Model save
Browse files
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: gemma
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- sft
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: google/gemma-7b
|
| 9 |
+
datasets:
|
| 10 |
+
- generator
|
| 11 |
+
model-index:
|
| 12 |
+
- name: gemma7b-summarize-claude3sonnet-30k
|
| 13 |
+
results: []
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 17 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 18 |
+
|
| 19 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/chansung18/huggingface/runs/62awpxu7)
|
| 20 |
+
# gemma7b-summarize-claude3sonnet-30k
|
| 21 |
+
|
| 22 |
+
This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b) on the generator dataset.
|
| 23 |
+
It achieves the following results on the evaluation set:
|
| 24 |
+
- Loss: 2.3268
|
| 25 |
+
|
| 26 |
+
## Model description
|
| 27 |
+
|
| 28 |
+
More information needed
|
| 29 |
+
|
| 30 |
+
## Intended uses & limitations
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Training and evaluation data
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training procedure
|
| 39 |
+
|
| 40 |
+
### Training hyperparameters
|
| 41 |
+
|
| 42 |
+
The following hyperparameters were used during training:
|
| 43 |
+
- learning_rate: 0.0002
|
| 44 |
+
- train_batch_size: 4
|
| 45 |
+
- eval_batch_size: 2
|
| 46 |
+
- seed: 42
|
| 47 |
+
- distributed_type: multi-GPU
|
| 48 |
+
- num_devices: 2
|
| 49 |
+
- gradient_accumulation_steps: 2
|
| 50 |
+
- total_train_batch_size: 16
|
| 51 |
+
- total_eval_batch_size: 4
|
| 52 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 53 |
+
- lr_scheduler_type: cosine
|
| 54 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 55 |
+
- num_epochs: 5
|
| 56 |
+
|
| 57 |
+
### Training results
|
| 58 |
+
|
| 59 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 60 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
| 61 |
+
| 0.7892 | 1.0 | 205 | 2.2406 |
|
| 62 |
+
| 0.7033 | 2.0 | 410 | 2.2004 |
|
| 63 |
+
| 0.6346 | 3.0 | 615 | 2.2458 |
|
| 64 |
+
| 0.5755 | 4.0 | 820 | 2.2935 |
|
| 65 |
+
| 0.5144 | 5.0 | 1025 | 2.3268 |
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
### Framework versions
|
| 69 |
+
|
| 70 |
+
- PEFT 0.11.1
|
| 71 |
+
- Transformers 4.41.0
|
| 72 |
+
- Pytorch 2.3.0+cu121
|
| 73 |
+
- Datasets 2.19.1
|
| 74 |
+
- Tokenizers 0.19.1
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 50056096
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d3dc520155fac336f0b6d9657bef3b45cc466efecf92a56f22d0b23fec196a9a
|
| 3 |
size 50056096
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"total_flos": 1.5670950020754964e+18,
|
| 4 |
+
"train_loss": 1.4066738275202308,
|
| 5 |
+
"train_runtime": 8016.4558,
|
| 6 |
+
"train_samples": 29489,
|
| 7 |
+
"train_samples_per_second": 2.044,
|
| 8 |
+
"train_steps_per_second": 0.128
|
| 9 |
+
}
|
runs/May19_11-55-30_deep-diver-main-fresh-rabbit-1-0-0/events.out.tfevents.1716134302.deep-diver-main-fresh-rabbit-1-0-0.385.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb7c5bf8534601f9379cfe693fb53baedbad721ad1e97deb6fcbaf390d4f03fb
|
| 3 |
+
size 50694
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"total_flos": 1.5670950020754964e+18,
|
| 4 |
+
"train_loss": 1.4066738275202308,
|
| 5 |
+
"train_runtime": 8016.4558,
|
| 6 |
+
"train_samples": 29489,
|
| 7 |
+
"train_samples_per_second": 2.044,
|
| 8 |
+
"train_steps_per_second": 0.128
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1524 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1025,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.004878048780487805,
|
| 13 |
+
"grad_norm": 616.0,
|
| 14 |
+
"learning_rate": 1.941747572815534e-06,
|
| 15 |
+
"loss": 39.9034,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.024390243902439025,
|
| 20 |
+
"grad_norm": 466.0,
|
| 21 |
+
"learning_rate": 9.70873786407767e-06,
|
| 22 |
+
"loss": 34.9341,
|
| 23 |
+
"step": 5
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.04878048780487805,
|
| 27 |
+
"grad_norm": 251.0,
|
| 28 |
+
"learning_rate": 1.941747572815534e-05,
|
| 29 |
+
"loss": 29.4015,
|
| 30 |
+
"step": 10
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07317073170731707,
|
| 34 |
+
"grad_norm": 56.25,
|
| 35 |
+
"learning_rate": 2.912621359223301e-05,
|
| 36 |
+
"loss": 17.9289,
|
| 37 |
+
"step": 15
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0975609756097561,
|
| 41 |
+
"grad_norm": 39.75,
|
| 42 |
+
"learning_rate": 3.883495145631068e-05,
|
| 43 |
+
"loss": 14.9717,
|
| 44 |
+
"step": 20
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12195121951219512,
|
| 48 |
+
"grad_norm": 15.75,
|
| 49 |
+
"learning_rate": 4.854368932038835e-05,
|
| 50 |
+
"loss": 12.4657,
|
| 51 |
+
"step": 25
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.14634146341463414,
|
| 55 |
+
"grad_norm": 5.15625,
|
| 56 |
+
"learning_rate": 5.825242718446602e-05,
|
| 57 |
+
"loss": 10.6748,
|
| 58 |
+
"step": 30
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.17073170731707318,
|
| 62 |
+
"grad_norm": 3.890625,
|
| 63 |
+
"learning_rate": 6.79611650485437e-05,
|
| 64 |
+
"loss": 9.811,
|
| 65 |
+
"step": 35
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.1951219512195122,
|
| 69 |
+
"grad_norm": 6.75,
|
| 70 |
+
"learning_rate": 7.766990291262136e-05,
|
| 71 |
+
"loss": 9.6309,
|
| 72 |
+
"step": 40
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.21951219512195122,
|
| 76 |
+
"grad_norm": 15.3125,
|
| 77 |
+
"learning_rate": 8.737864077669902e-05,
|
| 78 |
+
"loss": 8.5315,
|
| 79 |
+
"step": 45
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.24390243902439024,
|
| 83 |
+
"grad_norm": 31.75,
|
| 84 |
+
"learning_rate": 9.70873786407767e-05,
|
| 85 |
+
"loss": 6.2519,
|
| 86 |
+
"step": 50
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.2682926829268293,
|
| 90 |
+
"grad_norm": 5.25,
|
| 91 |
+
"learning_rate": 0.00010679611650485437,
|
| 92 |
+
"loss": 2.2665,
|
| 93 |
+
"step": 55
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.2926829268292683,
|
| 97 |
+
"grad_norm": 2.40625,
|
| 98 |
+
"learning_rate": 0.00011650485436893204,
|
| 99 |
+
"loss": 1.5008,
|
| 100 |
+
"step": 60
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.3170731707317073,
