Model save
Browse files- README.md +83 -0
- adapter_config.json +31 -0
- adapter_model.safetensors +3 -0
- all_results.json +8 -0
- special_tokens_map.json +30 -0
- tokenizer.json +0 -0
- tokenizer_config.json +43 -0
- train_results.json +8 -0
- trainer_state.json +1614 -0
- training_args.bin +3 -0
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 3 |
+
library_name: peft
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-dpo-lora
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-7b-dpo-lora
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.4932
|
| 22 |
+
- Rewards/chosen: -1.7916
|
| 23 |
+
- Rewards/rejected: -2.7322
|
| 24 |
+
- Rewards/accuracies: 0.7262
|
| 25 |
+
- Rewards/margins: 0.9407
|
| 26 |
+
- Logps/rejected: -535.5854
|
| 27 |
+
- Logps/chosen: -446.6503
|
| 28 |
+
- Logits/rejected: 0.9353
|
| 29 |
+
- Logits/chosen: 0.5257
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-06
|
| 49 |
+
- train_batch_size: 4
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 4
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 32
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 1
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.6087 | 0.1 | 100 | 0.6158 | -0.3136 | -0.5466 | 0.6726 | 0.2330 | -317.0252 | -298.8513 | -2.0360 | -2.1198 |
|
| 67 |
+
| 0.5463 | 0.21 | 200 | 0.5504 | -1.1262 | -1.6978 | 0.6925 | 0.5716 | -432.1413 | -380.1157 | -0.0431 | -0.2986 |
|
| 68 |
+
| 0.4949 | 0.31 | 300 | 0.5161 | -1.6535 | -2.4330 | 0.7183 | 0.7794 | -505.6621 | -432.8479 | 0.4034 | 0.1418 |
|
| 69 |
+
| 0.5239 | 0.42 | 400 | 0.5101 | -1.3693 | -2.0810 | 0.7302 | 0.7116 | -470.4624 | -404.4282 | 0.8585 | 0.5591 |
|
| 70 |
+
| 0.5272 | 0.52 | 500 | 0.5003 | -2.0358 | -2.9629 | 0.7381 | 0.9271 | -558.6534 | -471.0703 | 1.0404 | 0.7150 |
|
| 71 |
+
| 0.4886 | 0.63 | 600 | 0.4982 | -1.7739 | -2.6428 | 0.7262 | 0.8689 | -526.6414 | -444.8822 | 0.3752 | 0.0594 |
|
| 72 |
+
| 0.516 | 0.73 | 700 | 0.4933 | -2.0243 | -2.9388 | 0.7302 | 0.9144 | -556.2413 | -469.9273 | 0.8898 | 0.5312 |
|
| 73 |
+
| 0.495 | 0.84 | 800 | 0.4949 | -1.7382 | -2.6840 | 0.7262 | 0.9458 | -530.7620 | -441.3121 | 0.8308 | 0.4157 |
|
| 74 |
+
| 0.4866 | 0.94 | 900 | 0.4932 | -1.7916 | -2.7322 | 0.7262 | 0.9407 | -535.5854 | -446.6503 | 0.9353 | 0.5257 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- PEFT 0.7.1
|
| 80 |
+
- Transformers 4.38.2
|
| 81 |
+
- Pytorch 2.1.2+cu121
|
| 82 |
+
- Datasets 2.14.6
|
| 83 |
+
- Tokenizers 0.15.2
|
adapter_config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "mistralai/Mistral-7B-v0.1",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"fan_in_fan_out": false,
|
| 7 |
+
"inference_mode": true,
|
| 8 |
+
"init_lora_weights": true,
|
| 9 |
+
"layers_pattern": null,
|
| 10 |
+
"layers_to_transform": null,
|
| 11 |
+
"loftq_config": {},
|
| 12 |
+
"lora_alpha": 128,
|
| 13 |
+
"lora_dropout": 0.05,
|
| 14 |
+
"megatron_config": null,
|
| 15 |
+
"megatron_core": "megatron.core",
|
| 16 |
+
"modules_to_save": null,
|
| 17 |
+
"peft_type": "LORA",
|
| 18 |
+
"r": 64,
|
| 19 |
+
"rank_pattern": {},
|
| 20 |
+
"revision": null,
|
| 21 |
+
"target_modules": [
|
| 22 |
+
"q_proj",
|
| 23 |
+
"o_proj",
|
| 24 |
+
"gate_proj",
|
| 25 |
+
"up_proj",
|
| 26 |
+
"v_proj",
|
| 27 |
+
"k_proj",
|
| 28 |
+
"down_proj"
|
| 29 |
+
],
|
| 30 |
+
"task_type": "CAUSAL_LM"
|
| 31 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a30a7a4badef97c6dfcd8926c13ead9c59cb66a2a3eeb2b655d70e26c1cef415
|
| 3 |
+
size 335605144
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5264237219126436,
|
| 4 |
+
"train_runtime": 29801.8239,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 2.051,
|
| 7 |
+
"train_steps_per_second": 0.032
|
| 8 |
+
}
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "</s>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"0": {
|
| 6 |
+
"content": "<unk>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"1": {
|
| 14 |
+
"content": "<s>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"2": {
|
| 22 |
+
"content": "</s>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
}
|
| 29 |
+
},
|
| 30 |
+
"additional_special_tokens": [],
|
| 31 |
+
"bos_token": "<s>",
|
| 32 |
+
"chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n' + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
|
| 33 |
+
"clean_up_tokenization_spaces": false,
|
| 34 |
+
"eos_token": "</s>",
|
| 35 |
+
"legacy": true,
|
| 36 |
+
"model_max_length": 2048,
|
| 37 |
+
"pad_token": "</s>",
|
| 38 |
+
"sp_model_kwargs": {},
|
| 39 |
+
"spaces_between_special_tokens": false,
|
| 40 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 41 |
+
"unk_token": "<unk>",
|
| 42 |
+
"use_default_system_prompt": false
|
| 43 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5264237219126436,
|
| 4 |
+
"train_runtime": 29801.8239,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 2.051,
|
| 7 |
+
"train_steps_per_second": 0.032
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1614 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9997382884061764,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 955,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 1.6872971730727813,
|
| 14 |
+
"learning_rate": 5.208333333333333e-08,
|
| 15 |
+
"logits/chosen": -2.4102981090545654,
|
| 16 |
+
"logits/rejected": -2.4143850803375244,
|
| 17 |
+
"logps/chosen": -352.07745361328125,
|
| 18 |
+
"logps/rejected": -290.7293395996094,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01,
|
| 28 |
+
"grad_norm": 1.7117396873541209,
|
| 29 |
+
"learning_rate": 5.208333333333334e-07,
|
| 30 |
+
"logits/chosen": -2.2590177059173584,
|
| 31 |
+
"logits/rejected": -2.1774637699127197,
|
| 32 |
+
"logps/chosen": -236.37635803222656,
|
| 33 |
+
"logps/rejected": -228.0323944091797,
|
| 34 |
+
"loss": 0.6931,
|
| 35 |
+
"rewards/accuracies": 0.3819444477558136,
|
| 36 |
+
"rewards/chosen": -0.0001151897813542746,
|
| 37 |
+
"rewards/margins": -0.0005040301475673914,
|
| 38 |
+
"rewards/rejected": 0.00038884030072949827,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02,
|
| 43 |
+
"grad_norm": 1.5684536992232367,
|
| 44 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 45 |
+
"logits/chosen": -2.3643908500671387,
|
| 46 |
+
"logits/rejected": -2.298231840133667,
|
| 47 |
+
"logps/chosen": -282.7557067871094,
|
| 48 |
+
"logps/rejected": -264.20379638671875,
|
| 49 |
+
"loss": 0.6927,
|
| 50 |
+
"rewards/accuracies": 0.550000011920929,
|
| 51 |
+
"rewards/chosen": 0.0016558446222916245,
|
| 52 |
+
"rewards/margins": 0.000754863431211561,
|
| 53 |
+
"rewards/rejected": 0.0009009811910800636,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.03,
|
| 58 |
+
"grad_norm": 1.6320638446100388,
|
| 59 |
+
"learning_rate": 1.5625e-06,
|
| 60 |
+
"logits/chosen": -2.2148165702819824,
|
| 61 |
+
"logits/rejected": -2.1439080238342285,
|
| 62 |
+
"logps/chosen": -254.8279266357422,
|
| 63 |
+
"logps/rejected": -239.754150390625,
|
| 64 |
+
"loss": 0.6909,
|
| 65 |
+
"rewards/accuracies": 0.6875,
|
| 66 |
+
"rewards/chosen": 0.00770978769287467,
|
| 67 |
+
"rewards/margins": 0.005677036941051483,
|
| 68 |
+
"rewards/rejected": 0.002032750751823187,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04,
|
| 73 |
+
"grad_norm": 1.5505793749959536,
|
| 74 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 75 |
+
"logits/chosen": -2.4151053428649902,
|
| 76 |
+
"logits/rejected": -2.267972946166992,
|
| 77 |
+
"logps/chosen": -289.40142822265625,
|
| 78 |
+
"logps/rejected": -265.69000244140625,
|
| 79 |
+
"loss": 0.6867,
|
| 80 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 81 |
+
"rewards/chosen": 0.019891003146767616,
|
| 82 |
+
"rewards/margins": 0.008773349225521088,
|
| 83 |
+
"rewards/rejected": 0.011117652989923954,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.05,
|
| 88 |
+
"grad_norm": 1.6726051946382139,
|
| 89 |
+
"learning_rate": 2.604166666666667e-06,
|
| 90 |
+
"logits/chosen": -2.336714029312134,
|
| 91 |
+
"logits/rejected": -2.256179094314575,
|
| 92 |
+
"logps/chosen": -245.6073455810547,
|
| 93 |
+
"logps/rejected": -228.17562866210938,
|
| 94 |
+
"loss": 0.6793,
|
| 95 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 96 |
+
"rewards/chosen": 0.037440598011016846,
|
| 97 |
+
"rewards/margins": 0.02772732451558113,
|
| 98 |
+
"rewards/rejected": 0.009713277220726013,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.06,
|
| 103 |
+
"grad_norm": 1.7944993610396887,
|
| 104 |
+
"learning_rate": 3.125e-06,
|
| 105 |
+
"logits/chosen": -2.289384365081787,
|
| 106 |
+
"logits/rejected": -2.2603793144226074,
|
| 107 |
+
"logps/chosen": -252.38510131835938,
