Model save
Browse files- README.md +79 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +679 -0
README.md
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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-reward-scale-05
|
| 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-reward-scale-05
|
| 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: 0.0006
|
| 21 |
+
- Rewards/chosen: -1.9421
|
| 22 |
+
- Rewards/rejected: -75.0558
|
| 23 |
+
- Rewards/accuracies: 1.0
|
| 24 |
+
- Rewards/margins: 73.1137
|
| 25 |
+
- Logps/rejected: -8146.3696
|
| 26 |
+
- Logps/chosen: -561.1917
|
| 27 |
+
- Logits/rejected: 2.3223
|
| 28 |
+
- Logits/chosen: -0.9088
|
| 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: 1
|
| 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 |
+
| 0.0079 | 0.1420 | 50 | 0.0052 | -1.4109 | -51.3621 | 0.9960 | 49.9512 | -5776.9990 | -508.0702 | -2.4011 | -2.8844 |
|
| 66 |
+
| 0.0031 | 0.2841 | 100 | 0.0012 | -2.0608 | -82.9205 | 1.0 | 80.8596 | -8932.8350 | -573.0657 | 0.0172 | -2.4172 |
|
| 67 |
+
| 0.0016 | 0.4261 | 150 | 0.0008 | -2.0420 | -78.8194 | 1.0 | 76.7774 | -8522.7256 | -571.1802 | 1.8918 | -2.1991 |
|
| 68 |
+
| 0.0015 | 0.5682 | 200 | 0.0007 | -1.9757 | -79.8216 | 1.0 | 77.8459 | -8622.9443 | -564.5500 | 2.5732 | -0.8412 |
|
| 69 |
+
| 0.0016 | 0.7102 | 250 | 0.0008 | -1.8298 | -70.1341 | 1.0 | 68.3043 | -7654.1978 | -549.9620 | 2.2580 | -1.2005 |
|
| 70 |
+
| 0.0008 | 0.8523 | 300 | 0.0006 | -1.9079 | -74.3261 | 1.0 | 72.4182 | -8073.3999 | -557.7685 | 2.2921 | -0.9558 |
|
| 71 |
+
| 0.0029 | 0.9943 | 350 | 0.0006 | -1.9421 | -75.0558 | 1.0 | 73.1137 | -8146.3696 | -561.1917 | 2.3223 | -0.9088 |
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
### Framework versions
|
| 75 |
+
|
| 76 |
+
- Transformers 4.44.0.dev0
|
| 77 |
+
- Pytorch 2.1.2
|
| 78 |
+
- Datasets 2.20.0
|
| 79 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.0465552191166022,
|
| 5 |
+
"train_runtime": 10297.8682,
|
| 6 |
+
"train_samples": 45000,
|
| 7 |
+
"train_samples_per_second": 4.37,
|
| 8 |
+
"train_steps_per_second": 0.034
|
| 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": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.0465552191166022,
|
| 5 |
+
"train_runtime": 10297.8682,
|
| 6 |
+
"train_samples": 45000,
|
| 7 |
+
"train_samples_per_second": 4.37,
|
| 8 |
+
"train_steps_per_second": 0.034
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,679 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 352,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.028409090909090908,
|
| 13 |
+
"grad_norm": 31.782667280345194,
|
| 14 |
+
"learning_rate": 1.3888888888888888e-07,
|
| 15 |
+
"logits/chosen": -2.8591694831848145,
|
| 16 |
+
"logits/rejected": -2.6428685188293457,
|
| 17 |
+
"logps/chosen": -390.5384216308594,
|
| 18 |
+
"logps/rejected": -607.8155517578125,
|
| 19 |
+
"loss": 0.6897,
|
| 20 |
+
"rewards/accuracies": 0.581250011920929,
|
| 21 |
+
"rewards/chosen": 0.0022137626074254513,
|
| 22 |
+
"rewards/margins": 0.013292843475937843,
|
| 23 |
+
"rewards/rejected": -0.011079080402851105,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.056818181818181816,
|
| 28 |
+
"grad_norm": 18.732725603737485,
|
| 29 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 30 |
+
