+ "step": 93
+ },
+ {
+ "epoch": 2.90625,
+ "eval_loss": 0.7510760426521301,
+ "eval_runtime": 55.4361,
+ "eval_samples_per_second": 3.608,
+ "eval_steps_per_second": 0.235,
+ "step": 93
+ },
+ {
+ "epoch": 2.9375,
+ "grad_norm": 0.31302830623184313,
+ "learning_rate": 2e-05,
+ "loss": 0.871,
+ "step": 94
+ },
+ {
+ "epoch": 2.9375,
+ "eval_loss": 0.7500898838043213,
+ "eval_runtime": 55.3025,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 94
+ },
+ {
+ "epoch": 2.96875,
+ "grad_norm": 0.3132608568779145,
+ "learning_rate": 2e-05,
+ "loss": 0.8094,
+ "step": 95
+ },
+ {
+ "epoch": 2.96875,
+ "eval_loss": 0.7498895525932312,
+ "eval_runtime": 55.2402,
+ "eval_samples_per_second": 3.621,
+ "eval_steps_per_second": 0.235,
+ "step": 95
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 0.298645350091386,
+ "learning_rate": 2e-05,
+ "loss": 0.7673,
+ "step": 96
+ },
+ {
+ "epoch": 3.0,
+ "eval_loss": 0.7493192553520203,
+ "eval_runtime": 54.8718,
+ "eval_samples_per_second": 3.645,
+ "eval_steps_per_second": 0.237,
+ "step": 96
+ },
+ {
+ "epoch": 3.03125,
+ "grad_norm": 0.34042584783125357,
+ "learning_rate": 2e-05,
+ "loss": 0.7336,
+ "step": 97
+ },
+ {
+ "epoch": 3.03125,
+ "eval_loss": 0.7476670742034912,
+ "eval_runtime": 54.9305,
+ "eval_samples_per_second": 3.641,
+ "eval_steps_per_second": 0.237,
+ "step": 97
+ },
+ {
+ "epoch": 3.0625,
+ "grad_norm": 0.293099043801068,
+ "learning_rate": 2e-05,
+ "loss": 0.8088,
+ "step": 98
+ },
+ {
+ "epoch": 3.0625,
+ "eval_loss": 0.745802640914917,
+ "eval_runtime": 55.2051,
+ "eval_samples_per_second": 3.623,
+ "eval_steps_per_second": 0.235,
+ "step": 98
+ },
+ {
+ "epoch": 3.09375,
+ "grad_norm": 0.3042839507858426,
+ "learning_rate": 2e-05,
+ "loss": 0.787,
+ "step": 99
+ },
+ {
+ "epoch": 3.09375,
+ "eval_loss": 0.7439618110656738,
+ "eval_runtime": 55.0065,
+ "eval_samples_per_second": 3.636,
+ "eval_steps_per_second": 0.236,
+ "step": 99
+ },
+ {
+ "epoch": 3.125,
+ "grad_norm": 0.32992077073227005,
+ "learning_rate": 2e-05,
+ "loss": 0.8296,
+ "step": 100
+ },
+ {
+ "epoch": 3.125,
+ "eval_loss": 0.7424842715263367,
+ "eval_runtime": 55.1254,
+ "eval_samples_per_second": 3.628,
+ "eval_steps_per_second": 0.236,
+ "step": 100
+ },
+ {
+ "epoch": 3.15625,
+ "grad_norm": 0.2798839747424062,
+ "learning_rate": 2e-05,
+ "loss": 0.7642,
+ "step": 101
+ },
+ {
+ "epoch": 3.15625,
+ "eval_loss": 0.7414796948432922,
+ "eval_runtime": 49.183,
+ "eval_samples_per_second": 4.066,
+ "eval_steps_per_second": 0.264,
+ "step": 101
+ },
+ {
+ "epoch": 3.1875,
+ "grad_norm": 0.3046631191964983,
+ "learning_rate": 2e-05,
+ "loss": 0.8203,
+ "step": 102
+ },
+ {
+ "epoch": 3.1875,
+ "eval_loss": 0.7410265207290649,
+ "eval_runtime": 48.1541,
+ "eval_samples_per_second": 4.153,
+ "eval_steps_per_second": 0.27,
+ "step": 102
+ },
+ {
+ "epoch": 3.21875,
+ "grad_norm": 0.3117517214859861,
+ "learning_rate": 2e-05,
+ "loss": 0.8222,
+ "step": 103
+ },
+ {
+ "epoch": 3.21875,
+ "eval_loss": 0.7405675649642944,
+ "eval_runtime": 47.7145,
+ "eval_samples_per_second": 4.192,
+ "eval_steps_per_second": 0.272,
+ "step": 103
+ },
+ {
+ "epoch": 3.25,
+ "grad_norm": 0.3412709249466801,
+ "learning_rate": 2e-05,
+ "loss": 0.7459,
+ "step": 104
+ },
+ {
+ "epoch": 3.25,
+ "eval_loss": 0.7395681738853455,
+ "eval_runtime": 47.5855,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 104
+ },
+ {
+ "epoch": 3.28125,
+ "grad_norm": 0.2917443566507923,
+ "learning_rate": 2e-05,
+ "loss": 0.7849,
+ "step": 105
+ },
+ {
+ "epoch": 3.28125,
+ "eval_loss": 0.7387100458145142,
+ "eval_runtime": 47.6344,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 105
+ },
+ {
+ "epoch": 3.3125,
+ "grad_norm": 0.3054484743574741,
+ "learning_rate": 2e-05,
+ "loss": 0.8354,
+ "step": 106
+ },
+ {
+ "epoch": 3.3125,
+ "eval_loss": 0.7384718060493469,
+ "eval_runtime": 47.8373,
+ "eval_samples_per_second": 4.181,
+ "eval_steps_per_second": 0.272,
+ "step": 106
+ },
+ {
+ "epoch": 3.34375,
+ "grad_norm": 0.34986630381114014,
+ "learning_rate": 2e-05,
+ "loss": 0.7069,
+ "step": 107
+ },
+ {
+ "epoch": 3.34375,
+ "eval_loss": 0.737342357635498,
+ "eval_runtime": 47.5763,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 107
+ },
+ {
+ "epoch": 3.375,
+ "grad_norm": 0.32324403145716496,
+ "learning_rate": 2e-05,
+ "loss": 0.767,
+ "step": 108
+ },
+ {
+ "epoch": 3.375,
+ "eval_loss": 0.7360101938247681,
+ "eval_runtime": 47.5774,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 108
+ },
+ {
+ "epoch": 3.40625,
+ "grad_norm": 0.3795969851258545,
+ "learning_rate": 2e-05,
+ "loss": 0.7556,
+ "step": 109
+ },
+ {
+ "epoch": 3.40625,
+ "eval_loss": 0.7339167594909668,
+ "eval_runtime": 47.5818,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 109
+ },
+ {
+ "epoch": 3.4375,
+ "grad_norm": 0.34401062275458993,
+ "learning_rate": 2e-05,
+ "loss": 0.7494,
+ "step": 110
+ },
+ {
+ "epoch": 3.4375,
+ "eval_loss": 0.7321068644523621,
+ "eval_runtime": 47.7643,
+ "eval_samples_per_second": 4.187,
+ "eval_steps_per_second": 0.272,
+ "step": 110
+ },
+ {
+ "epoch": 3.46875,
+ "grad_norm": 0.3248480010385237,
+ "learning_rate": 2e-05,
+ "loss": 0.8103,
+ "step": 111
+ },
+ {
+ "epoch": 3.46875,
+ "eval_loss": 0.7309197783470154,
+ "eval_runtime": 49.5841,
+ "eval_samples_per_second": 4.034,
+ "eval_steps_per_second": 0.262,
+ "step": 111
+ },
+ {
+ "epoch": 3.5,
+ "grad_norm": 0.3572409124813593,
+ "learning_rate": 2e-05,
+ "loss": 0.7972,
+ "step": 112
+ },
+ {
+ "epoch": 3.5,
+ "eval_loss": 0.7301727533340454,
+ "eval_runtime": 49.3728,
+ "eval_samples_per_second": 4.051,
+ "eval_steps_per_second": 0.263,
+ "step": 112
+ },
+ {
+ "epoch": 3.53125,
+ "grad_norm": 0.37348522775103665,
+ "learning_rate": 2e-05,
+ "loss": 0.88,
+ "step": 113
+ },
+ {
+ "epoch": 3.53125,
+ "eval_loss": 0.7292957305908203,
+ "eval_runtime": 49.2192,
+ "eval_samples_per_second": 4.063,
+ "eval_steps_per_second": 0.264,
+ "step": 113
+ },
+ {
+ "epoch": 3.5625,
+ "grad_norm": 0.37667450960329546,
+ "learning_rate": 2e-05,
+ "loss": 0.7518,
+ "step": 114
+ },
+ {
+ "epoch": 3.5625,
+ "eval_loss": 0.728556215763092,
+ "eval_runtime": 49.0971,
+ "eval_samples_per_second": 4.074,
+ "eval_steps_per_second": 0.265,
+ "step": 114
+ },
+ {
+ "epoch": 3.59375,
+ "grad_norm": 0.3163628607304638,
+ "learning_rate": 2e-05,
+ "loss": 0.7948,
+ "step": 115
+ },
+ {
+ "epoch": 3.59375,
+ "eval_loss": 0.7287828326225281,
+ "eval_runtime": 49.0213,
+ "eval_samples_per_second": 4.08,
+ "eval_steps_per_second": 0.265,
+ "step": 115
+ },
+ {
+ "epoch": 3.625,
+ "grad_norm": 0.3038899302084592,
+ "learning_rate": 2e-05,
+ "loss": 0.7791,
+ "step": 116
+ },
+ {
+ "epoch": 3.625,
+ "eval_loss": 0.7294514179229736,
+ "eval_runtime": 51.9137,
+ "eval_samples_per_second": 3.853,
+ "eval_steps_per_second": 0.25,
+ "step": 116
+ },
+ {
+ "epoch": 3.65625,
+ "grad_norm": 0.3746448663122327,
+ "learning_rate": 2e-05,
+ "loss": 0.7863,
+ "step": 117
+ },
+ {
+ "epoch": 3.65625,
+ "eval_loss": 0.7289304137229919,
+ "eval_runtime": 51.3023,
+ "eval_samples_per_second": 3.898,
+ "eval_steps_per_second": 0.253,
+ "step": 117
+ },
+ {
+ "epoch": 3.6875,
+ "grad_norm": 0.4058937381299434,
+ "learning_rate": 2e-05,
+ "loss": 0.7907,
+ "step": 118
+ },
+ {
+ "epoch": 3.6875,
+ "eval_loss": 0.7281011343002319,
+ "eval_runtime": 50.8635,
+ "eval_samples_per_second": 3.932,
+ "eval_steps_per_second": 0.256,
+ "step": 118
+ },
+ {
+ "epoch": 3.71875,
+ "grad_norm": 0.31608065583227885,
+ "learning_rate": 2e-05,
+ "loss": 0.8348,
+ "step": 119
+ },
+ {
+ "epoch": 3.71875,
+ "eval_loss": 0.7280247211456299,
+ "eval_runtime": 50.4903,
+ "eval_samples_per_second": 3.961,
+ "eval_steps_per_second": 0.257,
+ "step": 119
+ },
+ {
+ "epoch": 3.75,
+ "grad_norm": 0.3375768031046084,
+ "learning_rate": 2e-05,
+ "loss": 0.7783,
+ "step": 120
+ },
+ {
+ "epoch": 3.75,
+ "eval_loss": 0.7281913757324219,
+ "eval_runtime": 50.5906,
+ "eval_samples_per_second": 3.953,
+ "eval_steps_per_second": 0.257,
+ "step": 120
+ },
+ {
+ "epoch": 3.78125,
+ "grad_norm": 0.36047493494859845,
+ "learning_rate": 2e-05,
+ "loss": 0.765,
+ "step": 121
+ },
+ {
+ "epoch": 3.78125,
+ "eval_loss": 0.7269737124443054,
+ "eval_runtime": 53.4722,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 121
+ },
+ {
+ "epoch": 3.8125,
+ "grad_norm": 0.389743860171921,