|
| 108 |
+
"logps/rejected": -252.2633819580078,
|
| 109 |
+
"loss": 0.6717,
|
| 110 |
+
"rewards/accuracies": 0.699999988079071,
|
| 111 |
+
"rewards/chosen": 0.04392694681882858,
|
| 112 |
+
"rewards/margins": 0.0502941831946373,
|
| 113 |
+
"rewards/rejected": -0.006367234978824854,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.07,
|
| 118 |
+
"grad_norm": 2.0272577815121133,
|
| 119 |
+
"learning_rate": 3.6458333333333333e-06,
|
| 120 |
+
"logits/chosen": -2.276283025741577,
|
| 121 |
+
"logits/rejected": -2.1802830696105957,
|
| 122 |
+
"logps/chosen": -252.1732635498047,
|
| 123 |
+
"logps/rejected": -269.9049987792969,
|
| 124 |
+
"loss": 0.6579,
|
| 125 |
+
"rewards/accuracies": 0.668749988079071,
|
| 126 |
+
"rewards/chosen": -0.014363644644618034,
|
| 127 |
+
"rewards/margins": 0.0627121776342392,
|
| 128 |
+
"rewards/rejected": -0.07707582414150238,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.08,
|
| 133 |
+
"grad_norm": 5.394650048474675,
|
| 134 |
+
"learning_rate": 4.166666666666667e-06,
|
| 135 |
+
"logits/chosen": -2.100991725921631,
|
| 136 |
+
"logits/rejected": -1.9991142749786377,
|
| 137 |
+
"logps/chosen": -265.56060791015625,
|
| 138 |
+
"logps/rejected": -263.40496826171875,
|
| 139 |
+
"loss": 0.6436,
|
| 140 |
+
"rewards/accuracies": 0.699999988079071,
|
| 141 |
+
"rewards/chosen": -0.027108073234558105,
|
| 142 |
+
"rewards/margins": 0.1258033663034439,
|
| 143 |
+
"rewards/rejected": -0.15291143953800201,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.09,
|
| 148 |
+
"grad_norm": 2.829581544220464,
|
| 149 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 150 |
+
"logits/chosen": -2.1768875122070312,
|
| 151 |
+
"logits/rejected": -2.1092872619628906,
|
| 152 |
+
"logps/chosen": -284.537841796875,
|
| 153 |
+
"logps/rejected": -302.475830078125,
|
| 154 |
+
"loss": 0.629,
|
| 155 |
+
"rewards/accuracies": 0.6875,
|
| 156 |
+
"rewards/chosen": -0.2255394160747528,
|
| 157 |
+
"rewards/margins": 0.15923205018043518,
|
| 158 |
+
"rewards/rejected": -0.384771466255188,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.1,
|
| 163 |
+
"grad_norm": 5.983992522046318,
|
| 164 |
+
"learning_rate": 4.9997324926814375e-06,
|
| 165 |
+
"logits/chosen": -2.1670639514923096,
|
| 166 |
+
"logits/rejected": -2.1124117374420166,
|
| 167 |
+
"logps/chosen": -293.4961242675781,
|
| 168 |
+
"logps/rejected": -312.59320068359375,
|
| 169 |
+
"loss": 0.6087,
|
| 170 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 171 |
+
"rewards/chosen": -0.15919722616672516,
|
| 172 |
+
"rewards/margins": 0.22884318232536316,
|
| 173 |
+
"rewards/rejected": -0.3880404233932495,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.1,
|
| 178 |
+
"eval_logits/chosen": -2.1197569370269775,
|
| 179 |
+
"eval_logits/rejected": -2.036003828048706,
|
| 180 |
+
"eval_logps/chosen": -298.85125732421875,
|
| 181 |
+
"eval_logps/rejected": -317.0252380371094,
|
| 182 |
+
"eval_loss": 0.6158178448677063,
|
| 183 |
+
"eval_rewards/accuracies": 0.6726190447807312,
|
| 184 |
+
"eval_rewards/chosen": -0.313564270734787,
|
| 185 |
+
"eval_rewards/margins": 0.23304378986358643,
|
| 186 |
+
"eval_rewards/rejected": -0.546608030796051,
|
| 187 |
+
"eval_runtime": 364.711,
|
| 188 |
+
"eval_samples_per_second": 5.484,
|
| 189 |
+
"eval_steps_per_second": 0.173,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.12,
|
| 194 |
+
"grad_norm": 4.31001099339475,
|
| 195 |
+
"learning_rate": 4.996723692767927e-06,
|
| 196 |
+
"logits/chosen": -2.2227556705474854,
|
| 197 |
+
"logits/rejected": -2.1177265644073486,
|
| 198 |
+
"logps/chosen": -341.78594970703125,
|
| 199 |
+
"logps/rejected": -359.0604553222656,
|
| 200 |
+
"loss": 0.6048,
|
| 201 |
+
"rewards/accuracies": 0.6875,
|
| 202 |
+
"rewards/chosen": -0.474813312292099,
|
| 203 |
+
"rewards/margins": 0.33163630962371826,
|
| 204 |
+
"rewards/rejected": -0.8064495921134949,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.13,
|
| 209 |
+
"grad_norm": 5.841735492614658,
|
| 210 |
+
"learning_rate": 4.9903757462135984e-06,
|
| 211 |
+
"logits/chosen": -2.2709505558013916,
|
| 212 |
+
"logits/rejected": -2.1858668327331543,
|
| 213 |
+
"logps/chosen": -327.3430480957031,
|
| 214 |
+
"logps/rejected": -359.780029296875,
|
| 215 |
+
"loss": 0.5976,
|
| 216 |
+
"rewards/accuracies": 0.706250011920929,
|
| 217 |
+
"rewards/chosen": -0.7242799401283264,
|
| 218 |
+
"rewards/margins": 0.3823621869087219,
|
| 219 |
+
"rewards/rejected": -1.1066421270370483,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.14,
|
| 224 |
+
"grad_norm": 4.058093924122145,
|
| 225 |
+
"learning_rate": 4.980697142834315e-06,
|
| 226 |
+
"logits/chosen": -1.9997293949127197,
|
| 227 |
+
"logits/rejected": -1.8148103952407837,
|
| 228 |
+
"logps/chosen": -391.48834228515625,
|
| 229 |
+
"logps/rejected": -384.39288330078125,
|
| 230 |
+
"loss": 0.5576,
|
| 231 |
+
"rewards/accuracies": 0.737500011920929,
|
| 232 |
+
"rewards/chosen": -0.850256621837616,
|
| 233 |
+
"rewards/margins": 0.467766135931015,
|
| 234 |
+
"rewards/rejected": -1.3180228471755981,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.15,
|
| 239 |
+
"grad_norm": 4.759314589639897,
|
| 240 |
+
"learning_rate": 4.967700826904229e-06,
|
| 241 |
+
"logits/chosen": -1.3535711765289307,
|
| 242 |
+
"logits/rejected": -1.2067102193832397,
|
| 243 |
+
"logps/chosen": -352.2384948730469,
|
| 244 |
+
"logps/rejected": -376.5408020019531,
|
| 245 |
+
"loss": 0.5621,
|
| 246 |
+
"rewards/accuracies": 0.731249988079071,
|
| 247 |
+
"rewards/chosen": -0.9079573750495911,
|
| 248 |
+
"rewards/margins": 0.47648367285728455,
|
| 249 |
+
"rewards/rejected": -1.3844410181045532,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.16,
|
| 254 |
+
"grad_norm": 6.966946629400433,
|
| 255 |
+
"learning_rate": 4.951404179843963e-06,
|
| 256 |
+
"logits/chosen": -1.4944788217544556,
|
| 257 |
+
"logits/rejected": -1.175698161125183,
|
| 258 |
+
"logps/chosen": -385.9867858886719,
|
| 259 |
+
"logps/rejected": -395.92889404296875,
|
| 260 |
+
"loss": 0.5526,
|
| 261 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 262 |
+
"rewards/chosen": -0.8049098253250122,
|
| 263 |
+
"rewards/margins": 0.6389679908752441,
|
| 264 |
+
"rewards/rejected": -1.443877935409546,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.17,
|
| 269 |
+
"grad_norm": 6.933994816745419,
|
| 270 |
+
"learning_rate": 4.931828996974498e-06,
|
| 271 |
+
"logits/chosen": -1.2824828624725342,
|
| 272 |
+
"logits/rejected": -1.0046826601028442,
|
| 273 |
+
"logps/chosen": -356.4523010253906,
|
| 274 |
+
"logps/rejected": -377.0785217285156,
|
| 275 |
+
"loss": 0.577,
|
| 276 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 277 |
+
"rewards/chosen": -1.0120383501052856,
|
| 278 |
+
"rewards/margins": 0.5240768790245056,
|
| 279 |
+
"rewards/rejected": -1.5361151695251465,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.18,
|
| 284 |
+
"grad_norm": 4.768436190577922,
|
| 285 |
+
"learning_rate": 4.909001458367867e-06,
|
| 286 |
+
"logits/chosen": -0.5933430790901184,
|
| 287 |
+
"logits/rejected": -0.33633238077163696,
|
| 288 |
+
"logps/chosen": -465.66510009765625,
|
| 289 |
+
"logps/rejected": -474.8177185058594,
|
| 290 |
+
"loss": 0.5755,
|
| 291 |
+
"rewards/accuracies": 0.731249988079071,
|
| 292 |
+
"rewards/chosen": -1.7751672267913818,
|
| 293 |
+
"rewards/margins": 0.543445885181427,
|
| 294 |
+
"rewards/rejected": -2.318613052368164,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.19,
|
| 299 |
+
"grad_norm": 4.556050170367665,
|
| 300 |
+
"learning_rate": 4.882952093833628e-06,
|
| 301 |
+
"logits/chosen": -0.9897077679634094,
|
| 302 |
+
"logits/rejected": -0.833814263343811,
|
| 303 |
+
"logps/chosen": -451.8089294433594,
|
| 304 |
+
"logps/rejected": -465.73565673828125,
|
| 305 |
+
"loss": 0.5727,
|
| 306 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 307 |
+
"rewards/chosen": -1.6387646198272705,
|
| 308 |
+
"rewards/margins": 0.44902724027633667,
|
| 309 |
+
"rewards/rejected": -2.087791919708252,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.2,
|
| 314 |
+
"grad_norm": 4.891304529530987,
|
| 315 |
+
"learning_rate": 4.853715742087947e-06,
|
| 316 |
+
"logits/chosen": -0.743046760559082,
|
| 317 |
+
"logits/rejected": -0.4523673951625824,
|
| 318 |
+
"logps/chosen": -378.919189453125,
|
| 319 |
+
"logps/rejected": -391.9801025390625,
|
| 320 |
+
"loss": 0.5486,
|
| 321 |
+
"rewards/accuracies": 0.6875,
|
| 322 |
+
"rewards/chosen": -1.2883106470108032,
|
| 323 |
+
"rewards/margins": 0.43378907442092896,
|
| 324 |
+
