Model save
Browse files- README.md +78 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +603 -0
README.md
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: zephyr-7b-dpo-full-magpi-low-bleu-3-epochs
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# zephyr-7b-dpo-full-magpi-low-bleu-3-epochs
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.0004
|
| 21 |
+
- Rewards/chosen: -1.8844
|
| 22 |
+
- Rewards/rejected: -46.8077
|
| 23 |
+
- Rewards/accuracies: 1.0
|
| 24 |
+
- Rewards/margins: 44.9232
|
| 25 |
+
- Logps/rejected: -5321.5576
|
| 26 |
+
- Logps/chosen: -555.4259
|
| 27 |
+
- Logits/rejected: 2.7529
|
| 28 |
+
- Logits/chosen: -1.2323
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 5e-07
|
| 48 |
+
- train_batch_size: 8
|
| 49 |
+
- eval_batch_size: 8
|
| 50 |
+
- seed: 55
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 8
|
| 53 |
+
- gradient_accumulation_steps: 2
|
| 54 |
+
- total_train_batch_size: 128
|
| 55 |
+
- total_eval_batch_size: 64
|
| 56 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
+
- lr_scheduler_type: cosine
|
| 58 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 59 |
+
- num_epochs: 3
|
| 60 |
+
|
| 61 |
+
### Training results
|
| 62 |
+
|
| 63 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 64 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 65 |
+
| 0.0066 | 0.4739 | 50 | 0.0028 | -1.0908 | -33.9616 | 0.9980 | 32.8709 | -4036.9529 | -476.0595 | -1.2144 | -1.9103 |
|
| 66 |
+
| 0.0177 | 0.9479 | 100 | 0.0006 | -1.6117 | -43.9541 | 1.0 | 42.3424 | -5036.1978 | -528.1522 | 1.4562 | -2.1299 |
|
| 67 |
+
| 0.0006 | 1.4218 | 150 | 0.0004 | -1.7244 | -46.1666 | 1.0 | 44.4422 | -5257.4517 | -539.4232 | 1.6969 | -1.9837 |
|
| 68 |
+
| 0.0002 | 1.8957 | 200 | 0.0005 | -1.7575 | -44.7450 | 1.0 | 42.9875 | -5115.2886 | -542.7341 | 2.1634 | -2.0033 |
|
| 69 |
+
| 0.0001 | 2.3697 | 250 | 0.0004 | -1.8985 | -46.5225 | 1.0 | 44.6240 | -5293.0405 | -556.8339 | 2.7114 | -1.2429 |
|
| 70 |
+
| 0.0001 | 2.8436 | 300 | 0.0004 | -1.8844 | -46.8077 | 1.0 | 44.9232 | -5321.5576 | -555.4259 | 2.7529 | -1.2323 |
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
### Framework versions
|
| 74 |
+
|
| 75 |
+
- Transformers 4.44.0.dev0
|
| 76 |
+
- Pytorch 2.1.2
|
| 77 |
+
- Datasets 2.20.0
|
| 78 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.985781990521327,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.04266955489773185,
|
| 5 |
+
"train_runtime": 9090.3989,
|
| 6 |
+
"train_samples": 13500,
|
| 7 |
+
"train_samples_per_second": 4.455,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.44.0.dev0"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.985781990521327,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.04266955489773185,
|
| 5 |
+
"train_runtime": 9090.3989,
|
| 6 |
+
"train_samples": 13500,
|
| 7 |
+
"train_samples_per_second": 4.455,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,603 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.985781990521327,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 315,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0947867298578199,
|
| 13 |
+
"grad_norm": 50.81444347836179,
|
| 14 |
+
"learning_rate": 1.5624999999999999e-07,
|
| 15 |
+
"logits/chosen": -2.8273773193359375,
|
| 16 |
+