+ "learning_rate": 2e-05,
+ "loss": 0.8269,
+ "step": 122
+ },
+ {
+ "epoch": 3.8125,
+ "eval_loss": 0.7251996397972107,
+ "eval_runtime": 53.4986,
+ "eval_samples_per_second": 3.738,
+ "eval_steps_per_second": 0.243,
+ "step": 122
+ },
+ {
+ "epoch": 3.84375,
+ "grad_norm": 0.33850935145960215,
+ "learning_rate": 2e-05,
+ "loss": 0.7497,
+ "step": 123
+ },
+ {
+ "epoch": 3.84375,
+ "eval_loss": 0.723595142364502,
+ "eval_runtime": 53.4196,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 123
+ },
+ {
+ "epoch": 3.875,
+ "grad_norm": 0.3166770012114478,
+ "learning_rate": 2e-05,
+ "loss": 0.7648,
+ "step": 124
+ },
+ {
+ "epoch": 3.875,
+ "eval_loss": 0.7223578095436096,
+ "eval_runtime": 52.6143,
+ "eval_samples_per_second": 3.801,
+ "eval_steps_per_second": 0.247,
+ "step": 124
+ },
+ {
+ "epoch": 3.90625,
+ "grad_norm": 0.41948670305268276,
+ "learning_rate": 2e-05,
+ "loss": 0.8306,
+ "step": 125
+ },
+ {
+ "epoch": 3.90625,
+ "eval_loss": 0.7206680774688721,
+ "eval_runtime": 52.3885,
+ "eval_samples_per_second": 3.818,
+ "eval_steps_per_second": 0.248,
+ "step": 125
+ },
+ {
+ "epoch": 3.9375,
+ "grad_norm": 0.35580041105853477,
+ "learning_rate": 2e-05,
+ "loss": 0.7945,
+ "step": 126
+ },
+ {
+ "epoch": 3.9375,
+ "eval_loss": 0.7196171283721924,
+ "eval_runtime": 55.1225,
+ "eval_samples_per_second": 3.628,
+ "eval_steps_per_second": 0.236,
+ "step": 126
+ },
+ {
+ "epoch": 3.96875,
+ "grad_norm": 0.38411890663257114,
+ "learning_rate": 2e-05,
+ "loss": 0.7466,
+ "step": 127
+ },
+ {
+ "epoch": 3.96875,
+ "eval_loss": 0.7188088297843933,
+ "eval_runtime": 55.3068,
+ "eval_samples_per_second": 3.616,
+ "eval_steps_per_second": 0.235,
+ "step": 127
+ },
+ {
+ "epoch": 4.0,
+ "grad_norm": 0.3682220575203032,
+ "learning_rate": 2e-05,
+ "loss": 0.6752,
+ "step": 128
+ },
+ {
+ "epoch": 4.0,
+ "eval_loss": 0.7181470990180969,
+ "eval_runtime": 53.9116,
+ "eval_samples_per_second": 3.71,
+ "eval_steps_per_second": 0.241,
+ "step": 128
+ },
+ {
+ "epoch": 4.03125,
+ "grad_norm": 0.34160763542661665,
+ "learning_rate": 2e-05,
+ "loss": 0.7788,
+ "step": 129
+ },
+ {
+ "epoch": 4.03125,
+ "eval_loss": 0.717949390411377,
+ "eval_runtime": 53.8446,
+ "eval_samples_per_second": 3.714,
+ "eval_steps_per_second": 0.241,
+ "step": 129
+ },
+ {
+ "epoch": 4.0625,
+ "grad_norm": 0.35709301353799944,
+ "learning_rate": 2e-05,
+ "loss": 0.8002,
+ "step": 130
+ },
+ {
+ "epoch": 4.0625,
+ "eval_loss": 0.7179380655288696,
+ "eval_runtime": 53.9299,
+ "eval_samples_per_second": 3.709,
+ "eval_steps_per_second": 0.241,
+ "step": 130
+ },
+ {
+ "epoch": 4.09375,
+ "grad_norm": 0.3503147340749238,
+ "learning_rate": 2e-05,
+ "loss": 0.7789,
+ "step": 131
+ },
+ {
+ "epoch": 4.09375,
+ "eval_loss": 0.7180312871932983,
+ "eval_runtime": 53.4091,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 131
+ },
+ {
+ "epoch": 4.125,
+ "grad_norm": 0.3931715546229069,
+ "learning_rate": 2e-05,
+ "loss": 0.762,
+ "step": 132
+ },
+ {
+ "epoch": 4.125,
+ "eval_loss": 0.717825710773468,
+ "eval_runtime": 53.6366,
+ "eval_samples_per_second": 3.729,
+ "eval_steps_per_second": 0.242,
+ "step": 132
+ },
+ {
+ "epoch": 4.15625,
+ "grad_norm": 0.36864033862644363,
+ "learning_rate": 2e-05,
+ "loss": 0.829,
+ "step": 133
+ },
+ {
+ "epoch": 4.15625,
+ "eval_loss": 0.7178698182106018,
+ "eval_runtime": 53.4891,
+ "eval_samples_per_second": 3.739,
+ "eval_steps_per_second": 0.243,
+ "step": 133
+ },
+ {
+ "epoch": 4.1875,
+ "grad_norm": 0.41393587587462155,
+ "learning_rate": 2e-05,
+ "loss": 0.7624,
+ "step": 134
+ },
+ {
+ "epoch": 4.1875,
+ "eval_loss": 0.7181968092918396,
+ "eval_runtime": 53.5395,
+ "eval_samples_per_second": 3.736,
+ "eval_steps_per_second": 0.243,
+ "step": 134
+ },
+ {
+ "epoch": 4.21875,
+ "grad_norm": 0.36727603900023204,
+ "learning_rate": 2e-05,
+ "loss": 0.7572,
+ "step": 135
+ },
+ {
+ "epoch": 4.21875,
+ "eval_loss": 0.7187527418136597,
+ "eval_runtime": 53.4818,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 135
+ },
+ {
+ "epoch": 4.25,
+ "grad_norm": 0.3684078795455007,
+ "learning_rate": 2e-05,
+ "loss": 0.7352,
+ "step": 136
+ },
+ {
+ "epoch": 4.25,
+ "eval_loss": 0.7194793820381165,
+ "eval_runtime": 53.4694,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 136
+ },
+ {
+ "epoch": 4.28125,
+ "grad_norm": 0.42414766562621153,
+ "learning_rate": 2e-05,
+ "loss": 0.7433,
+ "step": 137
+ },
+ {
+ "epoch": 4.28125,
+ "eval_loss": 0.7189603447914124,
+ "eval_runtime": 53.8049,
+ "eval_samples_per_second": 3.717,
+ "eval_steps_per_second": 0.242,
+ "step": 137
+ },
+ {
+ "epoch": 4.3125,
+ "grad_norm": 0.40420796619211563,
+ "learning_rate": 2e-05,
+ "loss": 0.7466,
+ "step": 138
+ },
+ {
+ "epoch": 4.3125,
+ "eval_loss": 0.7173956036567688,
+ "eval_runtime": 53.4014,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 138
+ },
+ {
+ "epoch": 4.34375,
+ "grad_norm": 0.36419740641344456,
+ "learning_rate": 2e-05,
+ "loss": 0.7045,
+ "step": 139
+ },
+ {
+ "epoch": 4.34375,
+ "eval_loss": 0.7153105139732361,
+ "eval_runtime": 53.285,
+ "eval_samples_per_second": 3.753,
+ "eval_steps_per_second": 0.244,
+ "step": 139
+ },
+ {
+ "epoch": 4.375,
+ "grad_norm": 0.384927357409491,
+ "learning_rate": 2e-05,
+ "loss": 0.7437,
+ "step": 140
+ },
+ {
+ "epoch": 4.375,
+ "eval_loss": 0.7135314345359802,
+ "eval_runtime": 53.4056,
+ "eval_samples_per_second": 3.745,
+ "eval_steps_per_second": 0.243,
+ "step": 140
+ },
+ {
+ "epoch": 4.40625,
+ "grad_norm": 0.37218579680263697,
+ "learning_rate": 2e-05,
+ "loss": 0.7693,
+ "step": 141
+ },
+ {
+ "epoch": 4.40625,
+ "eval_loss": 0.7120725512504578,
+ "eval_runtime": 53.5467,
+ "eval_samples_per_second": 3.735,
+ "eval_steps_per_second": 0.243,
+ "step": 141
+ },
+ {
+ "epoch": 4.4375,
+ "grad_norm": 0.38541382926033946,
+ "learning_rate": 2e-05,
+ "loss": 0.708,
+ "step": 142
+ },
+ {
+ "epoch": 4.4375,
+ "eval_loss": 0.7110380530357361,
+ "eval_runtime": 53.4119,
+ "eval_samples_per_second": 3.744,
+ "eval_steps_per_second": 0.243,
+ "step": 142
+ },
+ {
+ "epoch": 4.46875,
+ "grad_norm": 0.4028726453247759,
+ "learning_rate": 2e-05,
+ "loss": 0.7263,
+ "step": 143
+ },
+ {
+ "epoch": 4.46875,
+ "eval_loss": 0.7100683450698853,
+ "eval_runtime": 53.4337,
+ "eval_samples_per_second": 3.743,
+ "eval_steps_per_second": 0.243,
+ "step": 143
+ },
+ {
+ "epoch": 4.5,
+ "grad_norm": 0.3736204162232246,
+ "learning_rate": 2e-05,
+ "loss": 0.698,
+ "step": 144
+ },
+ {
+ "epoch": 4.5,
+ "eval_loss": 0.7093971371650696,
+ "eval_runtime": 53.4582,
+ "eval_samples_per_second": 3.741,
+ "eval_steps_per_second": 0.243,
+ "step": 144
+ },
+ {
+ "epoch": 4.53125,
+ "grad_norm": 0.4179284798304916,
+ "learning_rate": 2e-05,
+ "loss": 0.7611,
+ "step": 145
+ },
+ {
+ "epoch": 4.53125,
+ "eval_loss": 0.7089446783065796,
+ "eval_runtime": 53.4752,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 145
+ },
+ {
+ "epoch": 4.5625,
+ "grad_norm": 0.4038858950888911,
+ "learning_rate": 2e-05,
+ "loss": 0.6652,
+ "step": 146
+ },
+ {
+ "epoch": 4.5625,
+ "eval_loss": 0.7089542150497437,
+ "eval_runtime": 53.4741,
+ "eval_samples_per_second": 3.74,
+ "eval_steps_per_second": 0.243,
+ "step": 146
+ },
+ {
+ "epoch": 4.59375,
+ "grad_norm": 0.41740068710674544,
+ "learning_rate": 2e-05,
+ "loss": 0.7319,
+ "step": 147
+ },
+ {
+ "epoch": 4.59375,
+ "eval_loss": 0.7090431451797485,
+ "eval_runtime": 53.2419,
+ "eval_samples_per_second": 3.756,
+ "eval_steps_per_second": 0.244,
+ "step": 147
+ },
+ {
+ "epoch": 4.625,
+ "grad_norm": 0.4288335811568808,
+ "learning_rate": 2e-05,
+ "loss": 0.6837,
+ "step": 148
+ },
+ {
+ "epoch": 4.625,
+ "eval_loss": 0.7088204026222229,
+ "eval_runtime": 53.3614,
+ "eval_samples_per_second": 3.748,
+ "eval_steps_per_second": 0.244,
+ "step": 148
+ },
+ {
+ "epoch": 4.65625,
+ "grad_norm": 0.399955010119186,
+ "learning_rate": 2e-05,
+ "loss": 0.7989,
+ "step": 149
+ },
+ {
+ "epoch": 4.65625,
+ "eval_loss": 0.7084855437278748,
+ "eval_runtime": 53.4923,
+ "eval_samples_per_second": 3.739,
+ "eval_steps_per_second": 0.243,
+ "step": 149
+ },
+ {
+ "epoch": 4.6875,
+ "grad_norm": 0.41794643164255846,
+ "learning_rate": 2e-05,
+ "loss": 0.7194,
+ "step": 150
+ },
+ {
+ "epoch": 4.6875,
+ "eval_loss": 0.7080708146095276,
+ "eval_runtime": 53.639,
+ "eval_samples_per_second": 3.729,
+ "eval_steps_per_second": 0.242,
+ "step": 150
+ },