"rewards/rejected": -1.7220996618270874,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.21,
|
| 329 |
+
"grad_norm": 5.38693225816445,
|
| 330 |
+
"learning_rate": 4.821331504159906e-06,
|
| 331 |
+
"logits/chosen": -0.3933953046798706,
|
| 332 |
+
"logits/rejected": -0.05709639936685562,
|
| 333 |
+
"logps/chosen": -357.71466064453125,
|
| 334 |
+
"logps/rejected": -383.194580078125,
|
| 335 |
+
"loss": 0.5463,
|
| 336 |
+
"rewards/accuracies": 0.731249988079071,
|
| 337 |
+
"rewards/chosen": -1.0744948387145996,
|
| 338 |
+
"rewards/margins": 0.5533859729766846,
|
| 339 |
+
"rewards/rejected": -1.6278808116912842,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.21,
|
| 344 |
+
"eval_logits/chosen": -0.2985667288303375,
|
| 345 |
+
"eval_logits/rejected": -0.04311899468302727,
|
| 346 |
+
"eval_logps/chosen": -380.1156921386719,
|
| 347 |
+
"eval_logps/rejected": -432.1412658691406,
|
| 348 |
+
"eval_loss": 0.550410807132721,
|
| 349 |
+
"eval_rewards/accuracies": 0.692460298538208,
|
| 350 |
+
"eval_rewards/chosen": -1.1262083053588867,
|
| 351 |
+
"eval_rewards/margins": 0.5715598464012146,
|
| 352 |
+
"eval_rewards/rejected": -1.697768211364746,
|
| 353 |
+
"eval_runtime": 363.8367,
|
| 354 |
+
"eval_samples_per_second": 5.497,
|
| 355 |
+
"eval_steps_per_second": 0.173,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.22,
|
| 360 |
+
"grad_norm": 5.527241951911523,
|
| 361 |
+
"learning_rate": 4.7858426910973435e-06,
|
| 362 |
+
"logits/chosen": -0.14632602035999298,
|
| 363 |
+
"logits/rejected": 0.03931695967912674,
|
| 364 |
+
"logps/chosen": -388.56036376953125,
|
| 365 |
+
"logps/rejected": -445.367919921875,
|
| 366 |
+
"loss": 0.5336,
|
| 367 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 368 |
+
"rewards/chosen": -1.1834015846252441,
|
| 369 |
+
"rewards/margins": 0.731239914894104,
|
| 370 |
+
"rewards/rejected": -1.9146416187286377,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.23,
|
| 375 |
+
"grad_norm": 6.273093626258007,
|
| 376 |
+
"learning_rate": 4.747296766042161e-06,
|
| 377 |
+
"logits/chosen": 0.4761236310005188,
|
| 378 |
+
"logits/rejected": 0.6708475351333618,
|
| 379 |
+
"logps/chosen": -378.1613464355469,
|
| 380 |
+
"logps/rejected": -428.78643798828125,
|
| 381 |
+
"loss": 0.5169,
|
| 382 |
+
"rewards/accuracies": 0.6875,
|
| 383 |
+
"rewards/chosen": -1.4337002038955688,
|
| 384 |
+
"rewards/margins": 0.7104736566543579,
|
| 385 |
+
"rewards/rejected": -2.1441738605499268,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.24,
|
| 390 |
+
"grad_norm": 4.666490596901144,
|
| 391 |
+
"learning_rate": 4.705745280752586e-06,
|
| 392 |
+
"logits/chosen": -0.45128726959228516,
|
| 393 |
+
"logits/rejected": 0.05878635495901108,
|
| 394 |
+
"logps/chosen": -386.0497131347656,
|
| 395 |
+
"logps/rejected": -419.9007873535156,
|
| 396 |
+
"loss": 0.5351,
|
| 397 |
+
"rewards/accuracies": 0.762499988079071,
|
| 398 |
+
"rewards/chosen": -1.0124726295471191,
|
| 399 |
+
"rewards/margins": 0.6896177530288696,
|
| 400 |
+
"rewards/rejected": -1.7020905017852783,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.25,
|
| 405 |
+
"grad_norm": 6.120721061963155,
|
| 406 |
+
"learning_rate": 4.661243806657256e-06,
|
| 407 |
+
"logits/chosen": -0.053895045071840286,
|
| 408 |
+
"logits/rejected": 0.2541065514087677,
|
| 409 |
+
"logps/chosen": -380.99114990234375,
|
| 410 |
+
"logps/rejected": -433.10809326171875,
|
| 411 |
+
"loss": 0.5416,
|
| 412 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 413 |
+
"rewards/chosen": -1.1518728733062744,
|
| 414 |
+
"rewards/margins": 0.6217032074928284,
|
| 415 |
+
"rewards/rejected": -1.7735761404037476,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.26,
|
| 420 |
+
"grad_norm": 5.583267704971781,
|
| 421 |
+
"learning_rate": 4.613851860533367e-06,
|
| 422 |
+
"logits/chosen": 0.5581181049346924,
|
| 423 |
+
"logits/rejected": 0.7085031867027283,
|
| 424 |
+
"logps/chosen": -440.7920837402344,
|
| 425 |
+
"logps/rejected": -470.03778076171875,
|
| 426 |
+
"loss": 0.5495,
|
| 427 |
+
"rewards/accuracies": 0.731249988079071,
|
| 428 |
+
"rewards/chosen": -1.6821489334106445,
|
| 429 |
+
"rewards/margins": 0.609348475933075,
|
| 430 |
+
"rewards/rejected": -2.291497230529785,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.27,
|
| 435 |
+
"grad_norm": 5.673344697563603,
|
| 436 |
+
"learning_rate": 4.563632824908252e-06,
|
| 437 |
+
"logits/chosen": 0.7213119864463806,
|
| 438 |
+
"logits/rejected": 0.8342201113700867,
|
| 439 |
+
"logps/chosen": -413.45794677734375,
|
| 440 |
+
"logps/rejected": -476.8807067871094,
|
| 441 |
+
"loss": 0.5347,
|
| 442 |
+
"rewards/accuracies": 0.706250011920929,
|
| 443 |
+
"rewards/chosen": -1.6307703256607056,
|
| 444 |
+
"rewards/margins": 0.5706671476364136,
|
| 445 |
+
"rewards/rejected": -2.201437473297119,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.28,
|
| 450 |
+
"grad_norm": 7.321550555658446,
|
| 451 |
+
"learning_rate": 4.510653863290871e-06,
|
| 452 |
+
"logits/chosen": 0.22780194878578186,
|
| 453 |
+
"logits/rejected": 0.8186622858047485,
|
| 454 |
+
"logps/chosen": -407.9789123535156,
|
| 455 |
+
"logps/rejected": -441.42291259765625,
|
| 456 |
+
"loss": 0.5285,
|
| 457 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 458 |
+
"rewards/chosen": -1.2799824476242065,
|
| 459 |
+
"rewards/margins": 0.812950611114502,
|
| 460 |
+
"rewards/rejected": -2.092933177947998,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.29,
|
| 465 |
+
"grad_norm": 6.59015517751076,
|
| 466 |
+
"learning_rate": 4.454985830346574e-06,
|
| 467 |
+
"logits/chosen": 0.8685849905014038,
|
| 468 |
+
"logits/rejected": 1.2805653810501099,
|
| 469 |
+
"logps/chosen": -383.39483642578125,
|
| 470 |
+
"logps/rejected": -461.10528564453125,
|
| 471 |
+
"loss": 0.4941,
|
| 472 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 473 |
+
"rewards/chosen": -1.3261792659759521,
|
| 474 |
+
"rewards/margins": 0.7511085271835327,
|
| 475 |
+
"rewards/rejected": -2.0772879123687744,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.3,
|
| 480 |
+
"grad_norm": 5.290937758978531,
|
| 481 |
+
"learning_rate": 4.396703177135262e-06,
|
| 482 |
+
"logits/chosen": 0.7316430807113647,
|
| 483 |
+
"logits/rejected": 1.184604525566101,
|
| 484 |
+
"logps/chosen": -394.62908935546875,
|
| 485 |
+
"logps/rejected": -475.59637451171875,
|
| 486 |
+
"loss": 0.5394,
|
| 487 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 488 |
+
"rewards/chosen": -1.2810670137405396,
|
| 489 |
+
"rewards/margins": 0.8331844210624695,
|
| 490 |
+
"rewards/rejected": -2.1142513751983643,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.31,
|
| 495 |
+
"grad_norm": 6.38047854311955,
|
| 496 |
+
"learning_rate": 4.335883851539693e-06,
|
| 497 |
+
"logits/chosen": -0.5117667317390442,
|
| 498 |
+
"logits/rejected": -0.2342231273651123,
|
| 499 |
+
"logps/chosen": -363.1314697265625,
|
| 500 |
+
"logps/rejected": -406.3230285644531,
|
| 501 |
+
"loss": 0.4949,
|
| 502 |
+
"rewards/accuracies": 0.731249988079071,
|
| 503 |
+
"rewards/chosen": -1.074188232421875,
|
| 504 |
+
"rewards/margins": 0.7227994799613953,
|
| 505 |
+
"rewards/rejected": -1.796987771987915,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.31,
|
| 510 |
+
"eval_logits/chosen": 0.14181683957576752,
|
| 511 |
+
"eval_logits/rejected": 0.4033640921115875,
|
| 512 |
+
"eval_logps/chosen": -432.8479309082031,
|
| 513 |
+
"eval_logps/rejected": -505.6621398925781,
|
| 514 |
+
"eval_loss": 0.5161137580871582,
|
| 515 |
+
"eval_rewards/accuracies": 0.7182539701461792,
|
| 516 |
+
"eval_rewards/chosen": -1.6535308361053467,
|
| 517 |
+
"eval_rewards/margins": 0.7794461250305176,
|
| 518 |
+
"eval_rewards/rejected": -2.4329771995544434,
|
| 519 |
+
"eval_runtime": 363.6104,
|
| 520 |
+
"eval_samples_per_second": 5.5,
|
| 521 |
+
"eval_steps_per_second": 0.173,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.32,
|
| 526 |
+
"grad_norm": 6.269453976354009,
|
| 527 |
+
"learning_rate": 4.2726091940171055e-06,
|
| 528 |
+
"logits/chosen": 0.015387272462248802,
|
| 529 |
+
"logits/rejected": 0.4361787736415863,
|
| 530 |
+
"logps/chosen": -471.3013610839844,
|
| 531 |
+
"logps/rejected": -514.4930419921875,
|
| 532 |
+
"loss": 0.5181,
|
| 533 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 534 |
+
"rewards/chosen": -1.809822678565979,
|
| 535 |
+
"rewards/margins": 0.7652148604393005,
|
| 536 |
+
"rewards/rejected": -2.5750374794006348,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.33,
|
| 541 |
+
"grad_norm": 7.128031303328463,
|
| 542 |
+
"learning_rate": 4.206963828813555e-06,
|
| 543 |
+
"logits/chosen": -0.014968380331993103,
|
| 544 |