|
| 104 |
+
"grad_norm": 1.2578125,
|
| 105 |
+
"learning_rate": 0.00012621359223300972,
|
| 106 |
+
"loss": 1.3289,
|
| 107 |
+
"step": 65
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.34146341463414637,
|
| 111 |
+
"grad_norm": 2.796875,
|
| 112 |
+
"learning_rate": 0.0001359223300970874,
|
| 113 |
+
"loss": 1.2198,
|
| 114 |
+
"step": 70
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.36585365853658536,
|
| 118 |
+
"grad_norm": 1.4375,
|
| 119 |
+
"learning_rate": 0.00014563106796116506,
|
| 120 |
+
"loss": 1.1493,
|
| 121 |
+
"step": 75
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.3902439024390244,
|
| 125 |
+
"grad_norm": 8.1875,
|
| 126 |
+
"learning_rate": 0.0001553398058252427,
|
| 127 |
+
"loss": 1.0862,
|
| 128 |
+
"step": 80
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.4146341463414634,
|
| 132 |
+
"grad_norm": 3.0625,
|
| 133 |
+
"learning_rate": 0.0001650485436893204,
|
| 134 |
+
"loss": 1.0768,
|
| 135 |
+
"step": 85
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.43902439024390244,
|
| 139 |
+
"grad_norm": 1.2109375,
|
| 140 |
+
"learning_rate": 0.00017475728155339805,
|
| 141 |
+
"loss": 1.0468,
|
| 142 |
+
"step": 90
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.4634146341463415,
|
| 146 |
+
"grad_norm": 4.21875,
|
| 147 |
+
"learning_rate": 0.00018446601941747576,
|
| 148 |
+
"loss": 0.9991,
|
| 149 |
+
"step": 95
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4878048780487805,
|
| 153 |
+
"grad_norm": 2.0625,
|
| 154 |
+
"learning_rate": 0.0001941747572815534,
|
| 155 |
+
"loss": 0.9596,
|
| 156 |
+
"step": 100
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.5121951219512195,
|
| 160 |
+
"grad_norm": 1.5078125,
|
| 161 |
+
"learning_rate": 0.00019999767797859854,
|
| 162 |
+
"loss": 0.9712,
|
| 163 |
+
"step": 105
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.5365853658536586,
|
| 167 |
+
"grad_norm": 0.890625,
|
| 168 |
+
"learning_rate": 0.0001999715564762413,
|
| 169 |
+
"loss": 0.9139,
|
| 170 |
+
"step": 110
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.5609756097560976,
|
| 174 |
+
"grad_norm": 1.1875,
|
| 175 |
+
"learning_rate": 0.00019991641855173097,
|
| 176 |
+
"loss": 0.9175,
|
| 177 |
+
"step": 115
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5853658536585366,
|
| 181 |
+
"grad_norm": 1.25,
|
| 182 |
+
"learning_rate": 0.00019983228020867242,
|
| 183 |
+
"loss": 0.8959,
|
| 184 |
+
"step": 120
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.6097560975609756,
|
| 188 |
+
"grad_norm": 1.3203125,
|
| 189 |
+
"learning_rate": 0.00019971916586794867,
|
| 190 |
+
"loss": 0.8605,
|
| 191 |
+
"step": 125
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.6341463414634146,
|
| 195 |
+
"grad_norm": 0.98828125,
|
| 196 |
+
"learning_rate": 0.00019957710836063263,
|
| 197 |
+
"loss": 0.8857,
|
| 198 |
+
"step": 130
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.6585365853658537,
|
| 202 |
+
"grad_norm": 1.8046875,
|
| 203 |
+
"learning_rate": 0.00019940614891845809,
|
| 204 |
+
"loss": 0.8533,
|
| 205 |
+
"step": 135
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.6829268292682927,
|
| 209 |
+
"grad_norm": 1.9140625,
|
| 210 |
+
"learning_rate": 0.00019920633716185226,
|
| 211 |
+
"loss": 0.8332,
|
| 212 |
+
"step": 140
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.7073170731707317,
|
| 216 |
+
"grad_norm": 0.9140625,
|
| 217 |
+
"learning_rate": 0.00019897773108553378,
|
| 218 |
+
"loss": 0.8485,
|
| 219 |
+
"step": 145
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.7317073170731707,
|
| 223 |
+
"grad_norm": 1.9296875,
|
| 224 |
+
"learning_rate": 0.00019872039704167964,
|
| 225 |
+
"loss": 0.8479,
|
| 226 |
+
"step": 150
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.7560975609756098,
|
| 230 |
+
"grad_norm": 1.4921875,
|
| 231 |
+
"learning_rate": 0.00019843440972066697,
|
| 232 |
+
"loss": 0.8418,
|
| 233 |
+
"step": 155
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.7804878048780488,
|
| 237 |
+
"grad_norm": 1.9296875,
|
| 238 |
+
"learning_rate": 0.00019811985212939416,
|
| 239 |
+
"loss": 0.8513,
|
| 240 |
+
"step": 160
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.8048780487804879,
|
| 244 |
+
"grad_norm": 1.5234375,
|
| 245 |
+
"learning_rate": 0.00019777681556718864,
|
| 246 |
+
"loss": 0.8217,
|
| 247 |
+
"step": 165
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.8292682926829268,
|
| 251 |
+
"grad_norm": 1.8671875,
|
| 252 |
+
"learning_rate": 0.00019740539959930725,
|
| 253 |
+
"loss": 0.8169,
|
| 254 |
+
"step": 170
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.8536585365853658,
|
| 258 |
+
"grad_norm": 1.046875,
|
| 259 |
+
"learning_rate": 0.00019700571202803797,
|
| 260 |
+
"loss": 0.8386,
|
| 261 |
+
"step": 175
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.8780487804878049,
|
| 265 |
+
"grad_norm": 1.2421875,
|
| 266 |
+
"learning_rate": 0.00019657786886141052,
|
| 267 |
+
"loss": 0.8144,
|
| 268 |
+
"step": 180
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.9024390243902439,
|
| 272 |
+
"grad_norm": 4.15625,
|
| 273 |
+
"learning_rate": 0.00019612199427952552,
|
| 274 |
+
"loss": 0.8229,
|
| 275 |
+
"step": 185
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.926829268292683,
|
| 279 |
+
"grad_norm": 0.70703125,
|
| 280 |
+
"learning_rate": 0.00019563822059851145,
|
| 281 |
+
"loss": 0.8128,
|
| 282 |
+
"step": 190
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.9512195121951219,
|
| 286 |
+
"grad_norm": 0.97265625,
|
| 287 |
+
"learning_rate": 0.00019512668823212055,
|
| 288 |
+
"loss": 0.8072,
|
| 289 |
+
"step": 195
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.975609756097561,
|
| 293 |
+
"grad_norm": 0.6484375,
|
| 294 |
+
"learning_rate": 0.0001945875456509739,
|
| 295 |
+
"loss": 0.8097,
|
| 296 |
+
"step": 200
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.0,
|
| 300 |
+
"grad_norm": 0.6953125,
|
| 301 |
+
"learning_rate": 0.00019402094933946857,
|
| 302 |
+
"loss": 0.7892,
|
| 303 |
+
"step": 205
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 1.0,
|
| 307 |
+
"eval_loss": 2.2406020164489746,
|
| 308 |
+
"eval_runtime": 0.9998,
|