+ {
+ "epoch": 4.71875,
+ "grad_norm": 0.40953367303148197,
+ "learning_rate": 2e-05,
+ "loss": 0.7354,
+ "step": 151
+ },
+ {
+ "epoch": 4.71875,
+ "eval_loss": 0.7077429890632629,
+ "eval_runtime": 53.3837,
+ "eval_samples_per_second": 3.746,
+ "eval_steps_per_second": 0.244,
+ "step": 151
+ },
+ {
+ "epoch": 4.75,
+ "grad_norm": 0.5012282841513718,
+ "learning_rate": 2e-05,
+ "loss": 0.7662,
+ "step": 152
+ },
+ {
+ "epoch": 4.75,
+ "eval_loss": 0.7064151167869568,
+ "eval_runtime": 53.3549,
+ "eval_samples_per_second": 3.748,
+ "eval_steps_per_second": 0.244,
+ "step": 152
+ },
+ {
+ "epoch": 4.78125,
+ "grad_norm": 0.4210784420989087,
+ "learning_rate": 2e-05,
+ "loss": 0.7133,
+ "step": 153
+ },
+ {
+ "epoch": 4.78125,
+ "eval_loss": 0.7052726745605469,
+ "eval_runtime": 53.5059,
+ "eval_samples_per_second": 3.738,
+ "eval_steps_per_second": 0.243,
+ "step": 153
+ },
+ {
+ "epoch": 4.8125,
+ "grad_norm": 0.43520348530514996,
+ "learning_rate": 2e-05,
+ "loss": 0.729,
+ "step": 154
+ },
+ {
+ "epoch": 4.8125,
+ "eval_loss": 0.7045274972915649,
+ "eval_runtime": 53.8352,
+ "eval_samples_per_second": 3.715,
+ "eval_steps_per_second": 0.241,
+ "step": 154
+ },
+ {
+ "epoch": 4.84375,
+ "grad_norm": 0.4287647569802656,
+ "learning_rate": 2e-05,
+ "loss": 0.6727,
+ "step": 155
+ },
+ {
+ "epoch": 4.84375,
+ "eval_loss": 0.7041358947753906,
+ "eval_runtime": 53.7435,
+ "eval_samples_per_second": 3.721,
+ "eval_steps_per_second": 0.242,
+ "step": 155
+ },
+ {
+ "epoch": 4.875,
+ "grad_norm": 0.41883715320456333,
+ "learning_rate": 2e-05,
+ "loss": 0.7755,
+ "step": 156
+ },
+ {
+ "epoch": 4.875,
+ "eval_loss": 0.7037128210067749,
+ "eval_runtime": 53.8035,
+ "eval_samples_per_second": 3.717,
+ "eval_steps_per_second": 0.242,
+ "step": 156
+ },
+ {
+ "epoch": 4.90625,
+ "grad_norm": 0.40617584505395354,
+ "learning_rate": 2e-05,
+ "loss": 0.7776,
+ "step": 157
+ },
+ {
+ "epoch": 4.90625,
+ "eval_loss": 0.703965425491333,
+ "eval_runtime": 53.8731,
+ "eval_samples_per_second": 3.712,
+ "eval_steps_per_second": 0.241,
+ "step": 157
+ },
+ {
+ "epoch": 4.9375,
+ "grad_norm": 0.4085802225532245,
+ "learning_rate": 2e-05,
+ "loss": 0.7628,
+ "step": 158
+ },
+ {
+ "epoch": 4.9375,
+ "eval_loss": 0.7040860056877136,
+ "eval_runtime": 53.9059,
+ "eval_samples_per_second": 3.71,
+ "eval_steps_per_second": 0.241,
+ "step": 158
+ },
+ {
+ "epoch": 4.96875,
+ "grad_norm": 0.418039298119887,
+ "learning_rate": 2e-05,
+ "loss": 0.7221,
+ "step": 159
+ },
+ {
+ "epoch": 4.96875,
+ "eval_loss": 0.7039948105812073,
+ "eval_runtime": 53.7323,
+ "eval_samples_per_second": 3.722,
+ "eval_steps_per_second": 0.242,
+ "step": 159
+ },
+ {
+ "epoch": 5.0,
+ "grad_norm": 0.46118870048713073,
+ "learning_rate": 2e-05,
+ "loss": 0.7029,
+ "step": 160
+ },
+ {
+ "epoch": 5.0,
+ "eval_loss": 0.703814685344696,
+ "eval_runtime": 53.8975,
+ "eval_samples_per_second": 3.711,
+ "eval_steps_per_second": 0.241,
+ "step": 160
+ },
+ {
+ "epoch": 5.03125,
+ "grad_norm": 0.431474386110294,
+ "learning_rate": 2e-05,
+ "loss": 0.6772,
+ "step": 161
+ },
+ {
+ "epoch": 5.03125,
+ "eval_loss": 0.7034456133842468,
+ "eval_runtime": 51.1105,
+ "eval_samples_per_second": 3.913,
+ "eval_steps_per_second": 0.254,
+ "step": 161
+ },
+ {
+ "epoch": 5.0625,
+ "grad_norm": 0.39618929325750435,
+ "learning_rate": 2e-05,
+ "loss": 0.8219,
+ "step": 162
+ },
+ {
+ "epoch": 5.0625,
+ "eval_loss": 0.7042189240455627,
+ "eval_runtime": 47.2927,
+ "eval_samples_per_second": 4.229,
+ "eval_steps_per_second": 0.275,
+ "step": 162
+ },
+ {
+ "epoch": 5.09375,
+ "grad_norm": 0.4489132713249424,
+ "learning_rate": 2e-05,
+ "loss": 0.6387,
+ "step": 163
+ },
+ {
+ "epoch": 5.09375,
+ "eval_loss": 0.7061256170272827,
+ "eval_runtime": 47.387,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 163
+ },
+ {
+ "epoch": 5.125,
+ "grad_norm": 0.5100329637159183,
+ "learning_rate": 2e-05,
+ "loss": 0.7677,
+ "step": 164
+ },
+ {
+ "epoch": 5.125,
+ "eval_loss": 0.708121657371521,
+ "eval_runtime": 47.3311,
+ "eval_samples_per_second": 4.226,
+ "eval_steps_per_second": 0.275,
+ "step": 164
+ },
+ {
+ "epoch": 5.15625,
+ "grad_norm": 0.525511631981176,
+ "learning_rate": 2e-05,
+ "loss": 0.5956,
+ "step": 165
+ },
+ {
+ "epoch": 5.15625,
+ "eval_loss": 0.7091134786605835,
+ "eval_runtime": 47.2978,
+ "eval_samples_per_second": 4.229,
+ "eval_steps_per_second": 0.275,
+ "step": 165
+ },
+ {
+ "epoch": 5.1875,
+ "grad_norm": 0.534675354231597,
+ "learning_rate": 2e-05,
+ "loss": 0.7097,
+ "step": 166
+ },
+ {
+ "epoch": 5.1875,
+ "eval_loss": 0.7097848653793335,
+ "eval_runtime": 47.4095,
+ "eval_samples_per_second": 4.219,
+ "eval_steps_per_second": 0.274,
+ "step": 166
+ },
+ {
+ "epoch": 5.21875,
+ "grad_norm": 0.47286903698857446,
+ "learning_rate": 2e-05,
+ "loss": 0.7371,
+ "step": 167
+ },
+ {
+ "epoch": 5.21875,
+ "eval_loss": 0.7090296745300293,
+ "eval_runtime": 47.4487,
+ "eval_samples_per_second": 4.215,
+ "eval_steps_per_second": 0.274,
+ "step": 167
+ },
+ {
+ "epoch": 5.25,
+ "grad_norm": 0.4734705066820788,
+ "learning_rate": 2e-05,
+ "loss": 0.7652,
+ "step": 168
+ },
+ {
+ "epoch": 5.25,
+ "eval_loss": 0.7079525589942932,
+ "eval_runtime": 47.4101,
+ "eval_samples_per_second": 4.219,
+ "eval_steps_per_second": 0.274,
+ "step": 168
+ },
+ {
+ "epoch": 5.28125,
+ "grad_norm": 0.46209764763985184,
+ "learning_rate": 2e-05,
+ "loss": 0.6852,
+ "step": 169
+ },
+ {
+ "epoch": 5.28125,
+ "eval_loss": 0.7072803974151611,
+ "eval_runtime": 47.3704,
+ "eval_samples_per_second": 4.222,
+ "eval_steps_per_second": 0.274,
+ "step": 169
+ },
+ {
+ "epoch": 5.3125,
+ "grad_norm": 0.4828284708486433,
+ "learning_rate": 2e-05,
+ "loss": 0.6609,
+ "step": 170
+ },
+ {
+ "epoch": 5.3125,
+ "eval_loss": 0.7068901062011719,
+ "eval_runtime": 47.425,
+ "eval_samples_per_second": 4.217,
+ "eval_steps_per_second": 0.274,
+ "step": 170
+ },
+ {
+ "epoch": 5.34375,
+ "grad_norm": 0.5230116179180577,
+ "learning_rate": 2e-05,
+ "loss": 0.6872,
+ "step": 171
+ },
+ {
+ "epoch": 5.34375,
+ "eval_loss": 0.7058187127113342,
+ "eval_runtime": 47.5711,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 171
+ },
+ {
+ "epoch": 5.375,
+ "grad_norm": 0.48081340678536255,
+ "learning_rate": 2e-05,
+ "loss": 0.7694,
+ "step": 172
+ },
+ {
+ "epoch": 5.375,
+ "eval_loss": 0.7044984698295593,
+ "eval_runtime": 47.4233,
+ "eval_samples_per_second": 4.217,
+ "eval_steps_per_second": 0.274,
+ "step": 172
+ },
+ {
+ "epoch": 5.40625,
+ "grad_norm": 0.4787525602476421,
+ "learning_rate": 2e-05,
+ "loss": 0.7342,
+ "step": 173
+ },
+ {
+ "epoch": 5.40625,
+ "eval_loss": 0.7032212018966675,
+ "eval_runtime": 47.3534,
+ "eval_samples_per_second": 4.224,
+ "eval_steps_per_second": 0.275,
+ "step": 173
+ },
+ {
+ "epoch": 5.4375,
+ "grad_norm": 0.4871847582306217,
+ "learning_rate": 2e-05,
+ "loss": 0.7562,
+ "step": 174
+ },
+ {
+ "epoch": 5.4375,
+ "eval_loss": 0.7019696235656738,
+ "eval_runtime": 47.382,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 174
+ },
+ {
+ "epoch": 5.46875,
+ "grad_norm": 0.47999745025553603,
+ "learning_rate": 2e-05,
+ "loss": 0.7534,
+ "step": 175
+ },
+ {
+ "epoch": 5.46875,
+ "eval_loss": 0.7014529705047607,
+ "eval_runtime": 47.4435,
+ "eval_samples_per_second": 4.216,
+ "eval_steps_per_second": 0.274,
+ "step": 175
+ },
+ {
+ "epoch": 5.5,
+ "grad_norm": 0.5168030891996357,
+ "learning_rate": 2e-05,
+ "loss": 0.707,
+ "step": 176
+ },
+ {
+ "epoch": 5.5,
+ "eval_loss": 0.6993884444236755,
+ "eval_runtime": 47.4943,
+ "eval_samples_per_second": 4.211,
+ "eval_steps_per_second": 0.274,
+ "step": 176
+ },
+ {
+ "epoch": 5.53125,
+ "grad_norm": 0.536450206978984,
+ "learning_rate": 2e-05,
+ "loss": 0.7318,
+ "step": 177
+ },
+ {
+ "epoch": 5.53125,
+ "eval_loss": 0.6971662640571594,
+ "eval_runtime": 47.4193,
+ "eval_samples_per_second": 4.218,
+ "eval_steps_per_second": 0.274,
+ "step": 177
+ },
+ {
+ "epoch": 5.5625,
+ "grad_norm": 0.45352543205020696,
+ "learning_rate": 2e-05,
+ "loss": 0.7421,
+ "step": 178
+ },
+ {
+ "epoch": 5.5625,
+ "eval_loss": 0.6962605118751526,
+ "eval_runtime": 47.3798,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 178
+ },
+ {
+ "epoch": 5.59375,
+ "grad_norm": 0.5054883443109318,
+ "learning_rate": 2e-05,
+ "loss": 0.6668,
+ "step": 179
+ },
+ {
+ "epoch": 5.59375,
+ "eval_loss": 0.6970357298851013,