+
"logits/rejected": 0.18288832902908325,
|
| 545 |
+
"logps/chosen": -377.1259765625,
|
| 546 |
+
"logps/rejected": -439.23504638671875,
|
| 547 |
+
"loss": 0.5033,
|
| 548 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 549 |
+
"rewards/chosen": -1.4144322872161865,
|
| 550 |
+
"rewards/margins": 0.7219018936157227,
|
| 551 |
+
"rewards/rejected": -2.136334180831909,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.35,
|
| 556 |
+
"grad_norm": 6.507915372367444,
|
| 557 |
+
"learning_rate": 4.139035550786495e-06,
|
| 558 |
+
"logits/chosen": 0.25516384840011597,
|
| 559 |
+
"logits/rejected": 0.32111260294914246,
|
| 560 |
+
"logps/chosen": -384.0611877441406,
|
| 561 |
+
"logps/rejected": -484.21826171875,
|
| 562 |
+
"loss": 0.5141,
|
| 563 |
+
"rewards/accuracies": 0.768750011920929,
|
| 564 |
+
"rewards/chosen": -1.6149778366088867,
|
| 565 |
+
"rewards/margins": 0.7714108824729919,
|
| 566 |
+
"rewards/rejected": -2.3863887786865234,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.36,
|
| 571 |
+
"grad_norm": 7.117891747882256,
|
| 572 |
+
"learning_rate": 4.068915207986931e-06,
|
| 573 |
+
"logits/chosen": 0.793425977230072,
|
| 574 |
+
"logits/rejected": 1.1012945175170898,
|
| 575 |
+
"logps/chosen": -490.26702880859375,
|
| 576 |
+
"logps/rejected": -559.7706298828125,
|
| 577 |
+
"loss": 0.5288,
|
| 578 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 579 |
+
"rewards/chosen": -2.188723564147949,
|
| 580 |
+
"rewards/margins": 0.789432168006897,
|
| 581 |
+
"rewards/rejected": -2.9781556129455566,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.37,
|
| 586 |
+
"grad_norm": 6.767573615618988,
|
| 587 |
+
"learning_rate": 3.996696580158211e-06,
|
| 588 |
+
"logits/chosen": 0.038734257221221924,
|
| 589 |
+
"logits/rejected": 0.3444157540798187,
|
| 590 |
+
"logps/chosen": -393.08392333984375,
|
| 591 |
+
"logps/rejected": -440.9234924316406,
|
| 592 |
+
"loss": 0.543,
|
| 593 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 594 |
+
"rewards/chosen": -1.2908889055252075,
|
| 595 |
+
"rewards/margins": 0.7114801406860352,
|
| 596 |
+
"rewards/rejected": -2.0023691654205322,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.38,
|
| 601 |
+
"grad_norm": 5.745989852370514,
|
| 602 |
+
"learning_rate": 3.922476253313921e-06,
|
| 603 |
+
"logits/chosen": -0.3795308768749237,
|
| 604 |
+
"logits/rejected": -0.08608667552471161,
|
| 605 |
+
"logps/chosen": -352.70037841796875,
|
| 606 |
+
"logps/rejected": -411.849609375,
|
| 607 |
+
"loss": 0.5253,
|
| 608 |
+
"rewards/accuracies": 0.762499988079071,
|
| 609 |
+
"rewards/chosen": -1.2477184534072876,
|
| 610 |
+
"rewards/margins": 0.7360815405845642,
|
| 611 |
+
"rewards/rejected": -1.983799695968628,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.39,
|
| 616 |
+
"grad_norm": 7.421866262330383,
|
| 617 |
+
"learning_rate": 3.846353490562664e-06,
|
| 618 |
+
"logits/chosen": 0.32861948013305664,
|
| 619 |
+
"logits/rejected": 0.747407078742981,
|
| 620 |
+
"logps/chosen": -419.4422912597656,
|
| 621 |
+
"logps/rejected": -482.92218017578125,
|
| 622 |
+
"loss": 0.5213,
|
| 623 |
+
"rewards/accuracies": 0.768750011920929,
|
| 624 |
+
"rewards/chosen": -1.5996549129486084,
|
| 625 |
+
"rewards/margins": 0.861799418926239,
|
| 626 |
+
"rewards/rejected": -2.4614546298980713,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.4,
|
| 631 |
+
"grad_norm": 8.881319616744216,
|
| 632 |
+
"learning_rate": 3.768430099352445e-06,
|
| 633 |
+
"logits/chosen": 0.7479051351547241,
|
| 634 |
+
"logits/rejected": 0.9084765315055847,
|
| 635 |
+
"logps/chosen": -419.5804138183594,
|
| 636 |
+
"logps/rejected": -462.1929626464844,
|
| 637 |
+
"loss": 0.5316,
|
| 638 |
+
"rewards/accuracies": 0.637499988079071,
|
| 639 |
+
"rewards/chosen": -1.8022568225860596,
|
| 640 |
+
"rewards/margins": 0.5997826457023621,
|
| 641 |
+
"rewards/rejected": -2.4020392894744873,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.41,
|
| 646 |
+
"grad_norm": 6.8157951015425144,
|
| 647 |
+
"learning_rate": 3.6888102953122307e-06,
|
| 648 |
+
"logits/chosen": 0.3040899932384491,
|
| 649 |
+
"logits/rejected": 0.7399716973304749,
|
| 650 |
+
"logps/chosen": -431.64471435546875,
|
| 651 |
+
"logps/rejected": -468.09228515625,
|
| 652 |
+
"loss": 0.5349,
|
| 653 |
+
"rewards/accuracies": 0.768750011920929,
|
| 654 |
+
"rewards/chosen": -1.6253465414047241,
|
| 655 |
+
"rewards/margins": 0.7721388339996338,
|
| 656 |
+
"rewards/rejected": -2.3974852561950684,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.42,
|
| 661 |
+
"grad_norm": 7.719819400236357,
|
| 662 |
+
"learning_rate": 3.607600562872785e-06,
|
| 663 |
+
"logits/chosen": 0.025150710716843605,
|
| 664 |
+
"logits/rejected": 0.44670620560646057,
|
| 665 |
+
"logps/chosen": -389.49591064453125,
|
| 666 |
+
"logps/rejected": -436.03485107421875,
|
| 667 |
+
"loss": 0.5239,
|
| 668 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 669 |
+
"rewards/chosen": -1.289335012435913,
|
| 670 |
+
"rewards/margins": 0.5783864259719849,
|
| 671 |
+
"rewards/rejected": -1.8677213191986084,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.42,
|
| 676 |
+
"eval_logits/chosen": 0.5590775609016418,
|
| 677 |
+
"eval_logits/rejected": 0.8585302233695984,
|
| 678 |
+
"eval_logps/chosen": -404.4281921386719,
|
| 679 |
+
"eval_logps/rejected": -470.4624328613281,
|
| 680 |
+
"eval_loss": 0.5100502371788025,
|
| 681 |
+
"eval_rewards/accuracies": 0.7301587462425232,
|
| 682 |
+
"eval_rewards/chosen": -1.3693335056304932,
|
| 683 |
+
"eval_rewards/margins": 0.7116466164588928,
|
| 684 |
+
"eval_rewards/rejected": -2.080980062484741,
|
| 685 |
+
"eval_runtime": 363.5138,
|
| 686 |
+
"eval_samples_per_second": 5.502,
|
| 687 |
+
"eval_steps_per_second": 0.173,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.43,
|
| 692 |
+
"grad_norm": 7.970937869843318,
|
| 693 |
+
"learning_rate": 3.5249095128531863e-06,
|
| 694 |
+
"logits/chosen": 0.4714787006378174,
|
| 695 |
+
"logits/rejected": 0.9316293597221375,
|
| 696 |
+
"logps/chosen": -396.480224609375,
|
| 697 |
+
"logps/rejected": -434.74969482421875,
|
| 698 |
+
"loss": 0.5072,
|
| 699 |
+
"rewards/accuracies": 0.731249988079071,
|
| 700 |
+
"rewards/chosen": -1.2913535833358765,
|
| 701 |
+
"rewards/margins": 0.7052849531173706,
|
| 702 |
+
"rewards/rejected": -1.996638536453247,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.44,
|
| 707 |
+
"grad_norm": 6.303939907368478,
|
| 708 |
+
"learning_rate": 3.4408477372034743e-06,
|
| 709 |
+
"logits/chosen": 0.2637383043766022,
|
| 710 |
+
"logits/rejected": 0.8194657564163208,
|
| 711 |
+
"logps/chosen": -425.71044921875,
|
| 712 |
+
"logps/rejected": -472.1114807128906,
|
| 713 |
+
"loss": 0.506,
|
| 714 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 715 |
+
"rewards/chosen": -1.4857378005981445,
|
| 716 |
+
"rewards/margins": 0.8204676508903503,
|
| 717 |
+
"rewards/rejected": -2.3062055110931396,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.45,
|
| 722 |
+
"grad_norm": 4.180610112833693,
|
| 723 |
+
"learning_rate": 3.355527661097728e-06,
|
| 724 |
+
"logits/chosen": 0.5680927634239197,
|
| 725 |
+
"logits/rejected": 1.104353904724121,
|
| 726 |
+
"logps/chosen": -410.1541442871094,
|
| 727 |
+
"logps/rejected": -459.4964904785156,
|
| 728 |
+
"loss": 0.509,
|
| 729 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 730 |
+
"rewards/chosen": -1.484910249710083,
|
| 731 |
+
"rewards/margins": 0.6827523112297058,
|
| 732 |
+
"rewards/rejected": -2.1676626205444336,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.46,
|
| 737 |
+
"grad_norm": 8.034683384238669,
|
| 738 |
+
"learning_rate": 3.269063392575352e-06,
|
| 739 |
+
"logits/chosen": 0.9522289037704468,
|
| 740 |
+
"logits/rejected": 1.3713102340698242,
|
| 741 |
+
"logps/chosen": -410.77386474609375,
|
| 742 |
+
"logps/rejected": -469.934814453125,
|
| 743 |
+
"loss": 0.4919,
|
| 744 |
+
"rewards/accuracies": 0.731249988079071,
|
| 745 |
+
"rewards/chosen": -1.4788429737091064,
|
| 746 |
+
"rewards/margins": 0.84101402759552,
|
| 747 |
+
"rewards/rejected": -2.319856882095337,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.47,
|
| 752 |
+
"grad_norm": 7.933204029001868,
|
| 753 |
+
"learning_rate": 3.181570569931697e-06,
|
| 754 |
+
"logits/chosen": 1.7222553491592407,
|
| 755 |
+
"logits/rejected": 2.531193256378174,
|
| 756 |
+
"logps/chosen": -468.93035888671875,
|
| 757 |
+
"logps/rejected": -518.9406127929688,
|
| 758 |
+
"loss": 0.5071,
|
| 759 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 760 |
+
"rewards/chosen": -1.97989022731781,
|
| 761 |
+
"rewards/margins": 0.8815497159957886,