| 309 |
+
"eval_samples_per_second": 5.001,
|
| 310 |
+
"eval_steps_per_second": 2.0,
|
| 311 |
+
"step": 205
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.024390243902439,
|
| 315 |
+
"grad_norm": 0.6796875,
|
| 316 |
+
"learning_rate": 0.0001934270637503584,
|
| 317 |
+
"loss": 0.7753,
|
| 318 |
+
"step": 210
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.048780487804878,
|
| 322 |
+
"grad_norm": 0.71875,
|
| 323 |
+
"learning_rate": 0.00019280606125702203,
|
| 324 |
+
"loss": 0.7289,
|
| 325 |
+
"step": 215
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.0731707317073171,
|
| 329 |
+
"grad_norm": 0.625,
|
| 330 |
+
"learning_rate": 0.00019215812210343226,
|
| 331 |
+
"loss": 0.7656,
|
| 332 |
+
"step": 220
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.0975609756097562,
|
| 336 |
+
"grad_norm": 0.65625,
|
| 337 |
+
"learning_rate": 0.00019148343435184079,
|
| 338 |
+
"loss": 0.7603,
|
| 339 |
+
"step": 225
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.1219512195121952,
|
| 343 |
+
"grad_norm": 0.671875,
|
| 344 |
+
"learning_rate": 0.00019078219382819353,
|
| 345 |
+
"loss": 0.7491,
|
| 346 |
+
"step": 230
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.146341463414634,
|
| 350 |
+
"grad_norm": 0.56640625,
|
| 351 |
+
"learning_rate": 0.00019005460406529311,
|
| 352 |
+
"loss": 0.7249,
|
| 353 |
+
"step": 235
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.170731707317073,
|
| 357 |
+
"grad_norm": 0.57421875,
|
| 358 |
+
"learning_rate": 0.00018930087624372387,
|
| 359 |
+
"loss": 0.7755,
|
| 360 |
+
"step": 240
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.1951219512195121,
|
| 364 |
+
"grad_norm": 0.703125,
|
| 365 |
+
"learning_rate": 0.00018852122913055742,
|
| 366 |
+
"loss": 0.7553,
|
| 367 |
+
"step": 245
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 1.2195121951219512,
|
| 371 |
+
"grad_norm": 0.890625,
|
| 372 |
+
"learning_rate": 0.00018771588901585635,
|
| 373 |
+
"loss": 0.7482,
|
| 374 |
+
"step": 250
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.2439024390243902,
|
| 378 |
+
"grad_norm": 0.77734375,
|
| 379 |
+
"learning_rate": 0.00018688508964699404,
|
| 380 |
+
"loss": 0.7402,
|
| 381 |
+
"step": 255
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 1.2682926829268293,
|
| 385 |
+
"grad_norm": 1.015625,
|
| 386 |
+
"learning_rate": 0.00018602907216081044,
|
| 387 |
+
"loss": 0.7263,
|
| 388 |
+
"step": 260
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.2926829268292683,
|
| 392 |
+
"grad_norm": 1.3828125,
|
| 393 |
+
"learning_rate": 0.0001851480850136228,
|
| 394 |
+
"loss": 0.7435,
|
| 395 |
+
"step": 265
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.3170731707317074,
|
| 399 |
+
"grad_norm": 0.58984375,
|
| 400 |
+
"learning_rate": 0.00018424238390911198,
|
| 401 |
+
"loss": 0.7457,
|
| 402 |
+
"step": 270
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.3414634146341464,
|
| 406 |
+
"grad_norm": 1.0078125,
|
| 407 |
+
"learning_rate": 0.00018331223172410535,
|
| 408 |
+
"loss": 0.7483,
|
| 409 |
+
"step": 275
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.3658536585365852,
|
| 413 |
+
"grad_norm": 0.609375,
|
| 414 |
+
"learning_rate": 0.00018235789843227756,
|
| 415 |
+
"loss": 0.7486,
|
| 416 |
+
"step": 280
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.3902439024390243,
|
| 420 |
+
"grad_norm": 1.265625,
|
| 421 |
+
"learning_rate": 0.00018137966102579176,
|
| 422 |
+
"loss": 0.7548,
|
| 423 |
+
"step": 285
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.4146341463414633,
|
| 427 |
+
"grad_norm": 1.1953125,
|
| 428 |
+
"learning_rate": 0.00018037780343490312,
|
| 429 |
+
"loss": 0.7282,
|
| 430 |
+
"step": 290
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.4390243902439024,
|
| 434 |
+
"grad_norm": 0.6875,
|
| 435 |
+
"learning_rate": 0.00017935261644554942,
|
| 436 |
+
"loss": 0.7415,
|
| 437 |
+
"step": 295
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.4634146341463414,
|
| 441 |
+
"grad_norm": 0.73046875,
|
| 442 |
+
"learning_rate": 0.0001783043976149511,
|
| 443 |
+
"loss": 0.7362,
|
| 444 |
+
"step": 300
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.4878048780487805,
|
| 448 |
+
"grad_norm": 0.578125,
|
| 449 |
+
"learning_rate": 0.0001772334511852463,
|
| 450 |
+
"loss": 0.724,
|
| 451 |
+
"step": 305
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.5121951219512195,
|
| 455 |
+
"grad_norm": 0.59765625,
|
| 456 |
+
"learning_rate": 0.0001761400879951856,
|
| 457 |
+
"loss": 0.7249,
|
| 458 |
+
"step": 310
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.5365853658536586,
|
| 462 |
+
"grad_norm": 1.6953125,
|
| 463 |
+
"learning_rate": 0.00017502462538991205,
|
| 464 |
+
"loss": 0.7212,
|
| 465 |
+
"step": 315
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.5609756097560976,
|
| 469 |
+
"grad_norm": 0.6328125,
|
| 470 |
+
"learning_rate": 0.00017388738712885275,
|
| 471 |
+
"loss": 0.7194,
|
| 472 |
+
"step": 320
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.5853658536585367,
|
| 476 |
+
"grad_norm": 1.3125,
|
| 477 |
+
"learning_rate": 0.0001727287032917487,
|
| 478 |
+
"loss": 0.765,
|
| 479 |
+
"step": 325
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.6097560975609757,
|
| 483 |
+
"grad_norm": 1.578125,
|
| 484 |
+
"learning_rate": 0.00017154891018285028,
|
| 485 |
+
"loss": 0.7318,
|
| 486 |
+
"step": 330
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.6341463414634148,
|
| 490 |
+
"grad_norm": 0.50390625,
|
| 491 |
+
"learning_rate": 0.00017034835023330597,
|
| 492 |
+
"loss": 0.7396,
|
| 493 |
+
"step": 335
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.6585365853658538,
|
| 497 |
+
"grad_norm": 2.0625,
|
| 498 |
+
"learning_rate": 0.00016912737190177292,
|
| 499 |
+
"loss": 0.737,
|
| 500 |
+
"step": 340
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.6829268292682928,
|
| 504 |
+
"grad_norm": 0.59765625,
|
| 505 |
+
"learning_rate": 0.00016788632957327772,
|
| 506 |
+
"loss": 0.7302,
|
| 507 |
+
"step": 345
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.7073170731707317,
|
| 511 |
+
"grad_norm": 0.65234375,
|
| 512 |
+
"learning_rate": 0.00016662558345635753,
|
| 513 |
+
"loss": 0.7193,