"logits/rejected": -2.573636054992676,
|
| 17 |
+
"logps/chosen": -369.3688049316406,
|
| 18 |
+
"logps/rejected": -693.6748046875,
|
| 19 |
+
"loss": 0.6858,
|
| 20 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 21 |
+
"rewards/chosen": 0.0030312505550682545,
|
| 22 |
+
"rewards/margins": 0.015196545049548149,
|
| 23 |
+
"rewards/rejected": -0.012165295891463757,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.1895734597156398,
|
| 28 |
+
"grad_norm": 17.76028531231973,
|
| 29 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 30 |
+
"logits/chosen": -2.8040361404418945,
|
| 31 |
+
"logits/rejected": -2.5344460010528564,
|
| 32 |
+
"logps/chosen": -356.6896057128906,
|
| 33 |
+
"logps/rejected": -714.5941162109375,
|
| 34 |
+
"loss": 0.4666,
|
| 35 |
+
"rewards/accuracies": 1.0,
|
| 36 |
+
"rewards/chosen": 0.1258755326271057,
|
| 37 |
+
"rewards/margins": 0.69224613904953,
|
| 38 |
+
"rewards/rejected": -0.5663706064224243,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.2843601895734597,
|
| 43 |
+
"grad_norm": 3.027708678861818,
|
| 44 |
+
"learning_rate": 4.6874999999999996e-07,
|
| 45 |
+
"logits/chosen": -2.8109562397003174,
|
| 46 |
+
"logits/rejected": -2.5492990016937256,
|
| 47 |
+
"logps/chosen": -299.7785949707031,
|
| 48 |
+
"logps/rejected": -1061.5032958984375,
|
| 49 |
+
"loss": 0.124,
|
| 50 |
+
"rewards/accuracies": 1.0,
|
| 51 |
+
"rewards/chosen": 0.4659281373023987,
|
| 52 |
+
"rewards/margins": 4.801316261291504,
|
| 53 |
+
"rewards/rejected": -4.33538818359375,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.3791469194312796,
|
| 58 |
+
"grad_norm": 4.339445107844336,
|
| 59 |
+
"learning_rate": 4.990147841143461e-07,
|
| 60 |
+
"logits/chosen": -2.6756181716918945,
|
| 61 |
+
"logits/rejected": -2.3431499004364014,
|
| 62 |
+
"logps/chosen": -309.88055419921875,
|
| 63 |
+
"logps/rejected": -2287.322265625,
|
| 64 |
+
"loss": 0.0229,
|
| 65 |
+
"rewards/accuracies": 1.0,
|
| 66 |
+
"rewards/chosen": 0.4024983048439026,
|
| 67 |
+
"rewards/margins": 16.896318435668945,
|
| 68 |
+
"rewards/rejected": -16.493820190429688,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.47393364928909953,
|
| 73 |
+
"grad_norm": 1.4582859394909966,
|
| 74 |
+
"learning_rate": 4.950256493879794e-07,
|
| 75 |
+
"logits/chosen": -2.350309371948242,
|
| 76 |
+
"logits/rejected": -1.7835966348648071,
|
| 77 |
+
"logps/chosen": -439.63720703125,
|
| 78 |
+
"logps/rejected": -3742.43408203125,
|
| 79 |
+
"loss": 0.0066,
|
| 80 |
+
"rewards/accuracies": 1.0,
|
| 81 |
+
"rewards/chosen": -0.5562213659286499,
|
| 82 |
+
"rewards/margins": 29.479211807250977,
|
| 83 |
+
"rewards/rejected": -30.035430908203125,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.47393364928909953,
|
| 88 |
+
"eval_logits/chosen": -1.910264015197754,
|
| 89 |
+
"eval_logits/rejected": -1.2144216299057007,
|
| 90 |
+
"eval_logps/chosen": -476.05950927734375,
|
| 91 |
+
"eval_logps/rejected": -4036.952880859375,
|
| 92 |
+
"eval_loss": 0.00275122607126832,
|
| 93 |
+
"eval_rewards/accuracies": 0.9979838728904724,
|
| 94 |
+
"eval_rewards/chosen": -1.0907776355743408,
|
| 95 |
+
"eval_rewards/margins": 32.87086868286133,
|
| 96 |
+
"eval_rewards/rejected": -33.96164321899414,
|
| 97 |
+
"eval_runtime": 197.0098,
|
| 98 |
+