|
| 762 |
+
"rewards/rejected": -2.8614401817321777,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.48,
|
| 767 |
+
"grad_norm": 6.136879125450544,
|
| 768 |
+
"learning_rate": 3.09316620706208e-06,
|
| 769 |
+
"logits/chosen": 1.2675981521606445,
|
| 770 |
+
"logits/rejected": 1.6471933126449585,
|
| 771 |
+
"logps/chosen": -471.82501220703125,
|
| 772 |
+
"logps/rejected": -539.4986572265625,
|
| 773 |
+
"loss": 0.4972,
|
| 774 |
+
"rewards/accuracies": 0.71875,
|
| 775 |
+
"rewards/chosen": -1.9156773090362549,
|
| 776 |
+
"rewards/margins": 0.8640796542167664,
|
| 777 |
+
"rewards/rejected": -2.779757022857666,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.49,
|
| 782 |
+
"grad_norm": 6.202648384751733,
|
| 783 |
+
"learning_rate": 3.0039685369660785e-06,
|
| 784 |
+
"logits/chosen": 0.8523815274238586,
|
| 785 |
+
"logits/rejected": 1.4816639423370361,
|
| 786 |
+
"logps/chosen": -424.09759521484375,
|
| 787 |
+
"logps/rejected": -498.6441955566406,
|
| 788 |
+
"loss": 0.4932,
|
| 789 |
+
"rewards/accuracies": 0.737500011920929,
|
| 790 |
+
"rewards/chosen": -1.6259464025497437,
|
| 791 |
+
"rewards/margins": 0.9263278245925903,
|
| 792 |
+
"rewards/rejected": -2.552274227142334,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.5,
|
| 797 |
+
"grad_norm": 5.934005833493391,
|
| 798 |
+
"learning_rate": 2.91409685362137e-06,
|
| 799 |
+
"logits/chosen": 0.7408124208450317,
|
| 800 |
+
"logits/rejected": 1.0110493898391724,
|
| 801 |
+
"logps/chosen": -438.82781982421875,
|
| 802 |
+
"logps/rejected": -494.5777282714844,
|
| 803 |
+
"loss": 0.5068,
|
| 804 |
+
"rewards/accuracies": 0.675000011920929,
|
| 805 |
+
"rewards/chosen": -1.702741265296936,
|
| 806 |
+
"rewards/margins": 0.7638914585113525,
|
| 807 |
+
"rewards/rejected": -2.466632843017578,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.51,
|
| 812 |
+
"grad_norm": 6.134335934888188,
|
| 813 |
+
"learning_rate": 2.8236713524386085e-06,
|
| 814 |
+
"logits/chosen": 0.4550415873527527,
|
| 815 |
+
"logits/rejected": 0.7779414057731628,
|
| 816 |
+
"logps/chosen": -434.3086853027344,
|
| 817 |
+
"logps/rejected": -516.2806396484375,
|
| 818 |
+
"loss": 0.5089,
|
| 819 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 820 |
+
"rewards/chosen": -1.7570041418075562,
|
| 821 |
+
"rewards/margins": 0.8707243204116821,
|
| 822 |
+
"rewards/rejected": -2.6277284622192383,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.52,
|
| 827 |
+
"grad_norm": 7.785644515938578,
|
| 828 |
+
"learning_rate": 2.7328129695107205e-06,
|
| 829 |
+
"logits/chosen": 0.7499665021896362,
|
| 830 |
+
"logits/rejected": 1.3003952503204346,
|
| 831 |
+
"logps/chosen": -424.2498474121094,
|
| 832 |
+
"logps/rejected": -495.997802734375,
|
| 833 |
+
"loss": 0.5272,
|
| 834 |
+
"rewards/accuracies": 0.71875,
|
| 835 |
+
"rewards/chosen": -1.920672059059143,
|
| 836 |
+
"rewards/margins": 0.8888359069824219,
|
| 837 |
+
"rewards/rejected": -2.8095080852508545,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.52,
|
| 842 |
+
"eval_logits/chosen": 0.7149888277053833,
|
| 843 |
+
"eval_logits/rejected": 1.0403735637664795,
|
| 844 |
+
"eval_logps/chosen": -471.0703430175781,
|
| 845 |
+
"eval_logps/rejected": -558.6534423828125,
|
| 846 |
+
"eval_loss": 0.5003111362457275,
|
| 847 |
+
"eval_rewards/accuracies": 0.738095223903656,
|
| 848 |
+
"eval_rewards/chosen": -2.035755157470703,
|
| 849 |
+
"eval_rewards/margins": 0.9271355867385864,
|
| 850 |
+
"eval_rewards/rejected": -2.962890863418579,
|
| 851 |
+
"eval_runtime": 363.4067,
|
| 852 |
+
"eval_samples_per_second": 5.503,
|
| 853 |
+
"eval_steps_per_second": 0.173,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.53,
|
| 858 |
+
"grad_norm": 6.508475133675616,
|
| 859 |
+
"learning_rate": 2.641643219871597e-06,
|
| 860 |
+
"logits/chosen": 0.5101484060287476,
|
| 861 |
+
"logits/rejected": 0.9493732452392578,
|
| 862 |
+
"logps/chosen": -467.1604919433594,
|
| 863 |
+
"logps/rejected": -551.2037353515625,
|
| 864 |
+
"loss": 0.4865,
|
| 865 |
+
"rewards/accuracies": 0.75,
|
| 866 |
+
"rewards/chosen": -1.970902442932129,
|
| 867 |
+
"rewards/margins": 0.9815346002578735,
|
| 868 |
+
"rewards/rejected": -2.952437162399292,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.54,
|
| 873 |
+
"grad_norm": 5.49058038197313,
|
| 874 |
+
"learning_rate": 2.5502840349805074e-06,
|
| 875 |
+
"logits/chosen": 0.1692526787519455,
|
| 876 |
+
"logits/rejected": 0.5913276076316833,
|
| 877 |
+
"logps/chosen": -398.19561767578125,
|
| 878 |
+
"logps/rejected": -454.1546936035156,
|
| 879 |
+
"loss": 0.5012,
|
| 880 |
+
"rewards/accuracies": 0.75,
|
| 881 |
+
"rewards/chosen": -1.5394331216812134,
|
| 882 |
+
"rewards/margins": 0.9616721868515015,
|
| 883 |
+
"rewards/rejected": -2.501105546951294,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.55,
|
| 888 |
+
"grad_norm": 6.551673605459196,
|
| 889 |
+
"learning_rate": 2.4588575996495797e-06,
|
| 890 |
+
"logits/chosen": -0.2059365212917328,
|
| 891 |
+
"logits/rejected": 0.3154928684234619,
|
| 892 |
+
"logps/chosen": -433.8434143066406,
|
| 893 |
+
"logps/rejected": -497.22705078125,
|
| 894 |
+
"loss": 0.4805,
|
| 895 |
+
"rewards/accuracies": 0.824999988079071,
|
| 896 |
+
"rewards/chosen": -1.5971723794937134,
|
| 897 |
+
"rewards/margins": 1.0449589490890503,
|
| 898 |
+
"rewards/rejected": -2.6421313285827637,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.57,
|
| 903 |
+
"grad_norm": 7.008080314698571,
|
| 904 |
+
"learning_rate": 2.367486188632446e-06,
|
| 905 |
+
"logits/chosen": 0.24021968245506287,
|
| 906 |
+
"logits/rejected": 0.4527043402194977,
|
| 907 |
+
"logps/chosen": -493.78436279296875,
|
| 908 |
+
"logps/rejected": -600.9166259765625,
|
| 909 |
+
"loss": 0.4979,
|
| 910 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 911 |
+
"rewards/chosen": -2.1465725898742676,
|
| 912 |
+
"rewards/margins": 1.0573832988739014,
|
| 913 |
+
"rewards/rejected": -3.203955888748169,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.58,
|
| 918 |
+
"grad_norm": 7.130638434676378,
|
| 919 |
+
"learning_rate": 2.276292003092593e-06,
|
| 920 |
+
"logits/chosen": 0.10954128205776215,
|
| 921 |
+
"logits/rejected": 0.5284096002578735,
|
| 922 |
+
"logps/chosen": -523.6838989257812,
|
| 923 |
+
"logps/rejected": -586.87451171875,
|
| 924 |
+
"loss": 0.4699,
|
| 925 |
+
"rewards/accuracies": 0.737500011920929,
|
| 926 |
+
"rewards/chosen": -2.2369279861450195,
|
| 927 |
+
"rewards/margins": 0.9251989126205444,
|
| 928 |
+
"rewards/rejected": -3.1621270179748535,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.59,
|
| 933 |
+
"grad_norm": 6.527520826256094,
|
| 934 |
+
"learning_rate": 2.1853970071701415e-06,
|
| 935 |
+
"logits/chosen": 0.1212792843580246,
|
| 936 |
+
"logits/rejected": 0.621984601020813,
|
| 937 |
+
"logps/chosen": -492.78326416015625,
|
| 938 |
+
"logps/rejected": -556.6223754882812,
|
| 939 |
+
"loss": 0.4888,
|
| 940 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 941 |
+
"rewards/chosen": -1.9754877090454102,
|
| 942 |
+
"rewards/margins": 0.8828755617141724,
|
| 943 |
+
"rewards/rejected": -2.858363389968872,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.6,
|
| 948 |
+
"grad_norm": 7.521550053676498,
|
| 949 |
+
"learning_rate": 2.0949227648656194e-06,
|
| 950 |
+
"logits/chosen": 0.6495383977890015,
|
| 951 |
+
"logits/rejected": 0.7199844121932983,
|
| 952 |
+
"logps/chosen": -420.24951171875,
|
| 953 |
+
"logps/rejected": -494.35516357421875,
|
| 954 |
+
"loss": 0.5204,
|
| 955 |
+
"rewards/accuracies": 0.6875,
|
| 956 |
+
"rewards/chosen": -1.9334577322006226,
|
| 957 |
+
"rewards/margins": 0.878303050994873,
|
| 958 |
+
"rewards/rejected": -2.8117611408233643,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.61,
|
| 963 |
+
"grad_norm": 5.995967781657259,
|
| 964 |
+
"learning_rate": 2.00499027745888e-06,
|
| 965 |
+
"logits/chosen": -0.02193205989897251,
|
| 966 |
+
"logits/rejected": 0.2604701519012451,
|
| 967 |
+
"logps/chosen": -426.2652893066406,
|
| 968 |
+
"logps/rejected": -514.4281005859375,
|
| 969 |
+
"loss": 0.4959,
|
| 970 |
+
"rewards/accuracies": 0.731249988079071,
|
| 971 |
+
"rewards/chosen": -1.6685237884521484,
|
| 972 |
+
"rewards/margins": 0.8125013113021851,
|
| 973 |
+
"rewards/rejected": -2.481024980545044,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.62,
|
| 978 |
+
"grad_norm": 7.427361767129983,
|
| 979 |
+
"learning_rate": 1.915719821680624e-06,
|
| 980 |
+
"logits/chosen": -0.14169612526893616,
|
| 981 |
+
"logits/rejected": 0.3787733018398285,
|
| 982 |
+
"logps/chosen": -397.3866882324219,
|
| 983 |