+ "eval_runtime": 47.3311,
+ "eval_samples_per_second": 4.226,
+ "eval_steps_per_second": 0.275,
+ "step": 179
+ },
+ {
+ "epoch": 5.625,
+ "grad_norm": 0.49584660418833293,
+ "learning_rate": 2e-05,
+ "loss": 0.6548,
+ "step": 180
+ },
+ {
+ "epoch": 5.625,
+ "eval_loss": 0.6980059146881104,
+ "eval_runtime": 47.299,
+ "eval_samples_per_second": 4.228,
+ "eval_steps_per_second": 0.275,
+ "step": 180
+ },
+ {
+ "epoch": 5.65625,
+ "grad_norm": 0.5114381326491793,
+ "learning_rate": 2e-05,
+ "loss": 0.6691,
+ "step": 181
+ },
+ {
+ "epoch": 5.65625,
+ "eval_loss": 0.6995040774345398,
+ "eval_runtime": 47.3887,
+ "eval_samples_per_second": 4.22,
+ "eval_steps_per_second": 0.274,
+ "step": 181
+ },
+ {
+ "epoch": 5.6875,
+ "grad_norm": 0.48550125668870825,
+ "learning_rate": 2e-05,
+ "loss": 0.6525,
+ "step": 182
+ },
+ {
+ "epoch": 5.6875,
+ "eval_loss": 0.7020326256752014,
+ "eval_runtime": 47.3838,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 182
+ },
+ {
+ "epoch": 5.71875,
+ "grad_norm": 0.5860847796671736,
+ "learning_rate": 2e-05,
+ "loss": 0.674,
+ "step": 183
+ },
+ {
+ "epoch": 5.71875,
+ "eval_loss": 0.7027825713157654,
+ "eval_runtime": 47.3875,
+ "eval_samples_per_second": 4.221,
+ "eval_steps_per_second": 0.274,
+ "step": 183
+ },
+ {
+ "epoch": 5.75,
+ "grad_norm": 0.5535582209035479,
+ "learning_rate": 2e-05,
+ "loss": 0.6643,
+ "step": 184
+ },
+ {
+ "epoch": 5.75,
+ "eval_loss": 0.7025408148765564,
+ "eval_runtime": 47.5534,
+ "eval_samples_per_second": 4.206,
+ "eval_steps_per_second": 0.273,
+ "step": 184
+ },
+ {
+ "epoch": 5.78125,
+ "grad_norm": 0.5443574176405931,
+ "learning_rate": 2e-05,
+ "loss": 0.709,
+ "step": 185
+ },
+ {
+ "epoch": 5.78125,
+ "eval_loss": 0.7007840871810913,
+ "eval_runtime": 47.4469,
+ "eval_samples_per_second": 4.215,
+ "eval_steps_per_second": 0.274,
+ "step": 185
+ },
+ {
+ "epoch": 5.8125,
+ "grad_norm": 0.563830259704143,
+ "learning_rate": 2e-05,
+ "loss": 0.6884,
+ "step": 186
+ },
+ {
+ "epoch": 5.8125,
+ "eval_loss": 0.6979361176490784,
+ "eval_runtime": 49.1203,
+ "eval_samples_per_second": 4.072,
+ "eval_steps_per_second": 0.265,
+ "step": 186
+ },
+ {
+ "epoch": 5.84375,
+ "grad_norm": 0.5094956892765212,
+ "learning_rate": 2e-05,
+ "loss": 0.7318,
+ "step": 187
+ },
+ {
+ "epoch": 5.84375,
+ "eval_loss": 0.6962587237358093,
+ "eval_runtime": 49.1831,
+ "eval_samples_per_second": 4.066,
+ "eval_steps_per_second": 0.264,
+ "step": 187
+ },
+ {
+ "epoch": 5.875,
+ "grad_norm": 0.5264819980742595,
+ "learning_rate": 2e-05,
+ "loss": 0.6746,
+ "step": 188
+ },
+ {
+ "epoch": 5.875,
+ "eval_loss": 0.694776713848114,
+ "eval_runtime": 49.1994,
+ "eval_samples_per_second": 4.065,
+ "eval_steps_per_second": 0.264,
+ "step": 188
+ },
+ {
+ "epoch": 5.90625,
+ "grad_norm": 0.4737429304023209,
+ "learning_rate": 2e-05,
+ "loss": 0.664,
+ "step": 189
+ },
+ {
+ "epoch": 5.90625,
+ "eval_loss": 0.6939517855644226,
+ "eval_runtime": 49.2438,
+ "eval_samples_per_second": 4.061,
+ "eval_steps_per_second": 0.264,
+ "step": 189
+ },
+ {
+ "epoch": 5.9375,
+ "grad_norm": 0.494163934813738,
+ "learning_rate": 2e-05,
+ "loss": 0.6978,
+ "step": 190
+ },
+ {
+ "epoch": 5.9375,
+ "eval_loss": 0.6933834552764893,
+ "eval_runtime": 49.3494,
+ "eval_samples_per_second": 4.053,
+ "eval_steps_per_second": 0.263,
+ "step": 190
+ },
+ {
+ "epoch": 5.96875,
+ "grad_norm": 0.4945972278087299,
+ "learning_rate": 2e-05,
+ "loss": 0.6909,
+ "step": 191
+ },
+ {
+ "epoch": 5.96875,
+ "eval_loss": 0.6924250721931458,
+ "eval_runtime": 50.3255,
+ "eval_samples_per_second": 3.974,
+ "eval_steps_per_second": 0.258,
+ "step": 191
+ },
+ {
+ "epoch": 6.0,
+ "grad_norm": 0.48872556688745233,
+ "learning_rate": 2e-05,
+ "loss": 0.6622,
+ "step": 192
+ },
+ {
+ "epoch": 6.0,
+ "eval_loss": 0.6922193765640259,
+ "eval_runtime": 50.4561,
+ "eval_samples_per_second": 3.964,
+ "eval_steps_per_second": 0.258,
+ "step": 192
+ },
+ {
+ "epoch": 6.03125,
+ "grad_norm": 0.5013452255378538,
+ "learning_rate": 2e-05,
+ "loss": 0.7458,
+ "step": 193
+ },
+ {
+ "epoch": 6.03125,
+ "eval_loss": 0.6931161284446716,
+ "eval_runtime": 50.5049,
+ "eval_samples_per_second": 3.96,
+ "eval_steps_per_second": 0.257,
+ "step": 193
+ },
+ {
+ "epoch": 6.0625,
+ "grad_norm": 0.48271161232093784,
+ "learning_rate": 2e-05,
+ "loss": 0.7171,
+ "step": 194
+ },
+ {
+ "epoch": 6.0625,
+ "eval_loss": 0.6959040760993958,
+ "eval_runtime": 50.2441,
+ "eval_samples_per_second": 3.981,
+ "eval_steps_per_second": 0.259,
+ "step": 194
+ },
+ {
+ "epoch": 6.09375,
+ "grad_norm": 0.5414562703154852,
+ "learning_rate": 2e-05,
+ "loss": 0.6419,
+ "step": 195
+ },
+ {
+ "epoch": 6.09375,
+ "eval_loss": 0.7000604271888733,
+ "eval_runtime": 50.4261,
+ "eval_samples_per_second": 3.966,
+ "eval_steps_per_second": 0.258,
+ "step": 195
+ },
+ {
+ "epoch": 6.125,
+ "grad_norm": 0.5074661247335385,
+ "learning_rate": 2e-05,
+ "loss": 0.6881,
+ "step": 196
+ },
+ {
+ "epoch": 6.125,
+ "eval_loss": 0.7039622664451599,
+ "eval_runtime": 51.5214,
+ "eval_samples_per_second": 3.882,
+ "eval_steps_per_second": 0.252,
+ "step": 196
+ },
+ {
+ "epoch": 6.15625,
+ "grad_norm": 0.5603468534764365,
+ "learning_rate": 2e-05,
+ "loss": 0.7085,
+ "step": 197
+ },
+ {
+ "epoch": 6.15625,
+ "eval_loss": 0.7055023312568665,
+ "eval_runtime": 51.7102,
+ "eval_samples_per_second": 3.868,
+ "eval_steps_per_second": 0.251,
+ "step": 197
+ },
+ {
+ "epoch": 6.1875,
+ "grad_norm": 0.5992190802422799,
+ "learning_rate": 2e-05,
+ "loss": 0.7614,
+ "step": 198
+ },
+ {
+ "epoch": 6.1875,
+ "eval_loss": 0.7046856880187988,
+ "eval_runtime": 51.5464,
+ "eval_samples_per_second": 3.88,
+ "eval_steps_per_second": 0.252,
+ "step": 198
+ },
+ {
+ "epoch": 6.21875,
+ "grad_norm": 0.6293684167527106,
+ "learning_rate": 2e-05,
+ "loss": 0.6435,
+ "step": 199
+ },
+ {
+ "epoch": 6.21875,
+ "eval_loss": 0.7021151781082153,
+ "eval_runtime": 51.5328,
+ "eval_samples_per_second": 3.881,
+ "eval_steps_per_second": 0.252,
+ "step": 199
+ },
+ {
+ "epoch": 6.25,
+ "grad_norm": 0.591265449241434,
+ "learning_rate": 2e-05,
+ "loss": 0.688,
+ "step": 200
+ },
+ {
+ "epoch": 6.25,
+ "eval_loss": 0.7002359628677368,
+ "eval_runtime": 51.5812,
+ "eval_samples_per_second": 3.877,
+ "eval_steps_per_second": 0.252,
+ "step": 200
+ },
+ {
+ "epoch": 6.28125,
+ "grad_norm": 0.543141536526749,
+ "learning_rate": 2e-05,
+ "loss": 0.7027,
+ "step": 201
+ },
+ {
+ "epoch": 6.28125,
+ "eval_loss": 0.6986366510391235,
+ "eval_runtime": 52.6956,
+ "eval_samples_per_second": 3.795,
+ "eval_steps_per_second": 0.247,
+ "step": 201
+ },
+ {
+ "epoch": 6.3125,
+ "grad_norm": 0.5679656300203245,
+ "learning_rate": 2e-05,
+ "loss": 0.625,
+ "step": 202
+ },
+ {
+ "epoch": 6.3125,
+ "eval_loss": 0.698679506778717,
+ "eval_runtime": 52.5102,
+ "eval_samples_per_second": 3.809,
+ "eval_steps_per_second": 0.248,
+ "step": 202
+ },
+ {
+ "epoch": 6.34375,
+ "grad_norm": 0.5285839896523021,
+ "learning_rate": 2e-05,
+ "loss": 0.7687,
+ "step": 203
+ },
+ {
+ "epoch": 6.34375,
+ "eval_loss": 0.7005956768989563,
+ "eval_runtime": 52.6067,
+ "eval_samples_per_second": 3.802,
+ "eval_steps_per_second": 0.247,
+ "step": 203
+ },
+ {
+ "epoch": 6.375,
+ "grad_norm": 0.6512964945211068,
+ "learning_rate": 2e-05,
+ "loss": 0.623,
+ "step": 204
+ },
+ {
+ "epoch": 6.375,
+ "eval_loss": 0.7013595104217529,
+ "eval_runtime": 52.5428,
+ "eval_samples_per_second": 3.806,
+ "eval_steps_per_second": 0.247,
+ "step": 204
+ },
+ {
+ "epoch": 6.40625,
+ "grad_norm": 0.5295248631519638,
+ "learning_rate": 2e-05,
+ "loss": 0.5941,
+ "step": 205
+ },
+ {
+ "epoch": 6.40625,
+ "eval_loss": 0.7016547322273254,
+ "eval_runtime": 52.6142,
+ "eval_samples_per_second": 3.801,
+ "eval_steps_per_second": 0.247,
+ "step": 205
+ },
+ {
+ "epoch": 6.4375,
+ "grad_norm": 0.6134157701434021,
+ "learning_rate": 2e-05,
+ "loss": 0.6506,
+ "step": 206
+ },
+ {
+ "epoch": 6.4375,
+ "eval_loss": 0.7009623646736145,
+ "eval_runtime": 52.1942,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 206
+ },
+ {
+ "epoch": 6.46875,
+ "grad_norm": 0.57886797614996,
+ "learning_rate": 2e-05,
+ "loss": 0.6983,
+ "step": 207
+ },
+ {
+ "epoch": 6.46875,
+ "eval_loss": 0.6988092064857483,
+ "eval_runtime": 52.2577,
+ "eval_samples_per_second": 3.827,
+ "eval_steps_per_second": 0.249,
+ "step": 207
+ },
+ {
+ "epoch": 6.5,