|
| 514 |
+
"step": 350
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.7317073170731707,
|
| 518 |
+
"grad_norm": 0.76953125,
|
| 519 |
+
"learning_rate": 0.00016534549947851062,
|
| 520 |
+
"loss": 0.7147,
|
| 521 |
+
"step": 355
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.7560975609756098,
|
| 525 |
+
"grad_norm": 0.73828125,
|
| 526 |
+
"learning_rate": 0.00016404644917998698,
|
| 527 |
+
"loss": 0.7539,
|
| 528 |
+
"step": 360
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.7804878048780488,
|
| 532 |
+
"grad_norm": 0.9453125,
|
| 533 |
+
"learning_rate": 0.00016272880960595024,
|
| 534 |
+
"loss": 0.754,
|
| 535 |
+
"step": 365
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.8048780487804879,
|
| 539 |
+
"grad_norm": 0.90625,
|
| 540 |
+
"learning_rate": 0.00016139296319704117,
|
| 541 |
+
"loss": 0.7346,
|
| 542 |
+
"step": 370
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.8292682926829267,
|
| 546 |
+
"grad_norm": 0.89453125,
|
| 547 |
+
"learning_rate": 0.00016003929767837588,
|
| 548 |
+
"loss": 0.6961,
|
| 549 |
+
"step": 375
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.8536585365853657,
|
| 553 |
+
"grad_norm": 1.078125,
|
| 554 |
+
"learning_rate": 0.00015866820594700944,
|
| 555 |
+
"loss": 0.7259,
|
| 556 |
+
"step": 380
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.8780487804878048,
|
| 560 |
+
"grad_norm": 0.65625,
|
| 561 |
+
"learning_rate": 0.00015728008595789926,
|
| 562 |
+
"loss": 0.7198,
|
| 563 |
+
"step": 385
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.9024390243902438,
|
| 567 |
+
"grad_norm": 0.515625,
|
| 568 |
+
"learning_rate": 0.0001558753406083995,
|
| 569 |
+
"loss": 0.6858,
|
| 570 |
+
"step": 390
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.9268292682926829,
|
| 574 |
+
"grad_norm": 0.60546875,
|
| 575 |
+
"learning_rate": 0.00015445437762132174,
|
| 576 |
+
"loss": 0.6987,
|
| 577 |
+
"step": 395
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.951219512195122,
|
| 581 |
+
"grad_norm": 0.5625,
|
| 582 |
+
"learning_rate": 0.0001530176094265945,
|
| 583 |
+
"loss": 0.7054,
|
| 584 |
+
"step": 400
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.975609756097561,
|
| 588 |
+
"grad_norm": 0.62890625,
|
| 589 |
+
"learning_rate": 0.00015156545304155698,
|
| 590 |
+
"loss": 0.7226,
|
| 591 |
+
"step": 405
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.0,
|
| 595 |
+
"grad_norm": 0.52734375,
|
| 596 |
+
"learning_rate": 0.00015009832994992102,
|
| 597 |
+
"loss": 0.7033,
|
| 598 |
+
"step": 410
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 2.0,
|
| 602 |
+
"eval_loss": 2.2004494667053223,
|
| 603 |
+
"eval_runtime": 1.0005,
|
| 604 |
+
"eval_samples_per_second": 4.998,
|
| 605 |
+
"eval_steps_per_second": 1.999,
|
| 606 |
+
"step": 410
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 2.024390243902439,
|
| 610 |
+
"grad_norm": 0.6328125,
|
| 611 |
+
"learning_rate": 0.0001486166659794368,
|
| 612 |
+
"loss": 0.6398,
|
| 613 |
+
"step": 415
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 2.048780487804878,
|
| 617 |
+
"grad_norm": 0.65234375,
|
| 618 |
+
"learning_rate": 0.00014712089117829776,
|
| 619 |
+
"loss": 0.647,
|
| 620 |
+
"step": 420
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.073170731707317,
|
| 624 |
+
"grad_norm": 0.66796875,
|
| 625 |
+
"learning_rate": 0.0001456114396903204,
|
| 626 |
+
"loss": 0.6333,
|
| 627 |
+
"step": 425
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.097560975609756,
|
| 631 |
+
"grad_norm": 0.60546875,
|
| 632 |
+
"learning_rate": 0.0001440887496289356,
|
| 633 |
+
"loss": 0.6296,
|
| 634 |
+
"step": 430
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 2.1219512195121952,
|
| 638 |
+
"grad_norm": 0.546875,
|
| 639 |
+
"learning_rate": 0.00014255326295002754,
|
| 640 |
+
"loss": 0.6516,
|
| 641 |
+
"step": 435
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 2.1463414634146343,
|
| 645 |
+
"grad_norm": 0.671875,
|
| 646 |
+
"learning_rate": 0.00014100542532365724,
|
| 647 |
+
"loss": 0.6468,
|
| 648 |
+
"step": 440
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 2.1707317073170733,
|
| 652 |
+
"grad_norm": 0.578125,
|
| 653 |
+
"learning_rate": 0.0001394456860047086,
|
| 654 |
+
"loss": 0.6472,
|
| 655 |
+
"step": 445
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 2.1951219512195124,
|
| 659 |
+
"grad_norm": 0.77734375,
|
| 660 |
+
"learning_rate": 0.00013787449770249336,
|
| 661 |
+
"loss": 0.6587,
|
| 662 |
+
"step": 450
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.2195121951219514,
|
| 666 |
+
"grad_norm": 0.70703125,
|
| 667 |
+
"learning_rate": 0.0001362923164493538,
|
| 668 |
+
"loss": 0.6505,
|
| 669 |
+
"step": 455
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.2439024390243905,
|
| 673 |
+
"grad_norm": 0.69140625,
|
| 674 |
+
"learning_rate": 0.00013469960146830073,
|
| 675 |
+
"loss": 0.6463,
|
| 676 |
+
"step": 460
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 2.2682926829268295,
|
| 680 |
+
"grad_norm": 0.72265625,
|
| 681 |
+
"learning_rate": 0.00013309681503972565,
|
| 682 |
+
"loss": 0.6372,
|
| 683 |
+
"step": 465
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.292682926829268,
|
| 687 |
+
"grad_norm": 0.80078125,
|
| 688 |
+
"learning_rate": 0.00013148442236722506,
|
| 689 |
+
"loss": 0.6285,
|
| 690 |
+
"step": 470
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 2.317073170731707,
|
| 694 |
+
"grad_norm": 0.7578125,
|
| 695 |
+
"learning_rate": 0.00012986289144257705,
|
| 696 |
+
"loss": 0.6645,
|
| 697 |
+
"step": 475
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 2.341463414634146,
|
| 701 |
+
"grad_norm": 1.0234375,
|
| 702 |
+
"learning_rate": 0.00012823269290990777,
|
| 703 |
+
"loss": 0.6663,
|
| 704 |
+
"step": 480
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.3658536585365852,
|
| 708 |
+
"grad_norm": 1.0625,
|
| 709 |
+
"learning_rate": 0.0001265942999290887,
|
| 710 |
+
"loss": 0.6495,
|
| 711 |
+
"step": 485
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.3902439024390243,
|
| 715 |
+
"grad_norm": 1.0078125,
|
| 716 |
+
"learning_rate": 0.00012494818803840367,
|
| 717 |
+
"loss": 0.6429,
|
| 718 |
+
"step": 490