+
"logps/rejected": -437.279541015625,
|
| 984 |
+
"loss": 0.5169,
|
| 985 |
+
"rewards/accuracies": 0.706250011920929,
|
| 986 |
+
"rewards/chosen": -1.5220028162002563,
|
| 987 |
+
"rewards/margins": 0.7406032085418701,
|
| 988 |
+
"rewards/rejected": -2.262606143951416,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.63,
|
| 993 |
+
"grad_norm": 6.908180760937083,
|
| 994 |
+
"learning_rate": 1.8272307888529276e-06,
|
| 995 |
+
"logits/chosen": 0.0949440747499466,
|
| 996 |
+
"logits/rejected": 0.2515867352485657,
|
| 997 |
+
"logps/chosen": -374.9373474121094,
|
| 998 |
+
"logps/rejected": -469.41143798828125,
|
| 999 |
+
"loss": 0.4886,
|
| 1000 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1001 |
+
"rewards/chosen": -1.6092283725738525,
|
| 1002 |
+
"rewards/margins": 0.7828912138938904,
|
| 1003 |
+
"rewards/rejected": -2.3921194076538086,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.63,
|
| 1008 |
+
"eval_logits/chosen": 0.05941150337457657,
|
| 1009 |
+
"eval_logits/rejected": 0.37521931529045105,
|
| 1010 |
+
"eval_logps/chosen": -444.8821716308594,
|
| 1011 |
+
"eval_logps/rejected": -526.641357421875,
|
| 1012 |
+
"eval_loss": 0.49817371368408203,
|
| 1013 |
+
"eval_rewards/accuracies": 0.726190447807312,
|
| 1014 |
+
"eval_rewards/chosen": -1.7738730907440186,
|
| 1015 |
+
"eval_rewards/margins": 0.8688962459564209,
|
| 1016 |
+
"eval_rewards/rejected": -2.6427693367004395,
|
| 1017 |
+
"eval_runtime": 363.5537,
|
| 1018 |
+
"eval_samples_per_second": 5.501,
|
| 1019 |
+
"eval_steps_per_second": 0.173,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.64,
|
| 1024 |
+
"grad_norm": 7.791648443136137,
|
| 1025 |
+
"learning_rate": 1.739641525213929e-06,
|
| 1026 |
+
"logits/chosen": 0.37017613649368286,
|
| 1027 |
+
"logits/rejected": 0.7658742666244507,
|
| 1028 |
+
"logps/chosen": -431.961181640625,
|
| 1029 |
+
"logps/rejected": -511.739501953125,
|
| 1030 |
+
"loss": 0.4684,
|
| 1031 |
+
"rewards/accuracies": 0.78125,
|
| 1032 |
+
"rewards/chosen": -1.889151930809021,
|
| 1033 |
+
"rewards/margins": 0.9400160908699036,
|
| 1034 |
+
"rewards/rejected": -2.8291683197021484,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.65,
|
| 1039 |
+
"grad_norm": 6.033378581233917,
|
| 1040 |
+
"learning_rate": 1.6530691736402317e-06,
|
| 1041 |
+
"logits/chosen": 0.5889648795127869,
|
| 1042 |
+
"logits/rejected": 0.6707448959350586,
|
| 1043 |
+
"logps/chosen": -445.545654296875,
|
| 1044 |
+
"logps/rejected": -556.2078857421875,
|
| 1045 |
+
"loss": 0.4698,
|
| 1046 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1047 |
+
"rewards/chosen": -2.1388087272644043,
|
| 1048 |
+
"rewards/margins": 0.829494297504425,
|
| 1049 |
+
"rewards/rejected": -2.9683032035827637,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.66,
|
| 1054 |
+
"grad_norm": 8.258841660679291,
|
| 1055 |
+
"learning_rate": 1.5676295169786864e-06,
|
| 1056 |
+
"logits/chosen": 0.6791388392448425,
|
| 1057 |
+
"logits/rejected": 1.2308781147003174,
|
| 1058 |
+
"logps/chosen": -498.2019958496094,
|
| 1059 |
+
"logps/rejected": -589.25927734375,
|
| 1060 |
+
"loss": 0.4736,
|
| 1061 |
+
"rewards/accuracies": 0.75,
|
| 1062 |
+
"rewards/chosen": -2.2767367362976074,
|
| 1063 |
+
"rewards/margins": 1.188873291015625,
|
| 1064 |
+
"rewards/rejected": -3.4656097888946533,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.67,
|
| 1069 |
+
"grad_norm": 7.552790690651825,
|
| 1070 |
+
"learning_rate": 1.4834368231970922e-06,
|
| 1071 |
+
"logits/chosen": 0.6634274125099182,
|
| 1072 |
+
"logits/rejected": 0.95360267162323,
|
| 1073 |
+
"logps/chosen": -443.93304443359375,
|
| 1074 |
+
"logps/rejected": -516.3823852539062,
|
| 1075 |
+
"loss": 0.495,
|
| 1076 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1077 |
+
"rewards/chosen": -2.053490161895752,
|
| 1078 |
+
"rewards/margins": 0.9448803663253784,
|
| 1079 |
+
"rewards/rejected": -2.99837064743042,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.68,
|
| 1084 |
+
"grad_norm": 6.240039976432098,
|
| 1085 |
+
"learning_rate": 1.4006036925609245e-06,
|
| 1086 |
+
"logits/chosen": 0.2837061583995819,
|
| 1087 |
+
"logits/rejected": 0.6308731436729431,
|
| 1088 |
+
"logps/chosen": -462.76824951171875,
|
| 1089 |
+
"logps/rejected": -545.1807250976562,
|
| 1090 |
+
"loss": 0.4953,
|
| 1091 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1092 |
+
"rewards/chosen": -1.862541913986206,
|
| 1093 |
+
"rewards/margins": 0.9856799840927124,
|
| 1094 |
+
"rewards/rejected": -2.848222017288208,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.69,
|
| 1099 |
+
"grad_norm": 6.068196245695217,
|
| 1100 |
+
"learning_rate": 1.3192409070404582e-06,
|
| 1101 |
+
"logits/chosen": -0.018947910517454147,
|
| 1102 |
+
"logits/rejected": 0.6848478317260742,
|
| 1103 |
+
"logps/chosen": -459.00946044921875,
|
| 1104 |
+
"logps/rejected": -509.8402404785156,
|
| 1105 |
+
"loss": 0.4779,
|
| 1106 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1107 |
+
"rewards/chosen": -1.6749271154403687,
|
| 1108 |
+
"rewards/margins": 0.9239859580993652,
|
| 1109 |
+
"rewards/rejected": -2.5989129543304443,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.7,
|
| 1114 |
+
"grad_norm": 7.317258150817362,
|
| 1115 |
+
"learning_rate": 1.2394572821496953e-06,
|
| 1116 |
+
"logits/chosen": 0.4193340837955475,
|
| 1117 |
+
"logits/rejected": 0.5491575598716736,
|
| 1118 |
+
"logps/chosen": -433.359375,
|
| 1119 |
+
"logps/rejected": -529.31201171875,
|
| 1120 |
+
"loss": 0.5121,
|
| 1121 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1122 |
+
"rewards/chosen": -2.034470796585083,
|
| 1123 |
+
"rewards/margins": 0.8596637845039368,
|
| 1124 |
+
"rewards/rejected": -2.894134521484375,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.71,
|
| 1129 |
+
"grad_norm": 6.984303996772759,
|
| 1130 |
+
"learning_rate": 1.1613595214152713e-06,
|
| 1131 |
+
"logits/chosen": 0.375018447637558,
|
| 1132 |
+
"logits/rejected": 1.3137071132659912,
|
| 1133 |
+
"logps/chosen": -503.49188232421875,
|
| 1134 |
+
"logps/rejected": -537.77392578125,
|
| 1135 |
+
"loss": 0.4789,
|
| 1136 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1137 |
+
"rewards/chosen": -2.075040817260742,
|
| 1138 |
+
"rewards/margins": 0.832379162311554,
|
| 1139 |
+
"rewards/rejected": -2.9074199199676514,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.72,
|
| 1144 |
+
"grad_norm": 6.449594635756279,
|
| 1145 |
+
"learning_rate": 1.0850520736699362e-06,
|
| 1146 |
+
"logits/chosen": 0.4090999662876129,
|
| 1147 |
+
"logits/rejected": 0.8342447280883789,
|
| 1148 |
+
"logps/chosen": -482.46832275390625,
|
| 1149 |
+
"logps/rejected": -552.4658203125,
|
| 1150 |
+
"loss": 0.5044,
|
| 1151 |
+
"rewards/accuracies": 0.71875,
|
| 1152 |
+
"rewards/chosen": -1.9981062412261963,
|
| 1153 |
+
"rewards/margins": 0.9065272212028503,
|
| 1154 |
+
"rewards/rejected": -2.9046332836151123,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.73,
|
| 1159 |
+
"grad_norm": 7.762156396899746,
|
| 1160 |
+
"learning_rate": 1.0106369933615043e-06,
|
| 1161 |
+
"logits/chosen": 0.44636210799217224,
|
| 1162 |
+
"logits/rejected": 0.6373854875564575,
|
| 1163 |
+
"logps/chosen": -459.1854553222656,
|
| 1164 |
+
"logps/rejected": -532.9047241210938,
|
| 1165 |
+
"loss": 0.516,
|
| 1166 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1167 |
+
"rewards/chosen": -2.135340452194214,
|
| 1168 |
+
"rewards/margins": 0.8491575121879578,
|
| 1169 |
+
"rewards/rejected": -2.9844982624053955,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.73,
|
| 1174 |
+
"eval_logits/chosen": 0.531207799911499,
|
| 1175 |
+
"eval_logits/rejected": 0.8897786736488342,
|
| 1176 |
+
"eval_logps/chosen": -469.92730712890625,
|
| 1177 |
+
"eval_logps/rejected": -556.2413330078125,
|
| 1178 |
+
"eval_loss": 0.49328893423080444,
|
| 1179 |
+
"eval_rewards/accuracies": 0.7301587462425232,
|
| 1180 |
+
"eval_rewards/chosen": -2.024324893951416,
|
| 1181 |
+
"eval_rewards/margins": 0.9144444465637207,
|
| 1182 |
+
"eval_rewards/rejected": -2.938769578933716,
|
| 1183 |
+
"eval_runtime": 363.5561,
|
| 1184 |
+
"eval_samples_per_second": 5.501,
|
| 1185 |
+
"eval_steps_per_second": 0.173,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.74,
|
| 1190 |
+
"grad_norm": 6.854213571436283,
|
| 1191 |
+
"learning_rate": 9.382138040640714e-07,
|
| 1192 |
+
"logits/chosen": 0.5431604981422424,
|
| 1193 |
+
"logits/rejected": 1.0465270280838013,
|
| 1194 |
+
"logps/chosen": -448.469970703125,
|
| 1195 |
+
"logps/rejected": -540.0692749023438,
|
| 1196 |
+
"loss": 0.4754,
|
| 1197 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1198 |