"logits/chosen": -2.8422160148620605,
|
| 31 |
+
"logits/rejected": -2.694746494293213,
|
| 32 |
+
"logps/chosen": -328.73382568359375,
|
| 33 |
+
"logps/rejected": -775.1841430664062,
|
| 34 |
+
"loss": 0.556,
|
| 35 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 36 |
+
"rewards/chosen": 0.08411312848329544,
|
| 37 |
+
"rewards/margins": 0.6615578532218933,
|
| 38 |
+
"rewards/rejected": -0.5774446725845337,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.08522727272727272,
|
| 43 |
+
"grad_norm": 5.623138797985183,
|
| 44 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 45 |
+
"logits/chosen": -2.8572211265563965,
|
| 46 |
+
"logits/rejected": -2.6727840900421143,
|
| 47 |
+
"logps/chosen": -296.5441589355469,
|
| 48 |
+
"logps/rejected": -1129.43017578125,
|
| 49 |
+
"loss": 0.2375,
|
| 50 |
+
"rewards/accuracies": 1.0,
|
| 51 |
+
"rewards/chosen": 0.3573785722255707,
|
| 52 |
+
"rewards/margins": 4.872136116027832,
|
| 53 |
+
"rewards/rejected": -4.5147576332092285,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.11363636363636363,
|
| 58 |
+
"grad_norm": 1.8250836888812216,
|
| 59 |
+
"learning_rate": 4.998023493068254e-07,
|
| 60 |
+
"logits/chosen": -2.813652753829956,
|
| 61 |
+
"logits/rejected": -2.606616258621216,
|
| 62 |
+
"logps/chosen": -323.4709167480469,
|
| 63 |
+
"logps/rejected": -2253.32470703125,
|
| 64 |
+
"loss": 0.0646,
|
| 65 |
+
"rewards/accuracies": 1.0,
|
| 66 |
+
"rewards/chosen": 0.23284561932086945,
|
| 67 |
+
"rewards/margins": 15.6871337890625,
|
| 68 |
+
"rewards/rejected": -15.4542875289917,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.14204545454545456,
|
| 73 |
+
"grad_norm": 1.2011523237420525,
|
| 74 |
+
"learning_rate": 4.975823666181255e-07,
|
| 75 |
+
"logits/chosen": -2.769263744354248,
|
| 76 |
+
"logits/rejected": -2.4520652294158936,
|
| 77 |
+
"logps/chosen": -483.3968200683594,
|
| 78 |
+
"logps/rejected": -4912.21728515625,
|
| 79 |
+
"loss": 0.0079,
|
| 80 |
+
"rewards/accuracies": 1.0,
|
| 81 |
+
"rewards/chosen": -1.2076160907745361,
|
| 82 |
+
"rewards/margins": 41.37514877319336,
|
| 83 |
+
"rewards/rejected": -42.582763671875,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.14204545454545456,
|
| 88 |
+
"eval_logits/chosen": -2.8844106197357178,
|
| 89 |
+
"eval_logits/rejected": -2.4010605812072754,
|
| 90 |
+
"eval_logps/chosen": -508.0701904296875,
|
| 91 |
+
"eval_logps/rejected": -5776.9990234375,
|
| 92 |
+
"eval_loss": 0.005187372677028179,
|
| 93 |
+
"eval_rewards/accuracies": 0.9959677457809448,
|
| 94 |
+
"eval_rewards/chosen": -1.4108844995498657,
|
| 95 |
+
"eval_rewards/margins": 49.95121765136719,
|
| 96 |
+
"eval_rewards/rejected": -51.36210250854492,
|
| 97 |
+
"eval_runtime": 197.1461,
|
| 98 |
+
"eval_samples_per_second": 19.808,
|
| 99 |
+
"eval_steps_per_second": 0.314,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.17045454545454544,
|
| 104 |
+
"grad_norm": 0.511680193897244,
|
| 105 |
+
"learning_rate": 4.929173350101024e-07,
|
| 106 |
+
"logits/chosen": -2.8946661949157715,
|
| 107 |
+
"logits/rejected": -2.239091396331787,
|
| 108 |
+
"logps/chosen": -522.5350341796875,
|
| 109 |
+
"logps/rejected": -6444.09033203125,
|
| 110 |
+
"loss": 0.0077,
|
| 111 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 112 |
+