"eval_samples_per_second": 19.821,
|
| 99 |
+
"eval_steps_per_second": 0.315,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.5687203791469194,
|
| 104 |
+
"grad_norm": 0.2408524061043023,
|
| 105 |
+
"learning_rate": 4.88020090697132e-07,
|
| 106 |
+
"logits/chosen": -1.8117077350616455,
|
| 107 |
+
"logits/rejected": -0.6155702471733093,
|
| 108 |
+
"logps/chosen": -491.74359130859375,
|
| 109 |
+
"logps/rejected": -4422.6455078125,
|
| 110 |
+
"loss": 0.0021,
|
| 111 |
+
"rewards/accuracies": 1.0,
|
| 112 |
+
"rewards/chosen": -1.4004383087158203,
|
| 113 |
+
"rewards/margins": 36.08842086791992,
|
| 114 |
+
"rewards/rejected": -37.48885726928711,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.6635071090047393,
|
| 119 |
+
"grad_norm": 0.047142917188209066,
|
| 120 |
+
"learning_rate": 4.780843509929904e-07,
|
| 121 |
+
"logits/chosen": -1.8627986907958984,
|
| 122 |
+
"logits/rejected": 0.22284331917762756,
|
| 123 |
+
"logps/chosen": -509.3787536621094,
|
| 124 |
+
"logps/rejected": -4690.046875,
|
| 125 |
+
"loss": 0.0014,
|
| 126 |
+
"rewards/accuracies": 1.0,
|
| 127 |
+
"rewards/chosen": -1.4054739475250244,
|
| 128 |
+
"rewards/margins": 39.368186950683594,
|
| 129 |
+
"rewards/rejected": -40.773658752441406,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.7582938388625592,
|
| 134 |
+
"grad_norm": 9.120624486054943,
|
| 135 |
+
"learning_rate": 4.6534074564712217e-07,
|
| 136 |
+
"logits/chosen": -0.7828740477561951,
|
| 137 |
+
"logits/rejected": 1.5713117122650146,
|
| 138 |
+
"logps/chosen": -559.362548828125,
|
| 139 |
+
"logps/rejected": -5071.544921875,
|
| 140 |
+
"loss": 0.0022,
|
| 141 |
+
"rewards/accuracies": 1.0,
|
| 142 |
+
"rewards/chosen": -1.678344488143921,
|
| 143 |
+
"rewards/margins": 42.42586135864258,
|
| 144 |
+
"rewards/rejected": -44.10420227050781,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.8530805687203792,
|
| 149 |
+
"grad_norm": 0.11081628927270178,
|
| 150 |
+
"learning_rate": 4.4994615667026846e-07,
|
| 151 |
+
"logits/chosen": -1.3702471256256104,
|
| 152 |
+
"logits/rejected": 2.505129337310791,
|
| 153 |
+
"logps/chosen": -544.5760498046875,
|
| 154 |
+
"logps/rejected": -5206.58203125,
|
| 155 |
+
"loss": 0.0017,
|
| 156 |
+
"rewards/accuracies": 1.0,
|
| 157 |
+
"rewards/chosen": -1.5564204454421997,
|
| 158 |
+
"rewards/margins": 44.48157501220703,
|
| 159 |
+
"rewards/rejected": -46.03799057006836,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.9478672985781991,
|
| 164 |
+
"grad_norm": 0.3080840911945339,
|
| 165 |
+
"learning_rate": 4.320901013934887e-07,
|
| 166 |
+
"logits/chosen": -2.1306087970733643,
|
| 167 |
+
"logits/rejected": 1.8308042287826538,
|
| 168 |
+
"logps/chosen": -503.85980224609375,
|
| 169 |
+
"logps/rejected": -5392.4296875,
|
| 170 |
+
"loss": 0.0177,
|
| 171 |
+
"rewards/accuracies": 1.0,
|
| 172 |
+
"rewards/chosen": -1.653719186782837,
|
| 173 |
+
"rewards/margins": 45.56684112548828,
|
| 174 |
+
"rewards/rejected": -47.22056198120117,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.9478672985781991,
|
| 179 |
+
"eval_logits/chosen": -2.1299259662628174,
|
| 180 |
+
"eval_logits/rejected": 1.4562028646469116,
|
| 181 |
+
"eval_logps/chosen": -528.1521606445312,
|
| 182 |
+
"eval_logps/rejected": -5036.19775390625,
|