+ "grad_norm": 0.5593482836944472,
+ "learning_rate": 2e-05,
+ "loss": 0.6348,
+ "step": 208
+ },
+ {
+ "epoch": 6.5,
+ "eval_loss": 0.698823094367981,
+ "eval_runtime": 52.2296,
+ "eval_samples_per_second": 3.829,
+ "eval_steps_per_second": 0.249,
+ "step": 208
+ },
+ {
+ "epoch": 6.53125,
+ "grad_norm": 0.662802162179718,
+ "learning_rate": 2e-05,
+ "loss": 0.6206,
+ "step": 209
+ },
+ {
+ "epoch": 6.53125,
+ "eval_loss": 0.6990167498588562,
+ "eval_runtime": 52.4316,
+ "eval_samples_per_second": 3.814,
+ "eval_steps_per_second": 0.248,
+ "step": 209
+ },
+ {
+ "epoch": 6.5625,
+ "grad_norm": 0.6874374231122908,
+ "learning_rate": 2e-05,
+ "loss": 0.6033,
+ "step": 210
+ },
+ {
+ "epoch": 6.5625,
+ "eval_loss": 0.699796736240387,
+ "eval_runtime": 52.3193,
+ "eval_samples_per_second": 3.823,
+ "eval_steps_per_second": 0.248,
+ "step": 210
+ },
+ {
+ "epoch": 6.59375,
+ "grad_norm": 0.6625766736772473,
+ "learning_rate": 2e-05,
+ "loss": 0.6398,
+ "step": 211
+ },
+ {
+ "epoch": 6.59375,
+ "eval_loss": 0.6989737153053284,
+ "eval_runtime": 52.1885,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 211
+ },
+ {
+ "epoch": 6.625,
+ "grad_norm": 0.6563419096027812,
+ "learning_rate": 2e-05,
+ "loss": 0.6119,
+ "step": 212
+ },
+ {
+ "epoch": 6.625,
+ "eval_loss": 0.6973609924316406,
+ "eval_runtime": 52.1628,
+ "eval_samples_per_second": 3.834,
+ "eval_steps_per_second": 0.249,
+ "step": 212
+ },
+ {
+ "epoch": 6.65625,
+ "grad_norm": 0.5796353226697397,
+ "learning_rate": 2e-05,
+ "loss": 0.7041,
+ "step": 213
+ },
+ {
+ "epoch": 6.65625,
+ "eval_loss": 0.6957942247390747,
+ "eval_runtime": 52.2028,
+ "eval_samples_per_second": 3.831,
+ "eval_steps_per_second": 0.249,
+ "step": 213
+ },
+ {
+ "epoch": 6.6875,
+ "grad_norm": 0.5711947110504899,
+ "learning_rate": 2e-05,
+ "loss": 0.6465,
+ "step": 214
+ },
+ {
+ "epoch": 6.6875,
+ "eval_loss": 0.696739673614502,
+ "eval_runtime": 52.1849,
+ "eval_samples_per_second": 3.833,
+ "eval_steps_per_second": 0.249,
+ "step": 214
+ },
+ {
+ "epoch": 6.71875,
+ "grad_norm": 0.6619502413653232,
+ "learning_rate": 2e-05,
+ "loss": 0.6563,
+ "step": 215
+ },
+ {
+ "epoch": 6.71875,
+ "eval_loss": 0.6960940361022949,
+ "eval_runtime": 52.0996,
+ "eval_samples_per_second": 3.839,
+ "eval_steps_per_second": 0.25,
+ "step": 215
+ },
+ {
+ "epoch": 6.75,
+ "grad_norm": 0.6587126256919645,
+ "learning_rate": 2e-05,
+ "loss": 0.6505,
+ "step": 216
+ },
+ {
+ "epoch": 6.75,
+ "eval_loss": 0.6959022283554077,
+ "eval_runtime": 52.1062,
+ "eval_samples_per_second": 3.838,
+ "eval_steps_per_second": 0.249,
+ "step": 216
+ },
+ {
+ "epoch": 6.78125,
+ "grad_norm": 0.648164277941964,
+ "learning_rate": 2e-05,
+ "loss": 0.5969,
+ "step": 217
+ },
+ {
+ "epoch": 6.78125,
+ "eval_loss": 0.6999121308326721,
+ "eval_runtime": 51.9356,
+ "eval_samples_per_second": 3.851,
+ "eval_steps_per_second": 0.25,
+ "step": 217
+ },
+ {
+ "epoch": 6.8125,
+ "grad_norm": 0.6595860789738482,
+ "learning_rate": 2e-05,
+ "loss": 0.5945,
+ "step": 218
+ },
+ {
+ "epoch": 6.8125,
+ "eval_loss": 0.7028067111968994,
+ "eval_runtime": 52.2232,
+ "eval_samples_per_second": 3.83,
+ "eval_steps_per_second": 0.249,
+ "step": 218
+ },
+ {
+ "epoch": 6.84375,
+ "grad_norm": 0.7116894779822719,
+ "learning_rate": 2e-05,
+ "loss": 0.7027,
+ "step": 219
+ },
+ {
+ "epoch": 6.84375,
+ "eval_loss": 0.7035638689994812,
+ "eval_runtime": 52.1471,
+ "eval_samples_per_second": 3.835,
+ "eval_steps_per_second": 0.249,
+ "step": 219
+ },
+ {
+ "epoch": 6.875,
+ "grad_norm": 0.7581142336087988,
+ "learning_rate": 2e-05,
+ "loss": 0.7171,
+ "step": 220
+ },
+ {
+ "epoch": 6.875,
+ "eval_loss": 0.6981176733970642,
+ "eval_runtime": 52.1366,
+ "eval_samples_per_second": 3.836,
+ "eval_steps_per_second": 0.249,
+ "step": 220
+ },
+ {
+ "epoch": 6.90625,
+ "grad_norm": 0.6261292745909233,
+ "learning_rate": 2e-05,
+ "loss": 0.658,
+ "step": 221
+ },
+ {
+ "epoch": 6.90625,
+ "eval_loss": 0.6939045786857605,
+ "eval_runtime": 52.2211,
+ "eval_samples_per_second": 3.83,
+ "eval_steps_per_second": 0.249,
+ "step": 221
+ },
+ {
+ "epoch": 6.9375,
+ "grad_norm": 0.7256427809370966,
+ "learning_rate": 2e-05,
+ "loss": 0.6576,
+ "step": 222
+ },
+ {
+ "epoch": 6.9375,
+ "eval_loss": 0.6904327273368835,
+ "eval_runtime": 52.1829,
+ "eval_samples_per_second": 3.833,
+ "eval_steps_per_second": 0.249,
+ "step": 222
+ },
+ {
+ "epoch": 6.96875,
+ "grad_norm": 0.6653711103404113,
+ "learning_rate": 2e-05,
+ "loss": 0.6938,
+ "step": 223
+ },
+ {
+ "epoch": 6.96875,
+ "eval_loss": 0.6893274188041687,
+ "eval_runtime": 51.899,
+ "eval_samples_per_second": 3.854,
+ "eval_steps_per_second": 0.25,
+ "step": 223
+ },
+ {
+ "epoch": 7.0,
+ "grad_norm": 0.6730688267524797,
+ "learning_rate": 2e-05,
+ "loss": 0.7397,
+ "step": 224
+ },
+ {
+ "epoch": 7.0,
+ "eval_loss": 0.6895740032196045,
+ "eval_runtime": 52.1977,
+ "eval_samples_per_second": 3.832,
+ "eval_steps_per_second": 0.249,
+ "step": 224
+ },
+ {
+ "epoch": 7.03125,
+ "grad_norm": 0.5832904533111831,
+ "learning_rate": 2e-05,
+ "loss": 0.6366,
+ "step": 225
+ },
+ {
+ "epoch": 7.03125,
+ "eval_loss": 0.690305769443512,
+ "eval_runtime": 51.0898,
+ "eval_samples_per_second": 3.915,
+ "eval_steps_per_second": 0.254,
+ "step": 225
+ },
+ {
+ "epoch": 7.0625,
+ "grad_norm": 0.7244416322910332,
+ "learning_rate": 2e-05,
+ "loss": 0.5756,
+ "step": 226
+ },
+ {
+ "epoch": 7.0625,
+ "eval_loss": 0.6943302154541016,
+ "eval_runtime": 47.5876,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 226
+ },
+ {
+ "epoch": 7.09375,
+ "grad_norm": 0.6507055762944723,
+ "learning_rate": 2e-05,
+ "loss": 0.622,
+ "step": 227
+ },
+ {
+ "epoch": 7.09375,
+ "eval_loss": 0.7073258757591248,
+ "eval_runtime": 47.5809,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 227
+ },
+ {
+ "epoch": 7.125,
+ "grad_norm": 0.7122561204700196,
+ "learning_rate": 2e-05,
+ "loss": 0.5908,
+ "step": 228
+ },
+ {
+ "epoch": 7.125,
+ "eval_loss": 0.7263233065605164,
+ "eval_runtime": 47.544,
+ "eval_samples_per_second": 4.207,
+ "eval_steps_per_second": 0.273,
+ "step": 228
+ },
+ {
+ "epoch": 7.15625,
+ "grad_norm": 1.053512823308346,
+ "learning_rate": 2e-05,
+ "loss": 0.6193,
+ "step": 229
+ },
+ {
+ "epoch": 7.15625,
+ "eval_loss": 0.7283624410629272,
+ "eval_runtime": 47.5998,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 229
+ },
+ {
+ "epoch": 7.1875,
+ "grad_norm": 1.0167138351900848,
+ "learning_rate": 2e-05,
+ "loss": 0.5942,
+ "step": 230
+ },
+ {
+ "epoch": 7.1875,
+ "eval_loss": 0.7136476039886475,
+ "eval_runtime": 47.5738,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 230
+ },
+ {
+ "epoch": 7.21875,
+ "grad_norm": 0.7388726343392281,
+ "learning_rate": 2e-05,
+ "loss": 0.6898,
+ "step": 231
+ },
+ {
+ "epoch": 7.21875,
+ "eval_loss": 0.7017656564712524,
+ "eval_runtime": 47.5857,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 231
+ },
+ {
+ "epoch": 7.25,
+ "grad_norm": 0.6255681554939039,
+ "learning_rate": 2e-05,
+ "loss": 0.669,
+ "step": 232
+ },
+ {
+ "epoch": 7.25,
+ "eval_loss": 0.6967242956161499,
+ "eval_runtime": 47.7483,
+ "eval_samples_per_second": 4.189,
+ "eval_steps_per_second": 0.272,
+ "step": 232
+ },
+ {
+ "epoch": 7.28125,
+ "grad_norm": 0.7000438574267057,
+ "learning_rate": 2e-05,
+ "loss": 0.6143,
+ "step": 233
+ },
+ {
+ "epoch": 7.28125,
+ "eval_loss": 0.694460391998291,
+ "eval_runtime": 47.7828,
+ "eval_samples_per_second": 4.186,
+ "eval_steps_per_second": 0.272,
+ "step": 233
+ },
+ {
+ "epoch": 7.3125,
+ "grad_norm": 0.6658391411050186,
+ "learning_rate": 2e-05,
+ "loss": 0.6737,
+ "step": 234
+ },
+ {
+ "epoch": 7.3125,
+ "eval_loss": 0.6925583481788635,
+ "eval_runtime": 47.7913,
+ "eval_samples_per_second": 4.185,
+ "eval_steps_per_second": 0.272,
+ "step": 234
+ },
+ {
+ "epoch": 7.34375,
+ "grad_norm": 0.6473191970636399,
+ "learning_rate": 2e-05,
+ "loss": 0.6347,
+ "step": 235
+ },
+ {
+ "epoch": 7.34375,
+ "eval_loss": 0.6907203793525696,
+ "eval_runtime": 47.6866,
+ "eval_samples_per_second": 4.194,
+ "eval_steps_per_second": 0.273,
+ "step": 235
+ },
+ {
+ "epoch": 7.375,
+ "grad_norm": 0.703409963718735,
+ "learning_rate": 2e-05,
+ "loss": 0.5991,
+ "step": 236
+ },
+ {
+ "epoch": 7.375,
+ "eval_loss": 0.6898574829101562,
+ "eval_runtime": 47.6481,
+ "eval_samples_per_second": 4.197,
+ "eval_steps_per_second": 0.273,