"rewards/chosen": -1.6264005899429321,
|
| 113 |
+
"rewards/margins": 56.385963439941406,
|
| 114 |
+
"rewards/rejected": -58.012359619140625,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.19886363636363635,
|
| 119 |
+
"grad_norm": 0.2533965055751647,
|
| 120 |
+
"learning_rate": 4.858533249305336e-07,
|
| 121 |
+
"logits/chosen": -2.7772889137268066,
|
| 122 |
+
"logits/rejected": -1.765015959739685,
|
| 123 |
+
"logps/chosen": -555.8414306640625,
|
| 124 |
+
"logps/rejected": -6928.0419921875,
|
| 125 |
+
"loss": 0.0037,
|
| 126 |
+
"rewards/accuracies": 1.0,
|
| 127 |
+
"rewards/chosen": -1.7668079137802124,
|
| 128 |
+
"rewards/margins": 61.386253356933594,
|
| 129 |
+
"rewards/rejected": -63.1530647277832,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.22727272727272727,
|
| 134 |
+
"grad_norm": 5.78873598153483,
|
| 135 |
+
"learning_rate": 4.764600984163808e-07,
|
| 136 |
+
"logits/chosen": -2.8613221645355225,
|
| 137 |
+
"logits/rejected": -1.6336866617202759,
|
| 138 |
+
"logps/chosen": -575.8856201171875,
|
| 139 |
+
"logps/rejected": -8230.505859375,
|
| 140 |
+
"loss": 0.003,
|
| 141 |
+
"rewards/accuracies": 1.0,
|
| 142 |
+
"rewards/chosen": -1.6956470012664795,
|
| 143 |
+
"rewards/margins": 73.87208557128906,
|
| 144 |
+
"rewards/rejected": -75.56773376464844,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.2556818181818182,
|
| 149 |
+
"grad_norm": 0.05441796989473776,
|
| 150 |
+
"learning_rate": 4.6483042014491527e-07,
|
| 151 |
+
"logits/chosen": -2.866097927093506,
|
| 152 |
+
"logits/rejected": -1.7091907262802124,
|
| 153 |
+
"logps/chosen": -556.7591552734375,
|
| 154 |
+
"logps/rejected": -7405.0654296875,
|
| 155 |
+
"loss": 0.0112,
|
| 156 |
+
"rewards/accuracies": 1.0,
|
| 157 |
+
"rewards/chosen": -1.946458101272583,
|
| 158 |
+
"rewards/margins": 65.7303466796875,
|
| 159 |
+
"rewards/rejected": -67.67679595947266,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.2840909090909091,
|
| 164 |
+
"grad_norm": 1.0199863397394129,
|
| 165 |
+
"learning_rate": 4.510791413176912e-07,
|
| 166 |
+
"logits/chosen": -2.6974339485168457,
|
| 167 |
+
"logits/rejected": -0.42245426774024963,
|
| 168 |
+
"logps/chosen": -575.6227416992188,
|
| 169 |
+
"logps/rejected": -9247.611328125,
|
| 170 |
+
"loss": 0.0031,
|
| 171 |
+
"rewards/accuracies": 1.0,
|
| 172 |
+
"rewards/chosen": -2.0615272521972656,
|
| 173 |
+
"rewards/margins": 83.35676574707031,
|
| 174 |
+
"rewards/rejected": -85.41829681396484,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.2840909090909091,
|
| 179 |
+
"eval_logits/chosen": -2.4172048568725586,
|
| 180 |
+
"eval_logits/rejected": 0.017224134877324104,
|
| 181 |
+
"eval_logps/chosen": -573.0657348632812,
|
| 182 |
+
"eval_logps/rejected": -8932.8349609375,
|
| 183 |
+
"eval_loss": 0.001179259386844933,
|
| 184 |
+
"eval_rewards/accuracies": 1.0,
|
| 185 |
+
"eval_rewards/chosen": -2.060839891433716,
|
| 186 |
+
"eval_rewards/margins": 80.85962677001953,
|
| 187 |
+
"eval_rewards/rejected": -82.92045593261719,
|
| 188 |
+
"eval_runtime": 196.8335,
|
| 189 |
+
"eval_samples_per_second": 19.839,
|
| 190 |
+
"eval_steps_per_second": 0.315,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.3125,
|
| 195 |
+
"grad_norm": 0.0720935751683844,
|
| 196 |
+