| 183 |
+
"eval_loss": 0.0005923541029915214,
|
| 184 |
+
"eval_rewards/accuracies": 1.0,
|
| 185 |
+
"eval_rewards/chosen": -1.6117043495178223,
|
| 186 |
+
"eval_rewards/margins": 42.342384338378906,
|
| 187 |
+
"eval_rewards/rejected": -43.95408630371094,
|
| 188 |
+
"eval_runtime": 193.3087,
|
| 189 |
+
"eval_samples_per_second": 20.201,
|
| 190 |
+
"eval_steps_per_second": 0.321,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.042654028436019,
|
| 195 |
+
"grad_norm": 18.444713565032494,
|
| 196 |
+
"learning_rate": 4.119923993874379e-07,
|
| 197 |
+
"logits/chosen": -1.6922178268432617,
|
| 198 |
+
"logits/rejected": 2.0268759727478027,
|
| 199 |
+
"logps/chosen": -523.6668090820312,
|
| 200 |
+
"logps/rejected": -5431.6923828125,
|
| 201 |
+
"loss": 0.0056,
|
| 202 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 203 |
+
"rewards/chosen": -1.692284345626831,
|
| 204 |
+
"rewards/margins": 45.80804443359375,
|
| 205 |
+
"rewards/rejected": -47.500328063964844,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.1374407582938388,
|
| 210 |
+
"grad_norm": 0.667964087575959,
|
| 211 |
+
"learning_rate": 3.899004663415083e-07,
|
| 212 |
+
"logits/chosen": -1.945476770401001,
|
| 213 |
+
"logits/rejected": 2.063563346862793,
|
| 214 |
+
"logps/chosen": -512.0524291992188,
|
| 215 |
+
"logps/rejected": -5017.99755859375,
|
| 216 |
+
"loss": 0.0008,
|
| 217 |
+
"rewards/accuracies": 1.0,
|
| 218 |
+
"rewards/chosen": -1.6633787155151367,
|
| 219 |
+
"rewards/margins": 41.92264175415039,
|
| 220 |
+
"rewards/rejected": -43.586021423339844,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.2322274881516588,
|
| 225 |
+
"grad_norm": 0.07609835769363717,
|
| 226 |
+
"learning_rate": 3.6608626821692824e-07,
|
| 227 |
+
"logits/chosen": -1.3765870332717896,
|
| 228 |
+
"logits/rejected": 2.4521493911743164,
|
| 229 |
+
"logps/chosen": -511.48992919921875,
|
| 230 |
+
"logps/rejected": -5102.3095703125,
|
| 231 |
+
"loss": 0.0003,
|
| 232 |
+
"rewards/accuracies": 1.0,
|
| 233 |
+
"rewards/chosen": -1.574892282485962,
|
| 234 |
+
"rewards/margins": 42.903465270996094,
|
| 235 |
+
"rewards/rejected": -44.47835922241211,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.3270142180094786,
|
| 240 |
+
"grad_norm": 0.056132145026876815,
|
| 241 |
+
"learning_rate": 3.408429731701635e-07,
|
| 242 |
+
"logits/chosen": -1.673305869102478,
|
| 243 |
+
"logits/rejected": 2.8528292179107666,
|
| 244 |
+
"logps/chosen": -515.4207763671875,
|
| 245 |
+
"logps/rejected": -5185.7666015625,
|
| 246 |
+
"loss": 0.0021,
|
| 247 |
+
"rewards/accuracies": 1.0,
|
| 248 |
+
"rewards/chosen": -1.6433398723602295,
|
| 249 |
+
"rewards/margins": 44.202003479003906,
|
| 250 |
+
"rewards/rejected": -45.84534454345703,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.4218009478672986,
|
| 255 |
+
"grad_norm": 0.23322034769023395,
|
| 256 |
+
"learning_rate": 3.144813424636031e-07,
|
| 257 |
+
"logits/chosen": -2.0590405464172363,
|
| 258 |
+
"logits/rejected": 2.5171058177948,
|
| 259 |
+
"logps/chosen": -564.9078369140625,
|
| 260 |
+
"logps/rejected": -5421.3935546875,
|
| 261 |
+
"loss": 0.0006,
|
| 262 |
+
"rewards/accuracies": 1.0,
|
| 263 |
+
"rewards/chosen": -1.7184299230575562,
|
| 264 |
+
"rewards/margins": 46.220848083496094,
|