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 2.4146341463414633,
|
| 722 |
+
"grad_norm": 1.796875,
|
| 723 |
+
"learning_rate": 0.00012329483501652492,
|
| 724 |
+
"loss": 0.6573,
|
| 725 |
+
"step": 495
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 2.4390243902439024,
|
| 729 |
+
"grad_norm": 0.921875,
|
| 730 |
+
"learning_rate": 0.00012163472074383994,
|
| 731 |
+
"loss": 0.6487,
|
| 732 |
+
"step": 500
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 2.4634146341463414,
|
| 736 |
+
"grad_norm": 0.59375,
|
| 737 |
+
"learning_rate": 0.00011996832706316739,
|
| 738 |
+
"loss": 0.6726,
|
| 739 |
+
"step": 505
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 2.4878048780487805,
|
| 743 |
+
"grad_norm": 0.5546875,
|
| 744 |
+
"learning_rate": 0.00011829613763990384,
|
| 745 |
+
"loss": 0.6546,
|
| 746 |
+
"step": 510
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 2.5121951219512195,
|
| 750 |
+
"grad_norm": 0.57421875,
|
| 751 |
+
"learning_rate": 0.00011661863782164153,
|
| 752 |
+
"loss": 0.6306,
|
| 753 |
+
"step": 515
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 2.5365853658536586,
|
| 757 |
+
"grad_norm": 0.61328125,
|
| 758 |
+
"learning_rate": 0.00011493631449729767,
|
| 759 |
+
"loss": 0.6591,
|
| 760 |
+
"step": 520
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 2.5609756097560976,
|
| 764 |
+
"grad_norm": 0.578125,
|
| 765 |
+
"learning_rate": 0.00011324965595579666,
|
| 766 |
+
"loss": 0.6387,
|
| 767 |
+
"step": 525
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 2.5853658536585367,
|
| 771 |
+
"grad_norm": 0.60546875,
|
| 772 |
+
"learning_rate": 0.00011155915174434561,
|
| 773 |
+
"loss": 0.6539,
|
| 774 |
+
"step": 530
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 2.6097560975609757,
|
| 778 |
+
"grad_norm": 0.87109375,
|
| 779 |
+
"learning_rate": 0.00010986529252634503,
|
| 780 |
+
"loss": 0.6654,
|
| 781 |
+
"step": 535
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 2.6341463414634148,
|
| 785 |
+
"grad_norm": 0.859375,
|
| 786 |
+
"learning_rate": 0.00010816856993897522,
|
| 787 |
+
"loss": 0.6471,
|
| 788 |
+
"step": 540
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.658536585365854,
|
| 792 |
+
"grad_norm": 0.5703125,
|
| 793 |
+
"learning_rate": 0.00010646947645050023,
|
| 794 |
+
"loss": 0.6482,
|
| 795 |
+
"step": 545
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 2.682926829268293,
|
| 799 |
+
"grad_norm": 0.5703125,
|
| 800 |
+
"learning_rate": 0.00010476850521733048,
|
| 801 |
+
"loss": 0.6252,
|
| 802 |
+
"step": 550
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.7073170731707314,
|
| 806 |
+
"grad_norm": 0.69140625,
|
| 807 |
+
"learning_rate": 0.00010306614994088582,
|
| 808 |
+
"loss": 0.6605,
|
| 809 |
+
"step": 555
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 2.7317073170731705,
|
| 813 |
+
"grad_norm": 0.52734375,
|
| 814 |
+
"learning_rate": 0.00010136290472430013,
|
| 815 |
+
"loss": 0.646,
|
| 816 |
+
"step": 560
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 2.7560975609756095,
|
| 820 |
+
"grad_norm": 0.64453125,
|
| 821 |
+
"learning_rate": 9.965926392900956e-05,
|
| 822 |
+
"loss": 0.6492,
|
| 823 |
+
"step": 565
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 2.7804878048780486,
|
| 827 |
+
"grad_norm": 0.63671875,
|
| 828 |
+
"learning_rate": 9.795572203126573e-05,
|
| 829 |
+
"loss": 0.6477,
|
| 830 |
+
"step": 570
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 2.8048780487804876,
|
| 834 |
+
"grad_norm": 0.703125,
|
| 835 |
+
"learning_rate": 9.625277347861553e-05,
|
| 836 |
+
"loss": 0.6268,
|
| 837 |
+
"step": 575
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.8292682926829267,
|
| 841 |
+
"grad_norm": 0.80078125,
|
| 842 |
+
"learning_rate": 9.455091254638939e-05,
|
| 843 |
+
"loss": 0.6364,
|
| 844 |
+
"step": 580
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 2.8536585365853657,
|
| 848 |
+
"grad_norm": 0.62890625,
|
| 849 |
+
"learning_rate": 9.285063319423939e-05,
|
| 850 |
+
"loss": 0.6497,
|
| 851 |
+
"step": 585
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 2.8780487804878048,
|
| 855 |
+
"grad_norm": 0.58203125,
|
| 856 |
+
"learning_rate": 9.115242892276909e-05,
|
| 857 |
+
"loss": 0.6446,
|
| 858 |
+
"step": 590
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 2.902439024390244,
|
| 862 |
+
"grad_norm": 0.70703125,
|
| 863 |
+
"learning_rate": 8.945679263029661e-05,
|
| 864 |
+
"loss": 0.6657,
|
| 865 |
+
"step": 595
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 2.926829268292683,
|
| 869 |
+
"grad_norm": 0.64453125,
|
| 870 |
+
"learning_rate": 8.776421646979232e-05,
|
| 871 |
+
"loss": 0.6498,
|
| 872 |
+
"step": 600
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 2.951219512195122,
|
| 876 |
+
"grad_norm": 0.5546875,
|
| 877 |
+
"learning_rate": 8.607519170603328e-05,
|
| 878 |
+
"loss": 0.6623,
|
| 879 |
+
"step": 605
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 2.975609756097561,
|
| 883 |
+
"grad_norm": 0.55078125,
|
| 884 |
+
"learning_rate": 8.439020857301503e-05,
|
| 885 |
+
"loss": 0.6467,
|
| 886 |
+
"step": 610
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 3.0,
|
| 890 |
+
"grad_norm": 0.6875,
|
| 891 |
+
"learning_rate": 8.270975613166281e-05,
|
| 892 |
+
"loss": 0.6346,
|
| 893 |
+
"step": 615
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.0,
|
| 897 |
+
"eval_loss": 2.245811939239502,
|
| 898 |
+
"eval_runtime": 1.001,
|
| 899 |
+
"eval_samples_per_second": 4.995,
|
| 900 |
+
"eval_steps_per_second": 1.998,
|
| 901 |
+
"step": 615
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 3.024390243902439,
|
| 905 |
+
"grad_norm": 0.609375,
|
| 906 |
+
"learning_rate": 8.103432212788323e-05,
|
| 907 |
+
"loss": 0.5718,
|
| 908 |
+
"step": 620
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 3.048780487804878,
|
| 912 |
+
"grad_norm": 0.66796875,
|
| 913 |
+
"learning_rate": 7.936439285099752e-05,
|
| 914 |
+
"loss": 0.5989,
|
| 915 |
+
"step": 625
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 3.073170731707317,
|
| 919 |
+
"grad_norm": 0.62890625,
|
| 920 |
+
"learning_rate": 7.770045299259774e-05,
|
| 921 |
+
"loss": 0.598,
|
| 922 |
+
"step": 630