"learning_rate": 4.353420654246546e-07,
|
| 197 |
+
"logits/chosen": -2.386807441711426,
|
| 198 |
+
"logits/rejected": -0.0012192248832434416,
|
| 199 |
+
"logps/chosen": -584.6697998046875,
|
| 200 |
+
"logps/rejected": -7824.6484375,
|
| 201 |
+
"loss": 0.0034,
|
| 202 |
+
"rewards/accuracies": 1.0,
|
| 203 |
+
"rewards/chosen": -2.0924859046936035,
|
| 204 |
+
"rewards/margins": 70.16132354736328,
|
| 205 |
+
"rewards/rejected": -72.2538070678711,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.3409090909090909,
|
| 210 |
+
"grad_norm": 0.13283878655315423,
|
| 211 |
+
"learning_rate": 4.177746070897592e-07,
|
| 212 |
+
"logits/chosen": -2.329523801803589,
|
| 213 |
+
"logits/rejected": 0.7127262949943542,
|
| 214 |
+
"logps/chosen": -563.5079956054688,
|
| 215 |
+
"logps/rejected": -7580.98974609375,
|
| 216 |
+
"loss": 0.007,
|
| 217 |
+
"rewards/accuracies": 1.0,
|
| 218 |
+
"rewards/chosen": -1.8199115991592407,
|
| 219 |
+
"rewards/margins": 67.61772155761719,
|
| 220 |
+
"rewards/rejected": -69.43761444091797,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.3693181818181818,
|
| 225 |
+
"grad_norm": 0.24292729812234612,
|
| 226 |
+
"learning_rate": 3.9855025724292763e-07,
|
| 227 |
+
"logits/chosen": -2.526569366455078,
|
| 228 |
+
"logits/rejected": 1.3170499801635742,
|
| 229 |
+
"logps/chosen": -577.057861328125,
|
| 230 |
+
"logps/rejected": -8586.48046875,
|
| 231 |
+
"loss": 0.0012,
|
| 232 |
+
"rewards/accuracies": 1.0,
|
| 233 |
+
"rewards/chosen": -2.0660626888275146,
|
| 234 |
+
"rewards/margins": 77.39335632324219,
|
| 235 |
+
"rewards/rejected": -79.45941925048828,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.3977272727272727,
|
| 240 |
+
"grad_norm": 0.9183173633670716,
|
| 241 |
+
"learning_rate": 3.7785886977585555e-07,
|
| 242 |
+
"logits/chosen": -2.464595317840576,
|
| 243 |
+
"logits/rejected": 1.4278135299682617,
|
| 244 |
+
"logps/chosen": -540.9710083007812,
|
| 245 |
+
"logps/rejected": -10038.2763671875,
|
| 246 |
+
"loss": 0.0016,
|
| 247 |
+
"rewards/accuracies": 1.0,
|
| 248 |
+
"rewards/chosen": -1.749346375465393,
|
| 249 |
+
"rewards/margins": 92.21351623535156,
|
| 250 |
+
"rewards/rejected": -93.96287536621094,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.42613636363636365,
|
| 255 |
+
"grad_norm": 0.6993679548084264,
|
| 256 |
+
"learning_rate": 3.5590478660213206e-07,
|
| 257 |
+
"logits/chosen": -2.191847562789917,
|
| 258 |
+
"logits/rejected": 1.6170787811279297,
|
| 259 |
+
"logps/chosen": -609.4722900390625,
|
| 260 |
+
"logps/rejected": -9242.5185546875,
|
| 261 |
+
"loss": 0.0016,
|
| 262 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 263 |
+
"rewards/chosen": -2.3140909671783447,
|
| 264 |
+
"rewards/margins": 82.9913101196289,
|
| 265 |
+
"rewards/rejected": -85.30540466308594,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.42613636363636365,
|
| 270 |
+
"eval_logits/chosen": -2.19913911819458,
|
| 271 |
+
"eval_logits/rejected": 1.8917714357376099,
|
| 272 |
+
"eval_logps/chosen": -571.1802368164062,
|
| 273 |
+
"eval_logps/rejected": -8522.7255859375,
|
| 274 |
+
"eval_loss": 0.0007830065442249179,
|
| 275 |
+
"eval_rewards/accuracies": 1.0,
|
| 276 |
+
"eval_rewards/chosen": -2.041984796524048,
|
| 277 |
+
"eval_rewards/margins": 76.77738952636719,
|
| 278 |