| 265 |
+
"rewards/rejected": -47.93927764892578,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.4218009478672986,
|
| 270 |
+
"eval_logits/chosen": -1.983699083328247,
|
| 271 |
+
"eval_logits/rejected": 1.6969449520111084,
|
| 272 |
+
"eval_logps/chosen": -539.4231567382812,
|
| 273 |
+
"eval_logps/rejected": -5257.45166015625,
|
| 274 |
+
"eval_loss": 0.00044810696272179484,
|
| 275 |
+
"eval_rewards/accuracies": 1.0,
|
| 276 |
+
"eval_rewards/chosen": -1.7244139909744263,
|
| 277 |
+
"eval_rewards/margins": 44.44221878051758,
|
| 278 |
+
"eval_rewards/rejected": -46.166629791259766,
|
| 279 |
+
"eval_runtime": 194.364,
|
| 280 |
+
"eval_samples_per_second": 20.091,
|
| 281 |
+
"eval_steps_per_second": 0.319,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.5165876777251186,
|
| 286 |
+
"grad_norm": 0.12179595633717599,
|
| 287 |
+
"learning_rate": 2.8732590479375165e-07,
|
| 288 |
+
"logits/chosen": -1.6932016611099243,
|
| 289 |
+
"logits/rejected": 3.00923752784729,
|
| 290 |
+
"logps/chosen": -561.7379150390625,
|
| 291 |
+
"logps/rejected": -5284.3369140625,
|
| 292 |
+
"loss": 0.0002,
|
| 293 |
+
"rewards/accuracies": 1.0,
|
| 294 |
+
"rewards/chosen": -1.8306872844696045,
|
| 295 |
+
"rewards/margins": 45.2327995300293,
|
| 296 |
+
"rewards/rejected": -47.06348419189453,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 1.6113744075829384,
|
| 301 |
+
"grad_norm": 0.0933485824938723,
|
| 302 |
+
"learning_rate": 2.597109611334169e-07,
|
| 303 |
+
"logits/chosen": -1.3871994018554688,
|
| 304 |
+
"logits/rejected": 3.261793613433838,
|
| 305 |
+
"logps/chosen": -551.7293090820312,
|
| 306 |
+
"logps/rejected": -5060.6259765625,
|
| 307 |
+
"loss": 0.0005,
|
| 308 |
+
"rewards/accuracies": 1.0,
|
| 309 |
+
"rewards/chosen": -1.971683144569397,
|
| 310 |
+
"rewards/margins": 42.35912322998047,
|
| 311 |
+
"rewards/rejected": -44.33080291748047,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.7061611374407581,
|
| 316 |
+
"grad_norm": 3.1974087952825787,
|
| 317 |
+
"learning_rate": 2.3197646927086694e-07,
|
| 318 |
+
"logits/chosen": -1.4737141132354736,
|
| 319 |
+
"logits/rejected": 3.022137403488159,
|
| 320 |
+
"logps/chosen": -511.197509765625,
|
| 321 |
+
"logps/rejected": -5292.6005859375,
|
| 322 |
+
"loss": 0.0008,
|
| 323 |
+
"rewards/accuracies": 1.0,
|
| 324 |
+
"rewards/chosen": -1.733986258506775,
|
| 325 |
+
"rewards/margins": 45.09421157836914,
|
| 326 |
+
"rewards/rejected": -46.82819366455078,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 1.8009478672985781,
|
| 331 |
+
"grad_norm": 0.027408737628248046,
|
| 332 |
+
"learning_rate": 2.0446385870993467e-07,
|
| 333 |
+
"logits/chosen": -2.2061755657196045,
|
| 334 |
+
"logits/rejected": 2.45582914352417,
|
| 335 |
+
"logps/chosen": -533.607177734375,
|
| 336 |
+
"logps/rejected": -5150.46044921875,
|
| 337 |
+
"loss": 0.0003,
|
| 338 |
+
"rewards/accuracies": 1.0,
|
| 339 |
+
"rewards/chosen": -1.7912607192993164,
|
| 340 |
+
"rewards/margins": 43.65951919555664,
|
| 341 |
+
"rewards/rejected": -45.45077896118164,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.8957345971563981,
|
| 346 |
+
"grad_norm": 0.030194576861770926,
|
| 347 |
+
"learning_rate": 1.775118274523545e-07,
|
| 348 |
+
"logits/chosen": -2.1220927238464355,