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 3.097560975609756,
|
| 926 |
+
"grad_norm": 0.5625,
|
| 927 |
+
"learning_rate": 7.60429855058664e-05,
|
| 928 |
+
"loss": 0.5731,
|
| 929 |
+
"step": 635
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.1219512195121952,
|
| 933 |
+
"grad_norm": 0.56640625,
|
| 934 |
+
"learning_rate": 7.439247146540109e-05,
|
| 935 |
+
"loss": 0.6034,
|
| 936 |
+
"step": 640
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 3.1463414634146343,
|
| 940 |
+
"grad_norm": 0.6015625,
|
| 941 |
+
"learning_rate": 7.274938992758403e-05,
|
| 942 |
+
"loss": 0.5669,
|
| 943 |
+
"step": 645
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 3.1707317073170733,
|
| 947 |
+
"grad_norm": 0.59765625,
|
| 948 |
+
"learning_rate": 7.111421779153745e-05,
|
| 949 |
+
"loss": 0.5873,
|
| 950 |
+
"step": 650
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 3.1951219512195124,
|
| 954 |
+
"grad_norm": 0.60546875,
|
| 955 |
+
"learning_rate": 6.94874296607052e-05,
|
| 956 |
+
"loss": 0.5674,
|
| 957 |
+
"step": 655
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 3.2195121951219514,
|
| 961 |
+
"grad_norm": 0.54296875,
|
| 962 |
+
"learning_rate": 6.786949770510071e-05,
|
| 963 |
+
"loss": 0.5726,
|
| 964 |
+
"step": 660
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 3.2439024390243905,
|
| 968 |
+
"grad_norm": 0.578125,
|
| 969 |
+
"learning_rate": 6.626089152426097e-05,
|
| 970 |
+
"loss": 0.6006,
|
| 971 |
+
"step": 665
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 3.2682926829268295,
|
| 975 |
+
"grad_norm": 0.58984375,
|
| 976 |
+
"learning_rate": 6.4662078010947e-05,
|
| 977 |
+
"loss": 0.5729,
|
| 978 |
+
"step": 670
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 3.292682926829268,
|
| 982 |
+
"grad_norm": 0.546875,
|
| 983 |
+
"learning_rate": 6.307352121562949e-05,
|
| 984 |
+
"loss": 0.5719,
|
| 985 |
+
"step": 675
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 3.317073170731707,
|
| 989 |
+
"grad_norm": 0.5859375,
|
| 990 |
+
"learning_rate": 6.149568221179993e-05,
|
| 991 |
+
"loss": 0.571,
|
| 992 |
+
"step": 680
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 3.341463414634146,
|
| 996 |
+
"grad_norm": 0.5859375,
|
| 997 |
+
"learning_rate": 5.992901896214526e-05,
|
| 998 |
+
"loss": 0.5674,
|
| 999 |
+
"step": 685
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 3.3658536585365852,
|
| 1003 |
+
"grad_norm": 0.6640625,
|
| 1004 |
+
"learning_rate": 5.837398618562584e-05,
|
| 1005 |
+
"loss": 0.5772,
|
| 1006 |
+
"step": 690
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 3.3902439024390243,
|
| 1010 |
+
"grad_norm": 0.58203125,
|
| 1011 |
+
"learning_rate": 5.68310352254946e-05,
|
| 1012 |
+
"loss": 0.5893,
|
| 1013 |
+
"step": 695
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 3.4146341463414633,
|
| 1017 |
+
"grad_norm": 0.62109375,
|
| 1018 |
+
"learning_rate": 5.5300613918296295e-05,
|
| 1019 |
+
"loss": 0.5771,
|
| 1020 |
+
"step": 700
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 3.4390243902439024,
|
| 1024 |
+
"grad_norm": 0.58203125,
|
| 1025 |
+
"learning_rate": 5.378316646388424e-05,
|
| 1026 |
+
"loss": 0.5721,
|
| 1027 |
+
"step": 705
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 3.4634146341463414,
|
| 1031 |
+
"grad_norm": 0.578125,
|
| 1032 |
+
"learning_rate": 5.227913329649271e-05,
|
| 1033 |
+
"loss": 0.5788,
|
| 1034 |
+
"step": 710
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 3.4878048780487805,
|
| 1038 |
+
"grad_norm": 0.59765625,
|
| 1039 |
+
"learning_rate": 5.078895095690249e-05,
|
| 1040 |
+
"loss": 0.5693,
|
| 1041 |
+
"step": 715
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 3.5121951219512195,
|
| 1045 |
+
"grad_norm": 0.5703125,
|
| 1046 |
+
"learning_rate": 4.931305196573621e-05,
|
| 1047 |
+
"loss": 0.6107,
|
| 1048 |
+
"step": 720
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 3.5365853658536586,
|
| 1052 |
+
"grad_norm": 0.58203125,
|
| 1053 |
+
"learning_rate": 4.78518646979206e-05,
|
| 1054 |
+
"loss": 0.5695,
|
| 1055 |
+
"step": 725
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 3.5609756097560976,
|
| 1059 |
+
"grad_norm": 0.58984375,
|
| 1060 |
+
"learning_rate": 4.6405813258352135e-05,
|
| 1061 |
+
"loss": 0.5707,
|
| 1062 |
+
"step": 730
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 3.5853658536585367,
|
| 1066 |
+
"grad_norm": 0.62109375,
|
| 1067 |
+
"learning_rate": 4.4975317358801885e-05,
|
| 1068 |
+
"loss": 0.5635,
|
| 1069 |
+
"step": 735
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 3.6097560975609757,
|
| 1073 |
+
"grad_norm": 0.57421875,
|
| 1074 |
+
"learning_rate": 4.3560792196095543e-05,
|
| 1075 |
+
"loss": 0.5747,
|
| 1076 |
+
"step": 740
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 3.6341463414634148,
|
| 1080 |
+
"grad_norm": 0.57421875,
|
| 1081 |
+
"learning_rate": 4.216264833160396e-05,
|
| 1082 |
+
"loss": 0.5856,
|
| 1083 |
+
"step": 745
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 3.658536585365854,
|
| 1087 |
+
"grad_norm": 0.6015625,
|
| 1088 |
+
"learning_rate": 4.0781291572078806e-05,
|
| 1089 |
+
"loss": 0.5797,
|
| 1090 |
+
"step": 750
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 3.682926829268293,
|
| 1094 |
+
"grad_norm": 0.55859375,
|
| 1095 |
+
"learning_rate": 3.941712285186878e-05,
|
| 1096 |
+
"loss": 0.5643,
|
| 1097 |
+
"step": 755
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 3.7073170731707314,
|
| 1101 |
+
"grad_norm": 0.5859375,
|
| 1102 |
+
"learning_rate": 3.807053811654948e-05,
|
| 1103 |
+
"loss": 0.5723,
|
| 1104 |
+
"step": 760
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 3.7317073170731705,
|
| 1108 |
+
"grad_norm": 0.59375,
|
| 1109 |
+
"learning_rate": 3.674192820800156e-05,
|
| 1110 |
+
"loss": 0.5894,
|
| 1111 |
+
"step": 765
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 3.7560975609756095,
|
| 1115 |
+
"grad_norm": 0.55859375,
|
| 1116 |
+
"learning_rate": 3.543167875097013e-05,
|
| 1117 |
+
"loss": 0.5505,
|
| 1118 |
+
"step": 770
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 3.7804878048780486,
|
| 1122 |
+
"grad_norm": 0.5859375,
|
| 1123 |
+
"learning_rate": 3.4140170041138385e-05,
|
| 1124 |