+ "step": 236
+ },
+ {
+ "epoch": 7.40625,
+ "grad_norm": 0.6957469611517898,
+ "learning_rate": 2e-05,
+ "loss": 0.6428,
+ "step": 237
+ },
+ {
+ "epoch": 7.40625,
+ "eval_loss": 0.6922276020050049,
+ "eval_runtime": 47.7072,
+ "eval_samples_per_second": 4.192,
+ "eval_steps_per_second": 0.272,
+ "step": 237
+ },
+ {
+ "epoch": 7.4375,
+ "grad_norm": 0.7383281551578481,
+ "learning_rate": 2e-05,
+ "loss": 0.6272,
+ "step": 238
+ },
+ {
+ "epoch": 7.4375,
+ "eval_loss": 0.6988270282745361,
+ "eval_runtime": 47.5925,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 238
+ },
+ {
+ "epoch": 7.46875,
+ "grad_norm": 0.7113722006702997,
+ "learning_rate": 2e-05,
+ "loss": 0.6594,
+ "step": 239
+ },
+ {
+ "epoch": 7.46875,
+ "eval_loss": 0.7074680328369141,
+ "eval_runtime": 47.7257,
+ "eval_samples_per_second": 4.191,
+ "eval_steps_per_second": 0.272,
+ "step": 239
+ },
+ {
+ "epoch": 7.5,
+ "grad_norm": 0.7233836456752487,
+ "learning_rate": 2e-05,
+ "loss": 0.6003,
+ "step": 240
+ },
+ {
+ "epoch": 7.5,
+ "eval_loss": 0.7172031402587891,
+ "eval_runtime": 47.7463,
+ "eval_samples_per_second": 4.189,
+ "eval_steps_per_second": 0.272,
+ "step": 240
+ },
+ {
+ "epoch": 7.53125,
+ "grad_norm": 0.7452166529670862,
+ "learning_rate": 2e-05,
+ "loss": 0.6463,
+ "step": 241
+ },
+ {
+ "epoch": 7.53125,
+ "eval_loss": 0.7228195071220398,
+ "eval_runtime": 47.6283,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 241
+ },
+ {
+ "epoch": 7.5625,
+ "grad_norm": 0.88949489838851,
+ "learning_rate": 2e-05,
+ "loss": 0.6463,
+ "step": 242
+ },
+ {
+ "epoch": 7.5625,
+ "eval_loss": 0.7194420099258423,
+ "eval_runtime": 47.6221,
+ "eval_samples_per_second": 4.2,
+ "eval_steps_per_second": 0.273,
+ "step": 242
+ },
+ {
+ "epoch": 7.59375,
+ "grad_norm": 0.7592408002786533,
+ "learning_rate": 2e-05,
+ "loss": 0.6301,
+ "step": 243
+ },
+ {
+ "epoch": 7.59375,
+ "eval_loss": 0.7122278809547424,
+ "eval_runtime": 47.7549,
+ "eval_samples_per_second": 4.188,
+ "eval_steps_per_second": 0.272,
+ "step": 243
+ },
+ {
+ "epoch": 7.625,
+ "grad_norm": 0.910753798896517,
+ "learning_rate": 2e-05,
+ "loss": 0.7016,
+ "step": 244
+ },
+ {
+ "epoch": 7.625,
+ "eval_loss": 0.7019688487052917,
+ "eval_runtime": 47.5592,
+ "eval_samples_per_second": 4.205,
+ "eval_steps_per_second": 0.273,
+ "step": 244
+ },
+ {
+ "epoch": 7.65625,
+ "grad_norm": 0.7861795541835009,
+ "learning_rate": 2e-05,
+ "loss": 0.6107,
+ "step": 245
+ },
+ {
+ "epoch": 7.65625,
+ "eval_loss": 0.6964650750160217,
+ "eval_runtime": 47.5842,
+ "eval_samples_per_second": 4.203,
+ "eval_steps_per_second": 0.273,
+ "step": 245
+ },
+ {
+ "epoch": 7.6875,
+ "grad_norm": 0.7162378610377871,
+ "learning_rate": 2e-05,
+ "loss": 0.6474,
+ "step": 246
+ },
+ {
+ "epoch": 7.6875,
+ "eval_loss": 0.6934291124343872,
+ "eval_runtime": 47.4792,
+ "eval_samples_per_second": 4.212,
+ "eval_steps_per_second": 0.274,
+ "step": 246
+ },
+ {
+ "epoch": 7.71875,
+ "grad_norm": 0.7261823254305776,
+ "learning_rate": 2e-05,
+ "loss": 0.636,
+ "step": 247
+ },
+ {
+ "epoch": 7.71875,
+ "eval_loss": 0.6925876140594482,
+ "eval_runtime": 47.6623,
+ "eval_samples_per_second": 4.196,
+ "eval_steps_per_second": 0.273,
+ "step": 247
+ },
+ {
+ "epoch": 7.75,
+ "grad_norm": 0.6757318335309442,
+ "learning_rate": 2e-05,
+ "loss": 0.6249,
+ "step": 248
+ },
+ {
+ "epoch": 7.75,
+ "eval_loss": 0.6934402585029602,
+ "eval_runtime": 47.5464,
+ "eval_samples_per_second": 4.206,
+ "eval_steps_per_second": 0.273,
+ "step": 248
+ },
+ {
+ "epoch": 7.78125,
+ "grad_norm": 0.7182105984315053,
+ "learning_rate": 2e-05,
+ "loss": 0.6676,
+ "step": 249
+ },
+ {
+ "epoch": 7.78125,
+ "eval_loss": 0.6956924200057983,
+ "eval_runtime": 47.6014,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 249
+ },
+ {
+ "epoch": 7.8125,
+ "grad_norm": 0.7231439954921842,
+ "learning_rate": 2e-05,
+ "loss": 0.6719,
+ "step": 250
+ },
+ {
+ "epoch": 7.8125,
+ "eval_loss": 0.6987011432647705,
+ "eval_runtime": 47.64,
+ "eval_samples_per_second": 4.198,
+ "eval_steps_per_second": 0.273,
+ "step": 250
+ },
+ {
+ "epoch": 7.84375,
+ "grad_norm": 0.7938681326839265,
+ "learning_rate": 2e-05,
+ "loss": 0.584,
+ "step": 251
+ },
+ {
+ "epoch": 7.84375,
+ "eval_loss": 0.7026040554046631,
+ "eval_runtime": 47.6391,
+ "eval_samples_per_second": 4.198,
+ "eval_steps_per_second": 0.273,
+ "step": 251
+ },
+ {
+ "epoch": 7.875,
+ "grad_norm": 0.8011657536057513,
+ "learning_rate": 2e-05,
+ "loss": 0.594,
+ "step": 252
+ },
+ {
+ "epoch": 7.875,
+ "eval_loss": 0.7068576216697693,
+ "eval_runtime": 47.635,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 252
+ },
+ {
+ "epoch": 7.90625,
+ "grad_norm": 0.819763617578999,
+ "learning_rate": 2e-05,
+ "loss": 0.6758,
+ "step": 253
+ },
+ {
+ "epoch": 7.90625,
+ "eval_loss": 0.7079121470451355,
+ "eval_runtime": 47.6352,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 253
+ },
+ {
+ "epoch": 7.9375,
+ "grad_norm": 0.7697343122686975,
+ "learning_rate": 2e-05,
+ "loss": 0.6224,
+ "step": 254
+ },
+ {
+ "epoch": 7.9375,
+ "eval_loss": 0.7092974781990051,
+ "eval_runtime": 47.5993,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 254
+ },
+ {
+ "epoch": 7.96875,
+ "grad_norm": 0.8148531217392738,
+ "learning_rate": 2e-05,
+ "loss": 0.5579,
+ "step": 255
+ },
+ {
+ "epoch": 7.96875,
+ "eval_loss": 0.7090660333633423,
+ "eval_runtime": 47.5602,
+ "eval_samples_per_second": 4.205,
+ "eval_steps_per_second": 0.273,
+ "step": 255
+ },
+ {
+ "epoch": 8.0,
+ "grad_norm": 0.7576748044477204,
+ "learning_rate": 2e-05,
+ "loss": 0.609,
+ "step": 256
+ },
+ {
+ "epoch": 8.0,
+ "eval_loss": 0.7068901062011719,
+ "eval_runtime": 47.5944,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 256
+ },
+ {
+ "epoch": 8.03125,
+ "grad_norm": 0.814119412415159,
+ "learning_rate": 2e-05,
+ "loss": 0.5816,
+ "step": 257
+ },
+ {
+ "epoch": 8.03125,
+ "eval_loss": 0.7052778005599976,
+ "eval_runtime": 50.9012,
+ "eval_samples_per_second": 3.929,
+ "eval_steps_per_second": 0.255,
+ "step": 257
+ },
+ {
+ "epoch": 8.0625,
+ "grad_norm": 0.7940502590060119,
+ "learning_rate": 2e-05,
+ "loss": 0.5974,
+ "step": 258
+ },
+ {
+ "epoch": 8.0625,
+ "eval_loss": 0.7055818438529968,
+ "eval_runtime": 47.5726,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 258
+ },
+ {
+ "epoch": 8.09375,
+ "grad_norm": 0.7373690747574106,
+ "learning_rate": 2e-05,
+ "loss": 0.6267,
+ "step": 259
+ },
+ {
+ "epoch": 8.09375,
+ "eval_loss": 0.7084596753120422,
+ "eval_runtime": 47.5924,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 259
+ },
+ {
+ "epoch": 8.125,
+ "grad_norm": 0.8486372724795598,
+ "learning_rate": 2e-05,
+ "loss": 0.6349,
+ "step": 260
+ },
+ {
+ "epoch": 8.125,
+ "eval_loss": 0.7118301391601562,
+ "eval_runtime": 47.9994,
+ "eval_samples_per_second": 4.167,
+ "eval_steps_per_second": 0.271,
+ "step": 260
+ },
+ {
+ "epoch": 8.15625,
+ "grad_norm": 0.8391397763830329,
+ "learning_rate": 2e-05,
+ "loss": 0.5575,
+ "step": 261
+ },
+ {
+ "epoch": 8.15625,
+ "eval_loss": 0.7155640125274658,
+ "eval_runtime": 47.6071,
+ "eval_samples_per_second": 4.201,
+ "eval_steps_per_second": 0.273,
+ "step": 261
+ },
+ {
+ "epoch": 8.1875,
+ "grad_norm": 0.7928693737279656,
+ "learning_rate": 2e-05,
+ "loss": 0.6777,
+ "step": 262
+ },
+ {
+ "epoch": 8.1875,
+ "eval_loss": 0.7209051251411438,
+ "eval_runtime": 47.6324,
+ "eval_samples_per_second": 4.199,
+ "eval_steps_per_second": 0.273,
+ "step": 262
+ },
+ {
+ "epoch": 8.21875,
+ "grad_norm": 0.9171124624201488,
+ "learning_rate": 2e-05,
+ "loss": 0.5582,
+ "step": 263
+ },
+ {
+ "epoch": 8.21875,
+ "eval_loss": 0.7233929634094238,
+ "eval_runtime": 47.7509,
+ "eval_samples_per_second": 4.188,
+ "eval_steps_per_second": 0.272,
+ "step": 263
+ },
+ {
+ "epoch": 8.25,
+ "grad_norm": 0.9128766641132847,
+ "learning_rate": 2e-05,
+ "loss": 0.597,
+ "step": 264
+ },
+ {
+ "epoch": 8.25,
+ "eval_loss": 0.7227862477302551,
+ "eval_runtime": 47.5667,
+ "eval_samples_per_second": 4.205,
+ "eval_steps_per_second": 0.273,
+ "step": 264
+ },
+ {
+ "epoch": 8.28125,
+ "grad_norm": 1.0298171058788395,
+ "learning_rate": 2e-05,
+ "loss": 0.6262,
+ "step": 265
+ },
+ {
+ "epoch": 8.28125,
+ "eval_loss": 0.7159123420715332,
+ "eval_runtime": 47.6441,
+ "eval_samples_per_second": 4.198,
+ "eval_steps_per_second": 0.273,
+ "step": 265
+ },
+ {