|
| 349 |
+
"logits/rejected": 2.6763927936553955,
|
| 350 |
+
"logps/chosen": -580.6775512695312,
|
| 351 |
+
"logps/rejected": -5746.35791015625,
|
| 352 |
+
"loss": 0.0002,
|
| 353 |
+
"rewards/accuracies": 1.0,
|
| 354 |
+
"rewards/chosen": -1.894126534461975,
|
| 355 |
+
"rewards/margins": 48.489967346191406,
|
| 356 |
+
"rewards/rejected": -50.38408660888672,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 1.8957345971563981,
|
| 361 |
+
"eval_logits/chosen": -2.0032970905303955,
|
| 362 |
+
"eval_logits/rejected": 2.1634280681610107,
|
| 363 |
+
"eval_logps/chosen": -542.7340698242188,
|
| 364 |
+
"eval_logps/rejected": -5115.28857421875,
|
| 365 |
+
"eval_loss": 0.0005020965472795069,
|
| 366 |
+
"eval_rewards/accuracies": 1.0,
|
| 367 |
+
"eval_rewards/chosen": -1.7575234174728394,
|
| 368 |
+
"eval_rewards/margins": 42.98747634887695,
|
| 369 |
+
"eval_rewards/rejected": -44.744998931884766,
|
| 370 |
+
"eval_runtime": 192.9625,
|
| 371 |
+
"eval_samples_per_second": 20.237,
|
| 372 |
+
"eval_steps_per_second": 0.321,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 1.9905213270142181,
|
| 377 |
+
"grad_norm": 0.649351349290712,
|
| 378 |
+
"learning_rate": 1.514521724066537e-07,
|
| 379 |
+
"logits/chosen": -1.6789367198944092,
|
| 380 |
+
"logits/rejected": 3.430915355682373,
|
| 381 |
+
"logps/chosen": -524.8860473632812,
|
| 382 |
+
"logps/rejected": -4889.97998046875,
|
| 383 |
+
"loss": 0.0003,
|
| 384 |
+
"rewards/accuracies": 1.0,
|
| 385 |
+
"rewards/chosen": -1.7625246047973633,
|
| 386 |
+
"rewards/margins": 40.506309509277344,
|
| 387 |
+
"rewards/rejected": -42.268829345703125,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 2.085308056872038,
|
| 392 |
+
"grad_norm": 0.04035446049031933,
|
| 393 |
+
"learning_rate": 1.266057047539568e-07,
|
| 394 |
+
"logits/chosen": -1.4461164474487305,
|
| 395 |
+
"logits/rejected": 3.0876190662384033,
|
| 396 |
+
"logps/chosen": -525.4357299804688,
|
| 397 |
+
"logps/rejected": -4884.6865234375,
|
| 398 |
+
"loss": 0.0004,
|
| 399 |
+
"rewards/accuracies": 1.0,
|
| 400 |
+
"rewards/chosen": -1.8402020931243896,
|
| 401 |
+
"rewards/margins": 40.36837387084961,
|
| 402 |
+
"rewards/rejected": -42.208580017089844,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 2.1800947867298577,
|
| 407 |
+
"grad_norm": 0.015128682487430115,
|
| 408 |
+
"learning_rate": 1.032783005551884e-07,
|
| 409 |
+
"logits/chosen": -1.137474775314331,
|
| 410 |
+
"logits/rejected": 3.6721444129943848,
|
| 411 |
+
"logps/chosen": -568.20751953125,
|
| 412 |
+
"logps/rejected": -5129.24658203125,
|
| 413 |
+
"loss": 0.0001,
|
| 414 |
+
"rewards/accuracies": 1.0,
|
| 415 |
+
"rewards/chosen": -1.940818190574646,
|
| 416 |
+
"rewards/margins": 42.981773376464844,
|
| 417 |
+
"rewards/rejected": -44.92259216308594,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 2.2748815165876777,
|
| 422 |
+
"grad_norm": 0.037321954682957494,
|
| 423 |
+
"learning_rate": 8.175713521924976e-08,
|
| 424 |
+
"logits/chosen": -1.189206838607788,
|
| 425 |
+
"logits/rejected": 3.206519603729248,
|
| 426 |
+
"logps/chosen": -552.6511840820312,
|
| 427 |
+
"logps/rejected": -5748.79541015625,
|
| 428 |
+
"loss": 0.0001,
|
| 429 |
+
"rewards/accuracies": 1.0,
|
| 430 |
+
"rewards/chosen": -1.8525272607803345,