+
"rewards/chosen": -2.114922285079956,
|
| 1199 |
+
"rewards/margins": 0.9879555702209473,
|
| 1200 |
+
"rewards/rejected": -3.1028780937194824,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.75,
|
| 1205 |
+
"grad_norm": 5.476671160655268,
|
| 1206 |
+
"learning_rate": 8.678793653740633e-07,
|
| 1207 |
+
"logits/chosen": 0.13008640706539154,
|
| 1208 |
+
"logits/rejected": 0.6161590814590454,
|
| 1209 |
+
"logps/chosen": -476.4979553222656,
|
| 1210 |
+
"logps/rejected": -574.2578125,
|
| 1211 |
+
"loss": 0.4781,
|
| 1212 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1213 |
+
"rewards/chosen": -1.904258131980896,
|
| 1214 |
+
"rewards/margins": 1.040144681930542,
|
| 1215 |
+
"rewards/rejected": -2.9444031715393066,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.76,
|
| 1220 |
+
"grad_norm": 10.553613660329997,
|
| 1221 |
+
"learning_rate": 7.997277433690984e-07,
|
| 1222 |
+
"logits/chosen": 0.33887559175491333,
|
| 1223 |
+
"logits/rejected": 0.7562397718429565,
|
| 1224 |
+
"logps/chosen": -460.935302734375,
|
| 1225 |
+
"logps/rejected": -538.0240478515625,
|
| 1226 |
+
"loss": 0.4952,
|
| 1227 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1228 |
+
"rewards/chosen": -1.8783504962921143,
|
| 1229 |
+
"rewards/margins": 0.9283574819564819,
|
| 1230 |
+
"rewards/rejected": -2.8067078590393066,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 0.77,
|
| 1235 |
+
"grad_norm": 6.627806159178358,
|
| 1236 |
+
"learning_rate": 7.338500848029603e-07,
|
| 1237 |
+
"logits/chosen": 0.0666571706533432,
|
| 1238 |
+
"logits/rejected": 0.5915927290916443,
|
| 1239 |
+
"logps/chosen": -481.9137268066406,
|
| 1240 |
+
"logps/rejected": -545.0120849609375,
|
| 1241 |
+
"loss": 0.4847,
|
| 1242 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1243 |
+
"rewards/chosen": -1.9084794521331787,
|
| 1244 |
+
"rewards/margins": 0.9120246767997742,
|
| 1245 |
+
"rewards/rejected": -2.8205044269561768,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 0.79,
|
| 1250 |
+
"grad_norm": 5.781902544465049,
|
| 1251 |
+
"learning_rate": 6.70334495204884e-07,
|
| 1252 |
+
"logits/chosen": 0.2374829351902008,
|
| 1253 |
+
"logits/rejected": 0.428078830242157,
|
| 1254 |
+
"logps/chosen": -427.08099365234375,
|
| 1255 |
+
"logps/rejected": -511.5672302246094,
|
| 1256 |
+
"loss": 0.4783,
|
| 1257 |
+
"rewards/accuracies": 0.75,
|
| 1258 |
+
"rewards/chosen": -1.7517359256744385,
|
| 1259 |
+
"rewards/margins": 0.9133321642875671,
|
| 1260 |
+
"rewards/rejected": -2.6650681495666504,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.8,
|
| 1265 |
+
"grad_norm": 8.985975146227219,
|
| 1266 |
+
"learning_rate": 6.092659210462232e-07,
|
| 1267 |
+
"logits/chosen": -0.05799068883061409,
|
| 1268 |
+
"logits/rejected": 0.24734115600585938,
|
| 1269 |
+
"logps/chosen": -439.37841796875,
|
| 1270 |
+
"logps/rejected": -537.7448120117188,
|
| 1271 |
+
"loss": 0.5005,
|
| 1272 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1273 |
+
"rewards/chosen": -1.7082140445709229,
|
| 1274 |
+
"rewards/margins": 0.9354721903800964,
|
| 1275 |
+
"rewards/rejected": -2.643686294555664,
|
| 1276 |
+
"step": 760
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.81,
|
| 1280 |
+
"grad_norm": 8.444556330335592,
|
| 1281 |
+
"learning_rate": 5.507260361320738e-07,
|
| 1282 |
+
"logits/chosen": -0.10365080833435059,
|
| 1283 |
+
"logits/rejected": 0.6751923561096191,
|
| 1284 |
+
"logps/chosen": -425.59130859375,
|
| 1285 |
+
"logps/rejected": -492.73583984375,
|
| 1286 |
+
"loss": 0.4983,
|
| 1287 |
+
"rewards/accuracies": 0.8125,
|
| 1288 |
+
"rewards/chosen": -1.6130247116088867,
|
| 1289 |
+
"rewards/margins": 1.0618236064910889,
|
| 1290 |
+
"rewards/rejected": -2.6748480796813965,
|
| 1291 |
+
"step": 770
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.82,
|
| 1295 |
+
"grad_norm": 6.994205801584757,
|
| 1296 |
+
"learning_rate": 4.947931323697983e-07,
|
| 1297 |
+
"logits/chosen": -0.10879464447498322,
|
| 1298 |
+
"logits/rejected": 0.6311591863632202,
|
| 1299 |
+
"logps/chosen": -402.650634765625,
|
| 1300 |
+
"logps/rejected": -480.27978515625,
|
| 1301 |
+
"loss": 0.4858,
|
| 1302 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1303 |
+
"rewards/chosen": -1.6773712635040283,
|
| 1304 |
+
"rewards/margins": 0.953662097454071,
|
| 1305 |
+
"rewards/rejected": -2.631033420562744,
|
| 1306 |
+
"step": 780
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.83,
|
| 1310 |
+
"grad_norm": 7.034724618907778,
|
| 1311 |
+
"learning_rate": 4.4154201506053985e-07,
|
| 1312 |
+
"logits/chosen": -0.059246379882097244,
|
| 1313 |
+
"logits/rejected": 0.9515973925590515,
|
| 1314 |
+
"logps/chosen": -418.908935546875,
|
| 1315 |
+
"logps/rejected": -463.66497802734375,
|
| 1316 |
+
"loss": 0.4823,
|
| 1317 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1318 |
+
"rewards/chosen": -1.5309089422225952,
|
| 1319 |
+
"rewards/margins": 1.080568552017212,
|
| 1320 |
+
"rewards/rejected": -2.6114776134490967,
|
| 1321 |
+
"step": 790
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.84,
|
| 1325 |
+
"grad_norm": 11.254330628062345,
|
| 1326 |
+
"learning_rate": 3.910439028537638e-07,
|
| 1327 |
+
"logits/chosen": 0.3055418133735657,
|
| 1328 |
+
"logits/rejected": 0.5041495561599731,
|
| 1329 |
+
"logps/chosen": -393.4361267089844,
|
| 1330 |
+
"logps/rejected": -491.19537353515625,
|
| 1331 |
+
"loss": 0.495,
|
| 1332 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1333 |
+
"rewards/chosen": -1.6317393779754639,
|
| 1334 |
+
"rewards/margins": 0.9304509162902832,
|
| 1335 |
+
"rewards/rejected": -2.562190532684326,
|
| 1336 |
+
"step": 800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 0.84,
|
| 1340 |
+
"eval_logits/chosen": 0.4157378673553467,
|
| 1341 |
+
"eval_logits/rejected": 0.8307757377624512,
|
| 1342 |
+
"eval_logps/chosen": -441.31207275390625,
|
| 1343 |
+
"eval_logps/rejected": -530.7620239257812,
|
| 1344 |
+
"eval_loss": 0.49489909410476685,
|
| 1345 |
+
"eval_rewards/accuracies": 0.726190447807312,
|
| 1346 |
+
"eval_rewards/chosen": -1.7381722927093506,
|
| 1347 |
+
"eval_rewards/margins": 0.9458035230636597,
|
| 1348 |
+
"eval_rewards/rejected": -2.6839756965637207,
|
| 1349 |
+
"eval_runtime": 363.554,
|
| 1350 |
+
"eval_samples_per_second": 5.501,
|
| 1351 |
+
"eval_steps_per_second": 0.173,
|
| 1352 |
+
"step": 800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.85,
|
| 1356 |
+
"grad_norm": 10.131129540510356,
|
| 1357 |
+
"learning_rate": 3.4336633249862084e-07,
|
| 1358 |
+
"logits/chosen": 0.2840031385421753,
|
| 1359 |
+
"logits/rejected": 0.8334852457046509,
|
| 1360 |
+
"logps/chosen": -408.244384765625,
|
| 1361 |
+
"logps/rejected": -496.79852294921875,
|
| 1362 |
+
"loss": 0.5046,
|
| 1363 |
+
"rewards/accuracies": 0.75,
|
| 1364 |
+
"rewards/chosen": -1.7685251235961914,
|
| 1365 |
+
"rewards/margins": 0.8878759145736694,
|
| 1366 |
+
"rewards/rejected": -2.6564011573791504,
|
| 1367 |
+
"step": 810
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.86,
|
| 1371 |
+
"grad_norm": 6.762519528965419,
|
| 1372 |
+
"learning_rate": 2.98573068519539e-07,
|
| 1373 |
+
"logits/chosen": 0.10075119882822037,
|
| 1374 |
+
"logits/rejected": 0.3803648352622986,
|
| 1375 |
+
"logps/chosen": -466.81689453125,
|
| 1376 |
+
"logps/rejected": -508.10394287109375,
|
| 1377 |
+
"loss": 0.5171,
|
| 1378 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1379 |
+
"rewards/chosen": -1.9121501445770264,
|
| 1380 |
+
"rewards/margins": 0.6225390434265137,
|
| 1381 |
+
"rewards/rejected": -2.534689426422119,
|
| 1382 |
+
"step": 820
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.87,
|
| 1386 |
+
"grad_norm": 5.70932547678816,
|
| 1387 |
+
"learning_rate": 2.5672401793681854e-07,
|
| 1388 |
+
"logits/chosen": 0.38261884450912476,
|
| 1389 |
+
"logits/rejected": 0.6645594835281372,
|
| 1390 |
+
"logps/chosen": -452.214111328125,
|
| 1391 |
+
"logps/rejected": -503.60479736328125,
|
| 1392 |
+
"loss": 0.4928,
|
| 1393 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1394 |
+
"rewards/chosen": -1.784679651260376,
|
| 1395 |
+
"rewards/margins": 0.795635461807251,
|
| 1396 |
+
"rewards/rejected": -2.580315113067627,
|
| 1397 |
+
"step": 830
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.88,
|
| 1401 |
+
"grad_norm": 6.787751487161624,
|
| 1402 |
+
"learning_rate": 2.178751501463036e-07,
|
| 1403 |
+
"logits/chosen": 0.19089770317077637,
|
| 1404 |
+
"logits/rejected": 0.6150975227355957,
|
| 1405 |
+
"logps/chosen": -435.7149963378906,
|
| 1406 |
+
"logps/rejected": -512.2667236328125,
|
| 1407 |
+
"loss": 0.5099,