+ "epoch": 8.3125,
+ "grad_norm": 0.8345277253579861,
+ "learning_rate": 2e-05,
+ "loss": 0.5973,
+ "step": 266
+ },
+ {
+ "epoch": 8.3125,
+ "eval_loss": 0.7099489569664001,
+ "eval_runtime": 49.5358,
+ "eval_samples_per_second": 4.037,
+ "eval_steps_per_second": 0.262,
+ "step": 266
+ },
+ {
+ "epoch": 8.34375,
+ "grad_norm": 0.8270640865043484,
+ "learning_rate": 2e-05,
+ "loss": 0.5418,
+ "step": 267
+ },
+ {
+ "epoch": 8.34375,
+ "eval_loss": 0.7083099484443665,
+ "eval_runtime": 49.7373,
+ "eval_samples_per_second": 4.021,
+ "eval_steps_per_second": 0.261,
+ "step": 267
+ },
+ {
+ "epoch": 8.375,
+ "grad_norm": 0.8670483383004401,
+ "learning_rate": 2e-05,
+ "loss": 0.5935,
+ "step": 268
+ },
+ {
+ "epoch": 8.375,
+ "eval_loss": 0.7091077566146851,
+ "eval_runtime": 49.6764,
+ "eval_samples_per_second": 4.026,
+ "eval_steps_per_second": 0.262,
+ "step": 268
+ },
+ {
+ "epoch": 8.40625,
+ "grad_norm": 0.8373742279582174,
+ "learning_rate": 2e-05,
+ "loss": 0.5947,
+ "step": 269
+ },
+ {
+ "epoch": 8.40625,
+ "eval_loss": 0.709764301776886,
+ "eval_runtime": 49.5613,
+ "eval_samples_per_second": 4.035,
+ "eval_steps_per_second": 0.262,
+ "step": 269
+ },
+ {
+ "epoch": 8.4375,
+ "grad_norm": 0.9406584622840672,
+ "learning_rate": 2e-05,
+ "loss": 0.6079,
+ "step": 270
+ },
+ {
+ "epoch": 8.4375,
+ "eval_loss": 0.7089658379554749,
+ "eval_runtime": 49.6241,
+ "eval_samples_per_second": 4.03,
+ "eval_steps_per_second": 0.262,
+ "step": 270
+ },
+ {
+ "epoch": 8.46875,
+ "grad_norm": 0.9394463996884406,
+ "learning_rate": 2e-05,
+ "loss": 0.5102,
+ "step": 271
+ },
+ {
+ "epoch": 8.46875,
+ "eval_loss": 0.7126440405845642,
+ "eval_runtime": 50.6997,
+ "eval_samples_per_second": 3.945,
+ "eval_steps_per_second": 0.256,
+ "step": 271
+ },
+ {
+ "epoch": 8.5,
+ "grad_norm": 0.8618711805362732,
+ "learning_rate": 2e-05,
+ "loss": 0.5883,
+ "step": 272
+ },
+ {
+ "epoch": 8.5,
+ "eval_loss": 0.7210386395454407,
+ "eval_runtime": 47.7127,
+ "eval_samples_per_second": 4.192,
+ "eval_steps_per_second": 0.272,
+ "step": 272
+ },
+ {
+ "epoch": 8.53125,
+ "grad_norm": 0.9598465596200918,
+ "learning_rate": 2e-05,
+ "loss": 0.5958,
+ "step": 273
+ },
+ {
+ "epoch": 8.53125,
+ "eval_loss": 0.7250240445137024,
+ "eval_runtime": 47.5731,
+ "eval_samples_per_second": 4.204,
+ "eval_steps_per_second": 0.273,
+ "step": 273
+ },
+ {
+ "epoch": 8.5625,
+ "grad_norm": 0.9512065591304456,
+ "learning_rate": 2e-05,
+ "loss": 0.5701,
+ "step": 274
+ },
+ {
+ "epoch": 8.5625,
+ "eval_loss": 0.7265011072158813,
+ "eval_runtime": 47.611,
+ "eval_samples_per_second": 4.201,
+ "eval_steps_per_second": 0.273,
+ "step": 274
+ },
+ {
+ "epoch": 8.59375,
+ "grad_norm": 1.0268459491950561,
+ "learning_rate": 2e-05,
+ "loss": 0.6169,
+ "step": 275
+ },
+ {
+ "epoch": 8.59375,
+ "eval_loss": 0.723859965801239,
+ "eval_runtime": 47.5959,
+ "eval_samples_per_second": 4.202,
+ "eval_steps_per_second": 0.273,
+ "step": 275
+ },
+ {
+ "epoch": 8.625,
+ "grad_norm": 0.9424594037649877,
+ "learning_rate": 2e-05,
+ "loss": 0.6084,
+ "step": 276
+ },
+ {
+ "epoch": 8.625,
+ "eval_loss": 0.7198401093482971,
+ "eval_runtime": 49.4929,
+ "eval_samples_per_second": 4.041,
+ "eval_steps_per_second": 0.263,
+ "step": 276
+ },
+ {
+ "epoch": 8.65625,
+ "grad_norm": 0.9035217720347092,
+ "learning_rate": 2e-05,
+ "loss": 0.5512,
+ "step": 277
+ },
+ {
+ "epoch": 8.65625,
+ "eval_loss": 0.7168082594871521,
+ "eval_runtime": 49.6613,
+ "eval_samples_per_second": 4.027,
+ "eval_steps_per_second": 0.262,
+ "step": 277
+ },
+ {
+ "epoch": 8.6875,
+ "grad_norm": 0.8659031266239389,
+ "learning_rate": 2e-05,
+ "loss": 0.5863,
+ "step": 278
+ },
+ {
+ "epoch": 8.6875,
+ "eval_loss": 0.7159530520439148,
+ "eval_runtime": 49.5693,
+ "eval_samples_per_second": 4.035,
+ "eval_steps_per_second": 0.262,
+ "step": 278
+ },
+ {
+ "epoch": 8.71875,
+ "grad_norm": 0.8740167542953284,
+ "learning_rate": 2e-05,
+ "loss": 0.5667,
+ "step": 279
+ },
+ {
+ "epoch": 8.71875,
+ "eval_loss": 0.7145251631736755,
+ "eval_runtime": 49.4465,
+ "eval_samples_per_second": 4.045,
+ "eval_steps_per_second": 0.263,
+ "step": 279
+ },
+ {
+ "epoch": 8.75,
+ "grad_norm": 0.9263844516793406,
+ "learning_rate": 2e-05,
+ "loss": 0.6124,
+ "step": 280
+ },
+ {
+ "epoch": 8.75,
+ "eval_loss": 0.7149668335914612,
+ "eval_runtime": 49.6649,
+ "eval_samples_per_second": 4.027,
+ "eval_steps_per_second": 0.262,
+ "step": 280
+ },
+ {
+ "epoch": 8.78125,
+ "grad_norm": 0.8604543323600852,
+ "learning_rate": 2e-05,
+ "loss": 0.5688,
+ "step": 281
+ },
+ {
+ "epoch": 8.78125,
+ "eval_loss": 0.7160521149635315,
+ "eval_runtime": 50.6672,
+ "eval_samples_per_second": 3.947,
+ "eval_steps_per_second": 0.257,
+ "step": 281
+ },
+ {
+ "epoch": 8.8125,
+ "grad_norm": 0.9357009474127106,
+ "learning_rate": 2e-05,
+ "loss": 0.5463,
+ "step": 282
+ },
+ {
+ "epoch": 8.8125,
+ "eval_loss": 0.7187457084655762,
+ "eval_runtime": 50.6875,
+ "eval_samples_per_second": 3.946,
+ "eval_steps_per_second": 0.256,
+ "step": 282
+ },
+ {
+ "epoch": 8.84375,
+ "grad_norm": 0.8237087244624672,
+ "learning_rate": 2e-05,
+ "loss": 0.5393,
+ "step": 283
+ },
+ {
+ "epoch": 8.84375,
+ "eval_loss": 0.7205131649971008,
+ "eval_runtime": 50.5794,
+ "eval_samples_per_second": 3.954,
+ "eval_steps_per_second": 0.257,
+ "step": 283
+ },
+ {
+ "epoch": 8.875,
+ "grad_norm": 0.8962206816300475,
+ "learning_rate": 2e-05,
+ "loss": 0.484,
+ "step": 284
+ },
+ {
+ "epoch": 8.875,
+ "eval_loss": 0.7228506207466125,
+ "eval_runtime": 50.5953,
+ "eval_samples_per_second": 3.953,
+ "eval_steps_per_second": 0.257,
+ "step": 284
+ },
+ {
+ "epoch": 8.90625,
+ "grad_norm": 0.9983325109069782,
+ "learning_rate": 2e-05,
+ "loss": 0.5592,
+ "step": 285
+ },
+ {
+ "epoch": 8.90625,
+ "eval_loss": 0.7194100022315979,
+ "eval_runtime": 50.8657,
+ "eval_samples_per_second": 3.932,
+ "eval_steps_per_second": 0.256,
+ "step": 285
+ },
+ {
+ "epoch": 8.9375,
+ "grad_norm": 0.8875985843008509,
+ "learning_rate": 2e-05,
+ "loss": 0.6679,
+ "step": 286
+ },
+ {
+ "epoch": 8.9375,
+ "eval_loss": 0.7146596312522888,
+ "eval_runtime": 51.9576,
+ "eval_samples_per_second": 3.849,
+ "eval_steps_per_second": 0.25,
+ "step": 286
+ },
+ {
+ "epoch": 8.96875,
+ "grad_norm": 0.8611052694088349,
+ "learning_rate": 2e-05,
+ "loss": 0.5812,
+ "step": 287
+ },
+ {
+ "epoch": 8.96875,
+ "eval_loss": 0.710852861404419,
+ "eval_runtime": 51.9658,
+ "eval_samples_per_second": 3.849,
+ "eval_steps_per_second": 0.25,
+ "step": 287
+ },
+ {
+ "epoch": 9.0,
+ "grad_norm": 0.8497210900533776,
+ "learning_rate": 2e-05,
+ "loss": 0.5212,
+ "step": 288
+ },
+ {
+ "epoch": 9.0,
+ "eval_loss": 0.7121503353118896,
+ "eval_runtime": 51.6828,
+ "eval_samples_per_second": 3.87,
+ "eval_steps_per_second": 0.252,
+ "step": 288
+ },
+ {
+ "epoch": 9.03125,
+ "grad_norm": 0.8921157674462687,
+ "learning_rate": 2e-05,
+ "loss": 0.5437,
+ "step": 289
+ },
+ {
+ "epoch": 9.03125,
+ "eval_loss": 0.7179412841796875,
+ "eval_runtime": 51.9759,
+ "eval_samples_per_second": 3.848,
+ "eval_steps_per_second": 0.25,
+ "step": 289
+ },
+ {
+ "epoch": 9.0625,
+ "grad_norm": 0.9291292967074066,
+ "learning_rate": 2e-05,
+ "loss": 0.5679,
+ "step": 290
+ },
+ {
+ "epoch": 9.0625,
+ "eval_loss": 0.7306573390960693,
+ "eval_runtime": 51.603,
+ "eval_samples_per_second": 3.876,
+ "eval_steps_per_second": 0.252,
+ "step": 290
+ },
+ {
+ "epoch": 9.09375,
+ "grad_norm": 0.9871115113489229,
+ "learning_rate": 2e-05,
+ "loss": 0.5744,
+ "step": 291
+ },
+ {
+ "epoch": 9.09375,
+ "eval_loss": 0.74213707447052,
+ "eval_runtime": 51.5255,
+ "eval_samples_per_second": 3.882,
+ "eval_steps_per_second": 0.252,
+ "step": 291
+ },
+ {
+ "epoch": 9.125,
+ "grad_norm": 1.1662734879135015,
+ "learning_rate": 2e-05,
+ "loss": 0.5274,
+ "step": 292
+ },
+ {
+ "epoch": 9.125,
+ "eval_loss": 0.7484179139137268,
+ "eval_runtime": 51.3131,
+ "eval_samples_per_second": 3.898,
+ "eval_steps_per_second": 0.253,
+ "step": 292
+ },
+ {
+ "epoch": 9.15625,
+ "grad_norm": 1.096240777006249,
+ "learning_rate": 2e-05,
+ "loss": 0.5864,
+ "step": 293
+ },
+ {
+ "epoch": 9.15625,
+ "eval_loss": 0.745439887046814,
+ "eval_runtime": 51.1121,
+ "eval_samples_per_second": 3.913,
+ "eval_steps_per_second": 0.254,
+ "step": 293
+ },
+ {
+ "epoch": 9.1875,
+ "grad_norm": 0.944903135330694,
+ "learning_rate": 2e-05,