+
"eval_rewards/rejected": -78.81936645507812,
|
| 279 |
+
"eval_runtime": 195.9934,
|
| 280 |
+
"eval_samples_per_second": 19.924,
|
| 281 |
+
"eval_steps_per_second": 0.316,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.45454545454545453,
|
| 286 |
+
"grad_norm": 1.8645892720332131,
|
| 287 |
+
"learning_rate": 3.3290481963801696e-07,
|
| 288 |
+
"logits/chosen": -2.4417898654937744,
|
| 289 |
+
"logits/rejected": 1.9294321537017822,
|
| 290 |
+
"logps/chosen": -544.4041748046875,
|
| 291 |
+
"logps/rejected": -8048.359375,
|
| 292 |
+
"loss": 0.0004,
|
| 293 |
+
"rewards/accuracies": 1.0,
|
| 294 |
+
"rewards/chosen": -1.7426624298095703,
|
| 295 |
+
"rewards/margins": 72.57603454589844,
|
| 296 |
+
"rewards/rejected": -74.31868743896484,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.48295454545454547,
|
| 301 |
+
"grad_norm": 0.026359625456043548,
|
| 302 |
+
"learning_rate": 3.0908610963322626e-07,
|
| 303 |
+
"logits/chosen": -2.6247737407684326,
|
| 304 |
+
"logits/rejected": 1.3285863399505615,
|
| 305 |
+
"logps/chosen": -582.7529907226562,
|
| 306 |
+
"logps/rejected": -8232.6865234375,
|
| 307 |
+
"loss": 0.0027,
|
| 308 |
+
"rewards/accuracies": 1.0,
|
| 309 |
+
"rewards/chosen": -1.9513471126556396,
|
| 310 |
+
"rewards/margins": 73.0924072265625,
|
| 311 |
+
"rewards/rejected": -75.04375457763672,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.5113636363636364,
|
| 316 |
+
"grad_norm": 0.023938485830486744,
|
| 317 |
+
"learning_rate": 2.846838829972671e-07,
|
| 318 |
+
"logits/chosen": -2.2369513511657715,
|
| 319 |
+
"logits/rejected": 2.0209977626800537,
|
| 320 |
+
"logps/chosen": -574.5399780273438,
|
| 321 |
+
"logps/rejected": -7650.44775390625,
|
| 322 |
+
"loss": 0.0056,
|
| 323 |
+
"rewards/accuracies": 1.0,
|
| 324 |
+
"rewards/chosen": -2.136356830596924,
|
| 325 |
+
"rewards/margins": 68.23828887939453,
|
| 326 |
+
"rewards/rejected": -70.37464904785156,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.5397727272727273,
|
| 331 |
+
"grad_norm": 0.22173170076093188,
|
| 332 |
+
"learning_rate": 2.5993912877423147e-07,
|
| 333 |
+
"logits/chosen": -1.6255722045898438,
|
| 334 |
+
"logits/rejected": 1.8459784984588623,
|
| 335 |
+
"logps/chosen": -532.0623168945312,
|
| 336 |
+
"logps/rejected": -8512.8662109375,
|
| 337 |
+
"loss": 0.0009,
|
| 338 |
+
"rewards/accuracies": 1.0,
|
| 339 |
+
"rewards/chosen": -1.864465355873108,
|
| 340 |
+
"rewards/margins": 76.34809875488281,
|
| 341 |
+
"rewards/rejected": -78.21255493164062,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.5681818181818182,
|
| 346 |
+
"grad_norm": 0.7897968463196222,
|
| 347 |
+
"learning_rate": 2.3509621870754504e-07,
|
| 348 |
+
"logits/chosen": -1.029783010482788,
|
| 349 |
+
"logits/rejected": 2.887779474258423,
|
| 350 |
+
"logps/chosen": -551.5975952148438,
|
| 351 |
+
"logps/rejected": -8021.84912109375,
|
| 352 |
+
"loss": 0.0015,
|
| 353 |
+
"rewards/accuracies": 1.0,
|
| 354 |
+
"rewards/chosen": -1.8646786212921143,
|
| 355 |
+
"rewards/margins": 72.0453872680664,
|
| 356 |
+
"rewards/rejected": -73.91007232666016,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.5681818181818182,
|
| 361 |
+
"eval_logits/chosen": -0.8412286043167114,
|
| 362 |
+
"eval_logits/rejected": 2.5732452869415283,
|