|
| 431 |
+
"rewards/margins": 48.330223083496094,
|
| 432 |
+
"rewards/rejected": -50.1827507019043,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 2.3696682464454977,
|
| 437 |
+
"grad_norm": 0.006059762174940937,
|
| 438 |
+
"learning_rate": 6.230714818829733e-08,
|
| 439 |
+
"logits/chosen": -0.9478242993354797,
|
| 440 |
+
"logits/rejected": 3.6084961891174316,
|
| 441 |
+
"logps/chosen": -529.2486572265625,
|
| 442 |
+
"logps/rejected": -5455.07421875,
|
| 443 |
+
"loss": 0.0001,
|
| 444 |
+
"rewards/accuracies": 1.0,
|
| 445 |
+
"rewards/chosen": -1.7955585718154907,
|
| 446 |
+
"rewards/margins": 46.23430252075195,
|
| 447 |
+
"rewards/rejected": -48.02985382080078,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 2.3696682464454977,
|
| 452 |
+
"eval_logits/chosen": -1.2428650856018066,
|
| 453 |
+
"eval_logits/rejected": 2.7113900184631348,
|
| 454 |
+
"eval_logps/chosen": -556.8338623046875,
|
| 455 |
+
"eval_logps/rejected": -5293.04052734375,
|
| 456 |
+
"eval_loss": 0.00038583340938203037,
|
| 457 |
+
"eval_rewards/accuracies": 1.0,
|
| 458 |
+
"eval_rewards/chosen": -1.8985214233398438,
|
| 459 |
+
"eval_rewards/margins": 44.623992919921875,
|
| 460 |
+
"eval_rewards/rejected": -46.522518157958984,
|
| 461 |
+
"eval_runtime": 194.5012,
|
| 462 |
+
"eval_samples_per_second": 20.077,
|
| 463 |
+
"eval_steps_per_second": 0.319,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.4644549763033177,
|
| 468 |
+
"grad_norm": 0.03581328161307903,
|
| 469 |
+
"learning_rate": 4.516778136213037e-08,
|
| 470 |
+
"logits/chosen": -0.6857299208641052,
|
| 471 |
+
"logits/rejected": 4.016716003417969,
|
| 472 |
+
"logps/chosen": -546.9486083984375,
|
| 473 |
+
"logps/rejected": -6013.9873046875,
|
| 474 |
+
"loss": 0.0001,
|
| 475 |
+
"rewards/accuracies": 1.0,
|
| 476 |
+
"rewards/chosen": -2.029759407043457,
|
| 477 |
+
"rewards/margins": 51.3958740234375,
|
| 478 |
+
"rewards/rejected": -53.425636291503906,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 2.5592417061611377,
|
| 483 |
+
"grad_norm": 0.022468211975921346,
|
| 484 |
+
"learning_rate": 3.055003141378948e-08,
|
| 485 |
+
"logits/chosen": -1.1639906167984009,
|
| 486 |
+
"logits/rejected": 3.84511137008667,
|
| 487 |
+
"logps/chosen": -576.4962158203125,
|
| 488 |
+
"logps/rejected": -5156.3203125,
|
| 489 |
+
"loss": 0.0001,
|
| 490 |
+
"rewards/accuracies": 1.0,
|
| 491 |
+
"rewards/chosen": -1.8559118509292603,
|
| 492 |
+
"rewards/margins": 43.48760223388672,
|
| 493 |
+
"rewards/rejected": -45.34351348876953,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.654028436018957,
|
| 498 |
+
"grad_norm": 0.04077948274793423,
|
| 499 |
+
"learning_rate": 1.8633852284264508e-08,
|
| 500 |
+
"logits/chosen": -0.8976588249206543,
|
| 501 |
+
"logits/rejected": 3.54345440864563,
|
| 502 |
+
"logps/chosen": -536.1121826171875,
|
| 503 |
+
"logps/rejected": -5684.75048828125,
|
| 504 |
+
"loss": 0.0001,
|
| 505 |
+
"rewards/accuracies": 1.0,
|
| 506 |
+
"rewards/chosen": -1.8762388229370117,
|
| 507 |
+
"rewards/margins": 48.596927642822266,
|
| 508 |
+
"rewards/rejected": -50.473167419433594,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 2.748815165876777,
|
| 513 |
+
"grad_norm": 0.025754086632460663,
|
| 514 |
+