+ "loss": 0.5131,
+ "step": 294
+ },
+ {
+ "epoch": 9.1875,
+ "eval_loss": 0.7430945038795471,
+ "eval_runtime": 51.307,
+ "eval_samples_per_second": 3.898,
+ "eval_steps_per_second": 0.253,
+ "step": 294
+ },
+ {
+ "epoch": 9.21875,
+ "grad_norm": 1.0736115005040638,
+ "learning_rate": 2e-05,
+ "loss": 0.4866,
+ "step": 295
+ },
+ {
+ "epoch": 9.21875,
+ "eval_loss": 0.7417933940887451,
+ "eval_runtime": 51.2372,
+ "eval_samples_per_second": 3.903,
+ "eval_steps_per_second": 0.254,
+ "step": 295
+ },
+ {
+ "epoch": 9.25,
+ "grad_norm": 1.0688144195951634,
+ "learning_rate": 2e-05,
+ "loss": 0.509,
+ "step": 296
+ },
+ {
+ "epoch": 9.25,
+ "eval_loss": 0.7381229996681213,
+ "eval_runtime": 51.1494,
+ "eval_samples_per_second": 3.91,
+ "eval_steps_per_second": 0.254,
+ "step": 296
+ },
+ {
+ "epoch": 9.28125,
+ "grad_norm": 1.0276146013155785,
+ "learning_rate": 2e-05,
+ "loss": 0.5708,
+ "step": 297
+ },
+ {
+ "epoch": 9.28125,
+ "eval_loss": 0.7391738891601562,
+ "eval_runtime": 51.6779,
+ "eval_samples_per_second": 3.87,
+ "eval_steps_per_second": 0.252,
+ "step": 297
+ },
+ {
+ "epoch": 9.3125,
+ "grad_norm": 1.1618114955183,
+ "learning_rate": 2e-05,
+ "loss": 0.5337,
+ "step": 298
+ },
+ {
+ "epoch": 9.3125,
+ "eval_loss": 0.7411096096038818,
+ "eval_runtime": 51.5937,
+ "eval_samples_per_second": 3.876,
+ "eval_steps_per_second": 0.252,
+ "step": 298
+ },
+ {
+ "epoch": 9.34375,
+ "grad_norm": 1.08837375836462,
+ "learning_rate": 2e-05,
+ "loss": 0.5241,
+ "step": 299
+ },
+ {
+ "epoch": 9.34375,
+ "eval_loss": 0.7420552968978882,
+ "eval_runtime": 51.5437,
+ "eval_samples_per_second": 3.88,
+ "eval_steps_per_second": 0.252,
+ "step": 299
+ },
+ {
+ "epoch": 9.375,
+ "grad_norm": 1.0106379800787466,
+ "learning_rate": 2e-05,
+ "loss": 0.5198,
+ "step": 300
+ },
+ {
+ "epoch": 9.375,
+ "eval_loss": 0.7437419295310974,
+ "eval_runtime": 51.3565,
+ "eval_samples_per_second": 3.894,
+ "eval_steps_per_second": 0.253,
+ "step": 300
+ },
+ {
+ "epoch": 9.40625,
+ "grad_norm": 1.0700897207702011,
+ "learning_rate": 2e-05,
+ "loss": 0.5107,
+ "step": 301
+ },
+ {
+ "epoch": 9.40625,
+ "eval_loss": 0.7382708787918091,
+ "eval_runtime": 51.4533,
+ "eval_samples_per_second": 3.887,
+ "eval_steps_per_second": 0.253,
+ "step": 301
+ },
+ {
+ "epoch": 9.4375,
+ "grad_norm": 1.1021606769115393,
+ "learning_rate": 2e-05,
+ "loss": 0.5679,
+ "step": 302
+ },
+ {
+ "epoch": 9.4375,
+ "eval_loss": 0.7324429154396057,
+ "eval_runtime": 51.4117,
+ "eval_samples_per_second": 3.89,
+ "eval_steps_per_second": 0.253,
+ "step": 302
+ },
+ {
+ "epoch": 9.46875,
+ "grad_norm": 0.9792628984982289,
+ "learning_rate": 2e-05,
+ "loss": 0.5509,
+ "step": 303
+ },
+ {
+ "epoch": 9.46875,
+ "eval_loss": 0.7311490774154663,
+ "eval_runtime": 51.8022,
+ "eval_samples_per_second": 3.861,
+ "eval_steps_per_second": 0.251,
+ "step": 303
+ },
+ {
+ "epoch": 9.5,
+ "grad_norm": 0.9256898215171215,
+ "learning_rate": 2e-05,
+ "loss": 0.5824,
+ "step": 304
+ },
+ {
+ "epoch": 9.5,
+ "eval_loss": 0.736283540725708,
+ "eval_runtime": 51.7678,
+ "eval_samples_per_second": 3.863,
+ "eval_steps_per_second": 0.251,
+ "step": 304
+ },
+ {
+ "epoch": 9.53125,
+ "grad_norm": 0.993495109546069,
+ "learning_rate": 2e-05,
+ "loss": 0.5452,
+ "step": 305
+ },
+ {
+ "epoch": 9.53125,
+ "eval_loss": 0.7425567507743835,
+ "eval_runtime": 51.6022,
+ "eval_samples_per_second": 3.876,
+ "eval_steps_per_second": 0.252,
+ "step": 305
+ },
+ {
+ "epoch": 9.5625,
+ "grad_norm": 1.096995253097988,
+ "learning_rate": 2e-05,
+ "loss": 0.5359,
+ "step": 306
+ },
+ {
+ "epoch": 9.5625,
+ "eval_loss": 0.7483149766921997,
+ "eval_runtime": 51.5727,
+ "eval_samples_per_second": 3.878,
+ "eval_steps_per_second": 0.252,
+ "step": 306
+ },
+ {
+ "epoch": 9.59375,
+ "grad_norm": 1.1542996117677211,
+ "learning_rate": 2e-05,
+ "loss": 0.5229,
+ "step": 307
+ },
+ {
+ "epoch": 9.59375,
+ "eval_loss": 0.7505038380622864,
+ "eval_runtime": 51.846,
+ "eval_samples_per_second": 3.858,
+ "eval_steps_per_second": 0.251,
+ "step": 307
+ },
+ {
+ "epoch": 9.625,
+ "grad_norm": 1.1044494998416634,
+ "learning_rate": 2e-05,
+ "loss": 0.5718,
+ "step": 308
+ },
+ {
+ "epoch": 9.625,
+ "eval_loss": 0.7511885166168213,
+ "eval_runtime": 51.613,
+ "eval_samples_per_second": 3.875,
+ "eval_steps_per_second": 0.252,
+ "step": 308
+ },
+ {
+ "epoch": 9.65625,
+ "grad_norm": 1.0517094139644794,
+ "learning_rate": 2e-05,
+ "loss": 0.5395,
+ "step": 309
+ },
+ {
+ "epoch": 9.65625,
+ "eval_loss": 0.750588059425354,
+ "eval_runtime": 51.9083,
+ "eval_samples_per_second": 3.853,
+ "eval_steps_per_second": 0.25,
+ "step": 309
+ },
+ {
+ "epoch": 9.6875,
+ "grad_norm": 1.2320471917997522,
+ "learning_rate": 2e-05,
+ "loss": 0.5266,
+ "step": 310
+ },
+ {
+ "epoch": 9.6875,
+ "eval_loss": 0.7492180466651917,
+ "eval_runtime": 51.3612,
+ "eval_samples_per_second": 3.894,
+ "eval_steps_per_second": 0.253,
+ "step": 310
+ },
+ {
+ "epoch": 9.71875,
+ "grad_norm": 1.189122697506972,
+ "learning_rate": 2e-05,
+ "loss": 0.4893,
+ "step": 311
+ },
+ {
+ "epoch": 9.71875,
+ "eval_loss": 0.7448427081108093,
+ "eval_runtime": 51.8761,
+ "eval_samples_per_second": 3.855,
+ "eval_steps_per_second": 0.251,
+ "step": 311
+ },
+ {
+ "epoch": 9.75,
+ "grad_norm": 1.1250245833360049,
+ "learning_rate": 2e-05,
+ "loss": 0.5434,
+ "step": 312
+ },
+ {
+ "epoch": 9.75,
+ "eval_loss": 0.742850661277771,
+ "eval_runtime": 51.4442,
+ "eval_samples_per_second": 3.888,
+ "eval_steps_per_second": 0.253,
+ "step": 312
+ },
+ {
+ "epoch": 9.78125,
+ "grad_norm": 1.0320917220089818,
+ "learning_rate": 2e-05,
+ "loss": 0.539,
+ "step": 313
+ },
+ {
+ "epoch": 9.78125,
+ "eval_loss": 0.7389761209487915,
+ "eval_runtime": 51.609,
+ "eval_samples_per_second": 3.875,
+ "eval_steps_per_second": 0.252,
+ "step": 313
+ },
+ {
+ "epoch": 9.8125,
+ "grad_norm": 1.1419373892040323,
+ "learning_rate": 2e-05,
+ "loss": 0.5077,
+ "step": 314
+ },
+ {
+ "epoch": 9.8125,
+ "eval_loss": 0.7384924292564392,
+ "eval_runtime": 51.6937,
+ "eval_samples_per_second": 3.869,
+ "eval_steps_per_second": 0.251,
+ "step": 314
+ },
+ {
+ "epoch": 9.84375,
+ "grad_norm": 1.0260401820964369,
+ "learning_rate": 2e-05,
+ "loss": 0.534,
+ "step": 315
+ },
+ {
+ "epoch": 9.84375,
+ "eval_loss": 0.738023579120636,
+ "eval_runtime": 51.5428,
+ "eval_samples_per_second": 3.88,
+ "eval_steps_per_second": 0.252,
+ "step": 315
+ },
+ {
+ "epoch": 9.875,
+ "grad_norm": 1.0164514553564235,
+ "learning_rate": 2e-05,
+ "loss": 0.5514,
+ "step": 316
+ },
+ {
+ "epoch": 9.875,
+ "eval_loss": 0.7399526834487915,
+ "eval_runtime": 51.6232,
+ "eval_samples_per_second": 3.874,
+ "eval_steps_per_second": 0.252,
+ "step": 316
+ },
+ {
+ "epoch": 9.90625,
+ "grad_norm": 1.1847056085947891,
+ "learning_rate": 2e-05,
+ "loss": 0.5216,
+ "step": 317
+ },
+ {
+ "epoch": 9.90625,
+ "eval_loss": 0.7401251196861267,
+ "eval_runtime": 51.7617,
+ "eval_samples_per_second": 3.864,
+ "eval_steps_per_second": 0.251,
+ "step": 317
+ },
+ {
+ "epoch": 9.9375,
+ "grad_norm": 1.075888871715244,
+ "learning_rate": 2e-05,
+ "loss": 0.511,
+ "step": 318
+ },
+ {
+ "epoch": 9.9375,
+ "eval_loss": 0.739520788192749,
+ "eval_runtime": 51.7458,
+ "eval_samples_per_second": 3.865,
+ "eval_steps_per_second": 0.251,
+ "step": 318
+ },
+ {
+ "epoch": 9.96875,
+ "grad_norm": 1.16238118046427,
+ "learning_rate": 2e-05,
+ "loss": 0.546,
+ "step": 319
+ },
+ {
+ "epoch": 9.96875,
+ "eval_loss": 0.7371450662612915,
+ "eval_runtime": 51.4519,
+ "eval_samples_per_second": 3.887,
+ "eval_steps_per_second": 0.253,
+ "step": 319
+ },
+ {
+ "epoch": 10.0,
+ "grad_norm": 1.109611378591182,
+ "learning_rate": 2e-05,
+ "loss": 0.4855,
+ "step": 320
+ },
+ {
+ "epoch": 10.0,
+ "eval_loss": 0.7406165599822998,
+ "eval_runtime": 51.6984,
+ "eval_samples_per_second": 3.869,
+ "eval_steps_per_second": 0.251,
+ "step": 320
+ },
+ {
+ "epoch": 10.0,
+ "step": 320,
+ "total_flos": 461377729855488.0,
+ "train_loss": 0.11189348716288805,
+ "train_runtime": 4599.9883,
+ "train_samples_per_second": 2.174,
+ "train_steps_per_second": 0.07
+ }
+ ],
+ "logging_steps": 1.0,
+ "max_steps": 320,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 10,
+ "save_steps": 5,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 461377729855488.0,
+ "train_batch_size": 8,
+ "trial_name": null,
+ "trial_params": null
+}