| 363 |
+
"eval_logps/chosen": -564.5499877929688,
|
| 364 |
+
"eval_logps/rejected": -8622.9443359375,
|
| 365 |
+
"eval_loss": 0.0007432692218571901,
|
| 366 |
+
"eval_rewards/accuracies": 1.0,
|
| 367 |
+
"eval_rewards/chosen": -1.9756826162338257,
|
| 368 |
+
"eval_rewards/margins": 77.84588623046875,
|
| 369 |
+
"eval_rewards/rejected": -79.8215560913086,
|
| 370 |
+
"eval_runtime": 196.9649,
|
| 371 |
+
"eval_samples_per_second": 19.826,
|
| 372 |
+
"eval_steps_per_second": 0.315,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.5965909090909091,
|
| 377 |
+
"grad_norm": 0.4603520409149027,
|
| 378 |
+
"learning_rate": 2.1040049389819624e-07,
|
| 379 |
+
"logits/chosen": -0.8681282997131348,
|
| 380 |
+
"logits/rejected": 2.764690399169922,
|
| 381 |
+
"logps/chosen": -578.1253662109375,
|
| 382 |
+
"logps/rejected": -8362.7060546875,
|
| 383 |
+
"loss": 0.0032,
|
| 384 |
+
"rewards/accuracies": 1.0,
|
| 385 |
+
"rewards/chosen": -1.8748207092285156,
|
| 386 |
+
"rewards/margins": 75.43318939208984,
|
| 387 |
+
"rewards/rejected": -77.30801391601562,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.625,
|
| 392 |
+
"grad_norm": 2.35697501687302,
|
| 393 |
+
"learning_rate": 1.8609584188988133e-07,
|
| 394 |
+
"logits/chosen": -1.0785493850708008,
|
| 395 |
+
"logits/rejected": 2.1844732761383057,
|
| 396 |
+
"logps/chosen": -606.7586059570312,
|
| 397 |
+
"logps/rejected": -7175.0869140625,
|
| 398 |
+
"loss": 0.0036,
|
| 399 |
+
"rewards/accuracies": 1.0,
|
| 400 |
+
"rewards/chosen": -2.4215800762176514,
|
| 401 |
+
"rewards/margins": 62.83292770385742,
|
| 402 |
+
"rewards/rejected": -65.25450897216797,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.6534090909090909,
|
| 407 |
+
"grad_norm": 22.84103924366115,
|
| 408 |
+
"learning_rate": 1.624222881090439e-07,
|
| 409 |
+
"logits/chosen": -1.365192174911499,
|
| 410 |
+
"logits/rejected": 1.863221526145935,
|
| 411 |
+
"logps/chosen": -607.694091796875,
|
| 412 |
+
"logps/rejected": -8029.53759765625,
|
| 413 |
+
"loss": 0.0058,
|
| 414 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 415 |
+
"rewards/chosen": -2.130295515060425,
|
| 416 |
+
"rewards/margins": 71.30926513671875,
|
| 417 |
+
"rewards/rejected": -73.43955993652344,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.6818181818181818,
|
| 422 |
+
"grad_norm": 0.4015693433898369,
|
| 423 |
+
"learning_rate": 1.3961362544602212e-07,
|
| 424 |
+
"logits/chosen": -1.195718765258789,
|
| 425 |
+
"logits/rejected": 2.2943453788757324,
|
| 426 |
+
"logps/chosen": -563.174560546875,
|
| 427 |
+
"logps/rejected": -6934.9736328125,
|
| 428 |
+
"loss": 0.0015,
|
| 429 |
+
"rewards/accuracies": 1.0,
|
| 430 |
+
"rewards/chosen": -1.8716214895248413,
|
| 431 |
+
"rewards/margins": 61.537391662597656,
|
| 432 |
+
"rewards/rejected": -63.409019470214844,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.7102272727272727,
|
| 437 |
+
"grad_norm": 0.304169201785893,
|
| 438 |
+
"learning_rate": 1.1789510538684522e-07,
|
| 439 |
+
"logits/chosen": -1.2848708629608154,
|
| 440 |
+
"logits/rejected": 2.0380046367645264,
|
| 441 |
+
"logps/chosen": -532.541015625,
|
| 442 |
+
"logps/rejected": -8912.1201171875,
|
| 443 |
+
"loss": 0.0016,
|
| 444 |
+
"rewards/accuracies": 1.0,
|
| 445 |
+
"rewards/chosen": -1.8050743341445923,