+
"loss": 0.5752,
|
| 1125 |
+
"step": 775
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 3.8048780487804876,
|
| 1129 |
+
"grad_norm": 0.55859375,
|
| 1130 |
+
"learning_rate": 3.286777693474803e-05,
|
| 1131 |
+
"loss": 0.5629,
|
| 1132 |
+
"step": 780
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 3.8292682926829267,
|
| 1136 |
+
"grad_norm": 0.59765625,
|
| 1137 |
+
"learning_rate": 3.1614868739798495e-05,
|
| 1138 |
+
"loss": 0.5787,
|
| 1139 |
+
"step": 785
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 3.8536585365853657,
|
| 1143 |
+
"grad_norm": 0.6015625,
|
| 1144 |
+
"learning_rate": 3.0381809108856398e-05,
|
| 1145 |
+
"loss": 0.574,
|
| 1146 |
+
"step": 790
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 3.8780487804878048,
|
| 1150 |
+
"grad_norm": 0.6171875,
|
| 1151 |
+
"learning_rate": 2.9168955933506648e-05,
|
| 1152 |
+
"loss": 0.5826,
|
| 1153 |
+
"step": 795
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 3.902439024390244,
|
| 1157 |
+
"grad_norm": 0.578125,
|
| 1158 |
+
"learning_rate": 2.79766612404755e-05,
|
| 1159 |
+
"loss": 0.5725,
|
| 1160 |
+
"step": 800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 3.926829268292683,
|
| 1164 |
+
"grad_norm": 0.55859375,
|
| 1165 |
+
"learning_rate": 2.6805271089455986e-05,
|
| 1166 |
+
"loss": 0.5612,
|
| 1167 |
+
"step": 805
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 3.951219512195122,
|
| 1171 |
+
"grad_norm": 0.6328125,
|
| 1172 |
+
"learning_rate": 2.565512547266511e-05,
|
| 1173 |
+
"loss": 0.5721,
|
| 1174 |
+
"step": 810
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 3.975609756097561,
|
| 1178 |
+
"grad_norm": 0.5625,
|
| 1179 |
+
"learning_rate": 2.4526558216162322e-05,
|
| 1180 |
+
"loss": 0.5725,
|
| 1181 |
+
"step": 815
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 4.0,
|
| 1185 |
+
"grad_norm": 0.56640625,
|
| 1186 |
+
"learning_rate": 2.3419896882957527e-05,
|
| 1187 |
+
"loss": 0.5755,
|
| 1188 |
+
"step": 820
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 4.0,
|
| 1192 |
+
"eval_loss": 2.2934529781341553,
|
| 1193 |
+
"eval_runtime": 0.9995,
|
| 1194 |
+
"eval_samples_per_second": 5.002,
|
| 1195 |
+
"eval_steps_per_second": 2.001,
|
| 1196 |
+
"step": 820
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 4.024390243902439,
|
| 1200 |
+
"grad_norm": 0.55859375,
|
| 1201 |
+
"learning_rate": 2.2335462677936957e-05,
|
| 1202 |
+
"loss": 0.5324,
|
| 1203 |
+
"step": 825
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 4.048780487804878,
|
| 1207 |
+
"grad_norm": 0.61328125,
|
| 1208 |
+
"learning_rate": 2.1273570354634508e-05,
|
| 1209 |
+
"loss": 0.5386,
|
| 1210 |
+
"step": 830
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 4.073170731707317,
|
| 1214 |
+
"grad_norm": 0.55859375,
|
| 1215 |
+
"learning_rate": 2.023452812387555e-05,
|
| 1216 |
+
"loss": 0.5296,
|
| 1217 |
+
"step": 835
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 4.097560975609756,
|
| 1221 |
+
"grad_norm": 0.62890625,
|
| 1222 |
+
"learning_rate": 1.9218637564319696e-05,
|
| 1223 |
+
"loss": 0.5304,
|
| 1224 |
+
"step": 840
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 4.121951219512195,
|
| 1228 |
+
"grad_norm": 0.57421875,
|
| 1229 |
+
"learning_rate": 1.8226193534928604e-05,
|
| 1230 |
+
"loss": 0.5554,
|
| 1231 |
+
"step": 845
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 4.146341463414634,
|
| 1235 |
+
"grad_norm": 0.578125,
|
| 1236 |
+
"learning_rate": 1.725748408938408e-05,
|
| 1237 |
+
"loss": 0.5376,
|
| 1238 |
+
"step": 850
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 4.170731707317073,
|
| 1242 |
+
"grad_norm": 0.58984375,
|
| 1243 |
+
"learning_rate": 1.63127903924815e-05,
|
| 1244 |
+
"loss": 0.5335,
|
| 1245 |
+
"step": 855
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 4.195121951219512,
|
| 1249 |
+
"grad_norm": 0.5625,
|
| 1250 |
+
"learning_rate": 1.5392386638522482e-05,
|
| 1251 |
+
"loss": 0.5298,
|
| 1252 |
+
"step": 860
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 4.219512195121951,
|
| 1256 |
+
"grad_norm": 0.60546875,
|
| 1257 |
+
"learning_rate": 1.4496539971731026e-05,
|
| 1258 |
+
"loss": 0.5239,
|
| 1259 |
+
"step": 865
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 4.2439024390243905,
|
| 1263 |
+
"grad_norm": 0.57421875,
|
| 1264 |
+
"learning_rate": 1.3625510408715714e-05,
|
| 1265 |
+
"loss": 0.5255,
|
| 1266 |
+
"step": 870
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 4.2682926829268295,
|
| 1270 |
+
"grad_norm": 0.5859375,
|
| 1271 |
+
"learning_rate": 1.2779550763000703e-05,
|
| 1272 |
+
"loss": 0.5382,
|
| 1273 |
+
"step": 875
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 4.2926829268292686,
|
| 1277 |
+
"grad_norm": 0.609375,
|
| 1278 |
+
"learning_rate": 1.1958906571647421e-05,
|
| 1279 |
+
"loss": 0.5469,
|
| 1280 |
+
"step": 880
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 4.317073170731708,
|
| 1284 |
+
"grad_norm": 0.59375,
|
| 1285 |
+
"learning_rate": 1.1163816023988261e-05,
|
| 1286 |
+
"loss": 0.547,
|
| 1287 |
+
"step": 885
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 4.341463414634147,
|
| 1291 |
+
"grad_norm": 0.56640625,
|
| 1292 |
+
"learning_rate": 1.0394509892492833e-05,
|
| 1293 |
+
"loss": 0.5387,
|
| 1294 |
+
"step": 890
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 4.365853658536586,
|
| 1298 |
+
"grad_norm": 0.58203125,
|
| 1299 |
+
"learning_rate": 9.65121146578709e-06,
|
| 1300 |
+
"loss": 0.5487,
|
| 1301 |
+
"step": 895
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 4.390243902439025,
|
| 1305 |
+
"grad_norm": 0.57421875,
|
| 1306 |
+
"learning_rate": 8.934136483844391e-06,
|
| 1307 |
+
"loss": 0.5306,
|
| 1308 |
+
"step": 900
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 4.414634146341464,
|
| 1312 |
+
"grad_norm": 0.58984375,
|
| 1313 |
+
"learning_rate": 8.243493075367813e-06,
|
| 1314 |
+
"loss": 0.5259,
|
| 1315 |
+
"step": 905
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 4.439024390243903,
|
| 1319 |
+
"grad_norm": 0.5703125,
|
| 1320 |
+
"learning_rate": 7.579481697381363e-06,
|
| 1321 |
+
"loss": 0.5473,
|
| 1322 |
+
"step": 910
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 4.463414634146342,