|
| 1408 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1409 |
+
"rewards/chosen": -1.6579662561416626,
|
| 1410 |
+
"rewards/margins": 0.820970892906189,
|
| 1411 |
+
"rewards/rejected": -2.4789371490478516,
|
| 1412 |
+
"step": 840
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.89,
|
| 1416 |
+
"grad_norm": 7.2710507714696275,
|
| 1417 |
+
"learning_rate": 1.820784220652766e-07,
|
| 1418 |
+
"logits/chosen": 0.3307928442955017,
|
| 1419 |
+
"logits/rejected": 0.8749529719352722,
|
| 1420 |
+
"logps/chosen": -447.7725524902344,
|
| 1421 |
+
"logps/rejected": -521.2926025390625,
|
| 1422 |
+
"loss": 0.4866,
|
| 1423 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1424 |
+
"rewards/chosen": -1.6773738861083984,
|
| 1425 |
+
"rewards/margins": 1.1296848058700562,
|
| 1426 |
+
"rewards/rejected": -2.807058811187744,
|
| 1427 |
+
"step": 850
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.9,
|
| 1431 |
+
"grad_norm": 7.229633532426356,
|
| 1432 |
+
"learning_rate": 1.4938170864468636e-07,
|
| 1433 |
+
"logits/chosen": 0.37119001150131226,
|
| 1434 |
+
"logits/rejected": 0.9635122418403625,
|
| 1435 |
+
"logps/chosen": -403.9157409667969,
|
| 1436 |
+
"logps/rejected": -498.6942443847656,
|
| 1437 |
+
"loss": 0.4717,
|
| 1438 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1439 |
+
"rewards/chosen": -1.6513326168060303,
|
| 1440 |
+
"rewards/margins": 1.0772497653961182,
|
| 1441 |
+
"rewards/rejected": -2.7285823822021484,
|
| 1442 |
+
"step": 860
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.91,
|
| 1446 |
+
"grad_norm": 8.52193537691758,
|
| 1447 |
+
"learning_rate": 1.1982873884064466e-07,
|
| 1448 |
+
"logits/chosen": 0.4872625768184662,
|
| 1449 |
+
"logits/rejected": 0.821652889251709,
|
| 1450 |
+
"logps/chosen": -409.1003723144531,
|
| 1451 |
+
"logps/rejected": -532.8604736328125,
|
| 1452 |
+
"loss": 0.5032,
|
| 1453 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1454 |
+
"rewards/chosen": -1.784326195716858,
|
| 1455 |
+
"rewards/margins": 1.051481008529663,
|
| 1456 |
+
"rewards/rejected": -2.8358073234558105,
|
| 1457 |
+
"step": 870
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.92,
|
| 1461 |
+
"grad_norm": 6.240495703154818,
|
| 1462 |
+
"learning_rate": 9.345903713082305e-08,
|
| 1463 |
+
"logits/chosen": 0.43627986311912537,
|
| 1464 |
+
"logits/rejected": 0.9064377546310425,
|
| 1465 |
+
"logps/chosen": -447.2481994628906,
|
| 1466 |
+
"logps/rejected": -526.6138916015625,
|
| 1467 |
+
"loss": 0.4741,
|
| 1468 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1469 |
+
"rewards/chosen": -1.7614307403564453,
|
| 1470 |
+
"rewards/margins": 1.0818777084350586,
|
| 1471 |
+
"rewards/rejected": -2.843308210372925,
|
| 1472 |
+
"step": 880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.93,
|
| 1476 |
+
"grad_norm": 7.1651137983571624,
|
| 1477 |
+
"learning_rate": 7.030787065396866e-08,
|
| 1478 |
+
"logits/chosen": 0.2825869917869568,
|
| 1479 |
+
"logits/rejected": 0.7335230112075806,
|
| 1480 |
+
"logps/chosen": -467.37469482421875,
|
| 1481 |
+
"logps/rejected": -523.0452880859375,
|
| 1482 |
+
"loss": 0.4941,
|
| 1483 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1484 |
+
"rewards/chosen": -1.9135255813598633,
|
| 1485 |
+
"rewards/margins": 0.7427676320075989,
|
| 1486 |
+
"rewards/rejected": -2.6562931537628174,
|
| 1487 |
+
"step": 890
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.94,
|
| 1491 |
+
"grad_norm": 7.718450007612897,
|
| 1492 |
+
"learning_rate": 5.0406202043228604e-08,
|
| 1493 |
+
"logits/chosen": 0.40967661142349243,
|
| 1494 |
+
"logits/rejected": 0.6366448998451233,
|
| 1495 |
+
"logps/chosen": -437.923095703125,
|
| 1496 |
+
"logps/rejected": -538.6532592773438,
|
| 1497 |
+
"loss": 0.4866,
|
| 1498 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1499 |
+
"rewards/chosen": -1.7436615228652954,
|
| 1500 |
+
"rewards/margins": 0.9877888560295105,
|
| 1501 |
+
"rewards/rejected": -2.731450319290161,
|
| 1502 |
+
"step": 900
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.94,
|
| 1506 |
+
"eval_logits/chosen": 0.5257142782211304,
|
| 1507 |
+
"eval_logits/rejected": 0.935259997844696,
|
| 1508 |
+
"eval_logps/chosen": -446.6502990722656,
|
| 1509 |
+
"eval_logps/rejected": -535.5853881835938,
|
| 1510 |
+
"eval_loss": 0.49319854378700256,
|
| 1511 |
+
"eval_rewards/accuracies": 0.726190447807312,
|
| 1512 |
+
"eval_rewards/chosen": -1.79155433177948,
|
| 1513 |
+
"eval_rewards/margins": 0.9406552314758301,
|
| 1514 |
+
"eval_rewards/rejected": -2.7322094440460205,
|
| 1515 |
+
"eval_runtime": 363.747,
|
| 1516 |
+
"eval_samples_per_second": 5.498,
|
| 1517 |
+
"eval_steps_per_second": 0.173,
|
| 1518 |
+
"step": 900
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 0.95,
|
| 1522 |
+
"grad_norm": 7.916275121810713,
|
| 1523 |
+
"learning_rate": 3.378064801637687e-08,
|
| 1524 |
+
"logits/chosen": 0.09373348951339722,
|
| 1525 |
+
"logits/rejected": 0.7054184675216675,
|
| 1526 |
+
"logps/chosen": -458.953369140625,
|
| 1527 |
+
"logps/rejected": -513.7935791015625,
|
| 1528 |
+
"loss": 0.5184,
|
| 1529 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1530 |
+
"rewards/chosen": -1.85234797000885,
|
| 1531 |
+
"rewards/margins": 0.8529074788093567,
|
| 1532 |
+
"rewards/rejected": -2.7052555084228516,
|
| 1533 |
+
"step": 910
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.96,
|
| 1537 |
+
"grad_norm": 6.7231952954282646,
|
| 1538 |
+
"learning_rate": 2.0453443778310766e-08,
|
| 1539 |
+
"logits/chosen": 0.3935597240924835,
|
| 1540 |
+
"logits/rejected": 0.7541261315345764,
|
| 1541 |
+
"logps/chosen": -428.7496643066406,
|
| 1542 |
+
"logps/rejected": -517.533935546875,
|
| 1543 |
+
"loss": 0.4802,
|
| 1544 |
+
"rewards/accuracies": 0.78125,
|
| 1545 |
+
"rewards/chosen": -1.6958158016204834,
|
| 1546 |
+
"rewards/margins": 1.0528090000152588,
|
| 1547 |
+
"rewards/rejected": -2.748624801635742,
|
| 1548 |
+
"step": 920
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 0.97,
|
| 1552 |
+
"grad_norm": 6.355797363488898,
|
| 1553 |
+
"learning_rate": 1.0442413283435759e-08,
|
| 1554 |
+
"logits/chosen": 0.23696298897266388,
|
| 1555 |
+
"logits/rejected": 0.9657135009765625,
|
| 1556 |
+
"logps/chosen": -446.9468688964844,
|
| 1557 |
+
"logps/rejected": -511.3190002441406,
|
| 1558 |
+
"loss": 0.4773,
|
| 1559 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1560 |
+
"rewards/chosen": -1.799220323562622,
|
| 1561 |
+
"rewards/margins": 0.9288120269775391,
|
| 1562 |
+
"rewards/rejected": -2.7280325889587402,
|
| 1563 |
+
"step": 930
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.98,
|
| 1567 |
+
"grad_norm": 8.823642044844457,
|
| 1568 |
+
"learning_rate": 3.760945397705828e-09,
|
| 1569 |
+
"logits/chosen": 0.3560529053211212,
|
| 1570 |
+
"logits/rejected": 0.6830099821090698,
|
| 1571 |
+
"logps/chosen": -457.27587890625,
|
| 1572 |
+
"logps/rejected": -526.8748168945312,
|
| 1573 |
+
"loss": 0.513,
|
| 1574 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1575 |
+
"rewards/chosen": -1.8548386096954346,
|
| 1576 |
+
"rewards/margins": 0.7306956052780151,
|
| 1577 |
+
"rewards/rejected": -2.585533857345581,
|
| 1578 |
+
"step": 940
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.99,
|
| 1582 |
+
"grad_norm": 7.015844125023096,
|
| 1583 |
+
"learning_rate": 4.1797599220405605e-10,
|
| 1584 |
+
"logits/chosen": 0.21756580471992493,
|
| 1585 |
+
"logits/rejected": 0.7274689078330994,
|
| 1586 |
+
"logps/chosen": -462.048583984375,
|
| 1587 |
+
"logps/rejected": -528.304443359375,
|
| 1588 |
+
"loss": 0.4919,
|
| 1589 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1590 |
+
"rewards/chosen": -1.763311743736267,
|
| 1591 |
+
"rewards/margins": 0.9584550857543945,
|
| 1592 |
+
"rewards/rejected": -2.721766948699951,
|
| 1593 |
+
"step": 950
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 1.0,
|
| 1597 |
+
"step": 955,
|
| 1598 |
+
"total_flos": 0.0,
|
| 1599 |
+
"train_loss": 0.5264237219126436,
|
| 1600 |
+
"train_runtime": 29801.8239,
|
| 1601 |
+
"train_samples_per_second": 2.051,
|
| 1602 |
+
"train_steps_per_second": 0.032
|
| 1603 |
+
}
|
| 1604 |
+
],
|
| 1605 |
+
"logging_steps": 10,
|
| 1606 |
+
"max_steps": 955,
|
| 1607 |
+
"num_input_tokens_seen": 0,
|
| 1608 |
+
"num_train_epochs": 1,
|
| 1609 |
+
"save_steps": 100000,
|
| 1610 |
+
"total_flos": 0.0,
|
| 1611 |
+
"train_batch_size": 4,
|
| 1612 |
+
"trial_name": null,
|
| 1613 |
+
"trial_params": null
|
| 1614 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4fe19b748ca3fddccc7ee76272be28fd47fa1e74bf00b4cac21fa1ecd0626e9d
|
| 3 |
+
size 6200
|