|
| 446 |
+
"rewards/margins": 80.02030944824219,
|
| 447 |
+
"rewards/rejected": -81.82538604736328,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.7102272727272727,
|
| 452 |
+
"eval_logits/chosen": -1.2004659175872803,
|
| 453 |
+
"eval_logits/rejected": 2.2580416202545166,
|
| 454 |
+
"eval_logps/chosen": -549.9619750976562,
|
| 455 |
+
"eval_logps/rejected": -7654.19775390625,
|
| 456 |
+
"eval_loss": 0.0007973507163114846,
|
| 457 |
+
"eval_rewards/accuracies": 1.0,
|
| 458 |
+
"eval_rewards/chosen": -1.829802393913269,
|
| 459 |
+
"eval_rewards/margins": 68.30428314208984,
|
| 460 |
+
"eval_rewards/rejected": -70.13408660888672,
|
| 461 |
+
"eval_runtime": 195.5911,
|
| 462 |
+
"eval_samples_per_second": 19.965,
|
| 463 |
+
"eval_steps_per_second": 0.317,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.7386363636363636,
|
| 468 |
+
"grad_norm": 0.5568394356105595,
|
| 469 |
+
"learning_rate": 9.748121349736891e-08,
|
| 470 |
+
"logits/chosen": -1.168460488319397,
|
| 471 |
+
"logits/rejected": 2.0924360752105713,
|
| 472 |
+
"logps/chosen": -588.483642578125,
|
| 473 |
+
"logps/rejected": -7967.40478515625,
|
| 474 |
+
"loss": 0.0008,
|
| 475 |
+
"rewards/accuracies": 1.0,
|
| 476 |
+
"rewards/chosen": -2.046555995941162,
|
| 477 |
+
"rewards/margins": 70.84620666503906,
|
| 478 |
+
"rewards/rejected": -72.89276885986328,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.7670454545454546,
|
| 483 |
+
"grad_norm": 0.07097046676874795,
|
| 484 |
+
"learning_rate": 7.857355122839673e-08,
|
| 485 |
+
"logits/chosen": -1.3962290287017822,
|
| 486 |
+
"logits/rejected": 1.8883154392242432,
|
| 487 |
+
"logps/chosen": -566.6204833984375,
|
| 488 |
+
"logps/rejected": -7627.25634765625,
|
| 489 |
+
"loss": 0.0008,
|
| 490 |
+
"rewards/accuracies": 1.0,
|
| 491 |
+
"rewards/chosen": -1.8577096462249756,
|
| 492 |
+
"rewards/margins": 67.72525024414062,
|
| 493 |
+
"rewards/rejected": -69.58296203613281,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.7954545454545454,
|
| 498 |
+
"grad_norm": 0.05520595932113315,
|
| 499 |
+
"learning_rate": 6.135884496044244e-08,
|
| 500 |
+
"logits/chosen": -1.3116520643234253,
|
| 501 |
+
"logits/rejected": 2.1938819885253906,
|
| 502 |
+
"logps/chosen": -557.5162353515625,
|
| 503 |
+
"logps/rejected": -7626.7578125,
|
| 504 |
+
"loss": 0.0013,
|
| 505 |
+
"rewards/accuracies": 1.0,
|
| 506 |
+
"rewards/chosen": -1.7834333181381226,
|
| 507 |
+
"rewards/margins": 68.03524017333984,
|
| 508 |
+
"rewards/rejected": -69.81867980957031,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.8238636363636364,
|
| 513 |
+
"grad_norm": 0.05646545972920247,
|
| 514 |
+
"learning_rate": 4.600710195020982e-08,
|
| 515 |
+
"logits/chosen": -1.1199910640716553,
|
| 516 |
+
"logits/rejected": 2.300516128540039,
|
| 517 |
+
"logps/chosen": -577.44482421875,
|
| 518 |
+
"logps/rejected": -7700.6279296875,
|
| 519 |
+
"loss": 0.0024,
|
| 520 |
+
"rewards/accuracies": 1.0,
|
| 521 |
+
"rewards/chosen": -1.8830159902572632,
|
| 522 |
+
"rewards/margins": 68.71068572998047,
|
| 523 |
+
"rewards/rejected": -70.59370422363281,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.8522727272727273,
|
| 528 |
+
"grad_norm": 0.5065392892121787,
|
| 529 |
+
"learning_rate": 3.2669931390104374e-08,