|
| 1326 |
+
"grad_norm": 0.55859375,
|
| 1327 |
+
"learning_rate": 6.942295077048011e-06,
|
| 1328 |
+
"loss": 0.5329,
|
| 1329 |
+
"step": 915
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 4.487804878048781,
|
| 1333 |
+
"grad_norm": 0.60546875,
|
| 1334 |
+
"learning_rate": 6.3321181557312815e-06,
|
| 1335 |
+
"loss": 0.551,
|
| 1336 |
+
"step": 920
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 4.512195121951219,
|
| 1340 |
+
"grad_norm": 0.5703125,
|
| 1341 |
+
"learning_rate": 5.749128035316553e-06,
|
| 1342 |
+
"loss": 0.5508,
|
| 1343 |
+
"step": 925
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 4.536585365853659,
|
| 1347 |
+
"grad_norm": 0.59375,
|
| 1348 |
+
"learning_rate": 5.193493926807835e-06,
|
| 1349 |
+
"loss": 0.5348,
|
| 1350 |
+
"step": 930
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 4.560975609756097,
|
| 1354 |
+
"grad_norm": 0.58984375,
|
| 1355 |
+
"learning_rate": 4.665377101214863e-06,
|
| 1356 |
+
"loss": 0.5168,
|
| 1357 |
+
"step": 935
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 4.585365853658536,
|
| 1361 |
+
"grad_norm": 0.61328125,
|
| 1362 |
+
"learning_rate": 4.164930842744608e-06,
|
| 1363 |
+
"loss": 0.5442,
|
| 1364 |
+
"step": 940
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 4.609756097560975,
|
| 1368 |
+
"grad_norm": 0.59375,
|
| 1369 |
+
"learning_rate": 3.6923004043111444e-06,
|
| 1370 |
+
"loss": 0.5296,
|
| 1371 |
+
"step": 945
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 4.634146341463414,
|
| 1375 |
+
"grad_norm": 0.546875,
|
| 1376 |
+
"learning_rate": 3.2476229653763734e-06,
|
| 1377 |
+
"loss": 0.5566,
|
| 1378 |
+
"step": 950
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 4.658536585365853,
|
| 1382 |
+
"grad_norm": 0.58984375,
|
| 1383 |
+
"learning_rate": 2.8310275921341944e-06,
|
| 1384 |
+
"loss": 0.5219,
|
| 1385 |
+
"step": 955
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 4.682926829268292,
|
| 1389 |
+
"grad_norm": 0.578125,
|
| 1390 |
+
"learning_rate": 2.44263520004937e-06,
|
| 1391 |
+
"loss": 0.5577,
|
| 1392 |
+
"step": 960
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 4.7073170731707314,
|
| 1396 |
+
"grad_norm": 0.57421875,
|
| 1397 |
+
"learning_rate": 2.0825585187623007e-06,
|
| 1398 |
+
"loss": 0.5235,
|
| 1399 |
+
"step": 965
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 4.7317073170731705,
|
| 1403 |
+
"grad_norm": 0.578125,
|
| 1404 |
+
"learning_rate": 1.7509020593695302e-06,
|
| 1405 |
+
"loss": 0.547,
|
| 1406 |
+
"step": 970
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 4.7560975609756095,
|
| 1410 |
+
"grad_norm": 0.57421875,
|
| 1411 |
+
"learning_rate": 1.4477620840897766e-06,
|
| 1412 |
+
"loss": 0.5331,
|
| 1413 |
+
"step": 975
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 4.780487804878049,
|
| 1417 |
+
"grad_norm": 0.59375,
|
| 1418 |
+
"learning_rate": 1.1732265783241492e-06,
|
| 1419 |
+
"loss": 0.5251,
|
| 1420 |
+
"step": 980
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 4.804878048780488,
|
| 1424 |
+
"grad_norm": 0.60546875,
|
| 1425 |
+
"learning_rate": 9.273752251186096e-07,
|
| 1426 |
+
"loss": 0.527,
|
| 1427 |
+
"step": 985
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 4.829268292682927,
|
| 1431 |
+
"grad_norm": 0.57421875,
|
| 1432 |
+
"learning_rate": 7.102793820362829e-07,
|
| 1433 |
+
"loss": 0.5353,
|
| 1434 |
+
"step": 990
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 4.853658536585366,
|
| 1438 |
+
"grad_norm": 0.58203125,
|
| 1439 |
+
"learning_rate": 5.22002060446125e-07,
|
| 1440 |
+
"loss": 0.5347,
|
| 1441 |
+
"step": 995
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 4.878048780487805,
|
| 1445 |
+
"grad_norm": 0.58984375,
|
| 1446 |
+
"learning_rate": 3.6259790723409683e-07,
|
| 1447 |
+
"loss": 0.5436,
|
| 1448 |
+
"step": 1000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 4.902439024390244,
|
| 1452 |
+
"grad_norm": 0.5625,
|
| 1453 |
+
"learning_rate": 2.3211318894205136e-07,
|
| 1454 |
+
"loss": 0.535,
|
| 1455 |
+
"step": 1005
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 4.926829268292683,
|
| 1459 |
+
"grad_norm": 0.58984375,
|
| 1460 |
+
"learning_rate": 1.3058577833905404e-07,
|
| 1461 |
+
"loss": 0.5213,
|
| 1462 |
+
"step": 1010
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 4.951219512195122,
|
| 1466 |
+
"grad_norm": 0.55859375,
|
| 1467 |
+
"learning_rate": 5.804514342889755e-08,
|
| 1468 |
+
"loss": 0.5247,
|
| 1469 |
+
"step": 1015
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 4.975609756097561,
|
| 1473 |
+
"grad_norm": 0.60546875,
|
| 1474 |
+
"learning_rate": 1.4512338897121335e-08,
|
| 1475 |
+
"loss": 0.5333,
|
| 1476 |
+
"step": 1020
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 5.0,
|
| 1480 |
+
"grad_norm": 0.546875,
|
| 1481 |
+
"learning_rate": 0.0,
|
| 1482 |
+
"loss": 0.5144,
|
| 1483 |
+
"step": 1025
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 5.0,
|
| 1487 |
+
"eval_loss": 2.326845407485962,
|
| 1488 |
+
"eval_runtime": 1.0003,
|
| 1489 |
+
"eval_samples_per_second": 4.998,
|
| 1490 |
+
"eval_steps_per_second": 1.999,
|
| 1491 |
+
"step": 1025
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 5.0,
|
| 1495 |
+
"step": 1025,
|
| 1496 |
+
"total_flos": 1.5670950020754964e+18,
|
| 1497 |
+
"train_loss": 1.4066738275202308,
|
| 1498 |
+
"train_runtime": 8016.4558,
|
| 1499 |
+
"train_samples_per_second": 2.044,
|
| 1500 |
+
"train_steps_per_second": 0.128
|
| 1501 |
+
}
|
| 1502 |
+
],
|
| 1503 |
+
"logging_steps": 5,
|
| 1504 |
+
"max_steps": 1025,
|
| 1505 |
+
"num_input_tokens_seen": 0,
|
| 1506 |
+
"num_train_epochs": 5,
|
| 1507 |
+
"save_steps": 100,
|
| 1508 |
+
"stateful_callbacks": {
|
| 1509 |
+
"TrainerControl": {
|
| 1510 |
+
"args": {
|
| 1511 |
+
"should_epoch_stop": false,
|
| 1512 |
+
"should_evaluate": false,
|
| 1513 |
+
"should_log": false,
|
| 1514 |
+
"should_save": true,
|
| 1515 |
+
"should_training_stop": false
|
| 1516 |
+
},
|
| 1517 |
+
"attributes": {}
|
| 1518 |
+
}
|
| 1519 |
+
},
|
| 1520 |
+
"total_flos": 1.5670950020754964e+18,
|
| 1521 |
+
"train_batch_size": 4,
|
| 1522 |
+
"trial_name": null,
|
| 1523 |
+
"trial_params": null
|
| 1524 |
+
}
|