"learning_rate": 9.56593983327919e-09,
|
| 515 |
+
"logits/chosen": -1.061200737953186,
|
| 516 |
+
"logits/rejected": 3.624175548553467,
|
| 517 |
+
"logps/chosen": -544.1278076171875,
|
| 518 |
+
"logps/rejected": -5561.8818359375,
|
| 519 |
+
"loss": 0.0002,
|
| 520 |
+
"rewards/accuracies": 1.0,
|
| 521 |
+
"rewards/chosen": -1.8096107244491577,
|
| 522 |
+
"rewards/margins": 47.35784149169922,
|
| 523 |
+
"rewards/rejected": -49.16745376586914,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.843601895734597,
|
| 528 |
+
"grad_norm": 0.00972787744038685,
|
| 529 |
+
"learning_rate": 3.4579259185321398e-09,
|
| 530 |
+
"logits/chosen": -0.9289520978927612,
|
| 531 |
+
"logits/rejected": 3.9043102264404297,
|
| 532 |
+
"logps/chosen": -559.0933837890625,
|
| 533 |
+
"logps/rejected": -5231.408203125,
|
| 534 |
+
"loss": 0.0001,
|
| 535 |
+
"rewards/accuracies": 1.0,
|
| 536 |
+
"rewards/chosen": -1.9514116048812866,
|
| 537 |
+
"rewards/margins": 43.472190856933594,
|
| 538 |
+
"rewards/rejected": -45.423606872558594,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 2.843601895734597,
|
| 543 |
+
"eval_logits/chosen": -1.2322728633880615,
|
| 544 |
+
"eval_logits/rejected": 2.752917766571045,
|
| 545 |
+
"eval_logps/chosen": -555.4259033203125,
|
| 546 |
+
"eval_logps/rejected": -5321.5576171875,
|
| 547 |
+
"eval_loss": 0.0003751559997908771,
|
| 548 |
+
"eval_rewards/accuracies": 1.0,
|
| 549 |
+
"eval_rewards/chosen": -1.8844420909881592,
|
| 550 |
+
"eval_rewards/margins": 44.92324447631836,
|
| 551 |
+
"eval_rewards/rejected": -46.80768585205078,
|
| 552 |
+
"eval_runtime": 193.0161,
|
| 553 |
+
"eval_samples_per_second": 20.231,
|
| 554 |
+
"eval_steps_per_second": 0.321,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 2.938388625592417,
|
| 559 |
+
"grad_norm": 0.07731963429515382,
|
| 560 |
+
"learning_rate": 3.850041354441502e-10,
|
| 561 |
+
"logits/chosen": -0.9526262283325195,
|
| 562 |
+
"logits/rejected": 3.7660250663757324,
|
| 563 |
+
"logps/chosen": -527.6392822265625,
|
| 564 |
+
"logps/rejected": -5758.5341796875,
|
| 565 |
+
"loss": 0.0001,
|
| 566 |
+
"rewards/accuracies": 1.0,
|
| 567 |
+
"rewards/chosen": -1.9126548767089844,
|
| 568 |
+
"rewards/margins": 49.44573211669922,
|
| 569 |
+
"rewards/rejected": -51.3583869934082,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 2.985781990521327,
|
| 574 |
+
"step": 315,
|
| 575 |
+
"total_flos": 0.0,
|
| 576 |
+
"train_loss": 0.04266955489773185,
|
| 577 |
+
"train_runtime": 9090.3989,
|
| 578 |
+
"train_samples_per_second": 4.455,
|
| 579 |
+
"train_steps_per_second": 0.035
|
| 580 |
+
}
|
| 581 |
+
],
|
| 582 |
+
"logging_steps": 10,
|
| 583 |
+
"max_steps": 315,
|
| 584 |
+
"num_input_tokens_seen": 0,
|
| 585 |
+
"num_train_epochs": 3,
|
| 586 |
+
"save_steps": 100,
|
| 587 |
+
"stateful_callbacks": {
|
| 588 |
+
"TrainerControl": {
|
| 589 |
+
"args": {
|
| 590 |
+
"should_epoch_stop": false,
|
| 591 |
+
"should_evaluate": false,
|
| 592 |
+
"should_log": false,
|
| 593 |
+
"should_save": true,
|
| 594 |
+
"should_training_stop": true
|
| 595 |
+
},
|
| 596 |
+
"attributes": {}
|
| 597 |
+
}
|
| 598 |
+
},
|
| 599 |
+
"total_flos": 0.0,
|
| 600 |
+
"train_batch_size": 8,
|
| 601 |
+
"trial_name": null,
|
| 602 |
+
"trial_params": null
|
| 603 |
+
}
|