|
| 530 |
+
"logits/chosen": -1.1116163730621338,
|
| 531 |
+
"logits/rejected": 2.1315221786499023,
|
| 532 |
+
"logps/chosen": -531.3294677734375,
|
| 533 |
+
"logps/rejected": -8432.990234375,
|
| 534 |
+
"loss": 0.0008,
|
| 535 |
+
"rewards/accuracies": 1.0,
|
| 536 |
+
"rewards/chosen": -1.7507492303848267,
|
| 537 |
+
"rewards/margins": 76.37090301513672,
|
| 538 |
+
"rewards/rejected": -78.12163543701172,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.8522727272727273,
|
| 543 |
+
"eval_logits/chosen": -0.9558252096176147,
|
| 544 |
+
"eval_logits/rejected": 2.2921457290649414,
|
| 545 |
+
"eval_logps/chosen": -557.7684936523438,
|
| 546 |
+
"eval_logps/rejected": -8073.39990234375,
|
| 547 |
+
"eval_loss": 0.0005955722881481051,
|
| 548 |
+
"eval_rewards/accuracies": 1.0,
|
| 549 |
+
"eval_rewards/chosen": -1.907867670059204,
|
| 550 |
+
"eval_rewards/margins": 72.41824340820312,
|
| 551 |
+
"eval_rewards/rejected": -74.32611846923828,
|
| 552 |
+
"eval_runtime": 196.4288,
|
| 553 |
+
"eval_samples_per_second": 19.88,
|
| 554 |
+
"eval_steps_per_second": 0.316,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.8806818181818182,
|
| 559 |
+
"grad_norm": 0.11084443137467888,
|
| 560 |
+
"learning_rate": 2.147904716149135e-08,
|
| 561 |
+
"logits/chosen": -0.904153048992157,
|
| 562 |
+
"logits/rejected": 2.1967928409576416,
|
| 563 |
+
"logps/chosen": -541.1990966796875,
|
| 564 |
+
"logps/rejected": -8187.125,
|
| 565 |
+
"loss": 0.002,
|
| 566 |
+
"rewards/accuracies": 1.0,
|
| 567 |
+
"rewards/chosen": -1.7913249731063843,
|
| 568 |
+
"rewards/margins": 73.07491302490234,
|
| 569 |
+
"rewards/rejected": -74.86624145507812,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.9090909090909091,
|
| 574 |
+
"grad_norm": 0.06043015851279257,
|
| 575 |
+
"learning_rate": 1.254496706805433e-08,
|
| 576 |
+
"logits/chosen": -1.1182914972305298,
|
| 577 |
+
"logits/rejected": 2.186739206314087,
|
| 578 |
+
"logps/chosen": -570.1544189453125,
|
| 579 |
+
"logps/rejected": -8264.416015625,
|
| 580 |
+
"loss": 0.0005,
|
| 581 |
+
"rewards/accuracies": 1.0,
|
| 582 |
+
"rewards/chosen": -1.8748136758804321,
|
| 583 |
+
"rewards/margins": 74.24952697753906,
|
| 584 |
+
"rewards/rejected": -76.12433624267578,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.9375,
|
| 589 |
+
"grad_norm": 0.14908479424865714,
|
| 590 |
+
"learning_rate": 5.955921395237318e-09,
|
| 591 |
+
"logits/chosen": -0.9981291890144348,
|
| 592 |
+
"logits/rejected": 2.14846134185791,
|
| 593 |
+
"logps/chosen": -528.0042114257812,
|
| 594 |
+
"logps/rejected": -8246.408203125,
|
| 595 |
+
"loss": 0.0004,
|
| 596 |
+
"rewards/accuracies": 1.0,
|
| 597 |
+
"rewards/chosen": -1.727360486984253,
|
| 598 |
+
"rewards/margins": 74.25918579101562,
|
| 599 |
+
"rewards/rejected": -75.98654174804688,
|
| 600 |
+
"step": 330
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.9659090909090909,
|
| 604 |
+
"grad_norm": 0.18250364966205163,
|
| 605 |
+
"learning_rate": 1.7769815745066474e-09,
|
| 606 |
+
"logits/chosen": -1.248327612876892,
|
| 607 |
+
"logits/rejected": 2.005960702896118,
|
| 608 |
+
"logps/chosen": -543.0258178710938,
|
| 609 |
+
"logps/rejected": -7556.515625,
|
| 610 |
+
"loss": 0.0008,
|
| 611 |
+
"rewards/accuracies": 1.0,
|
| 612 |
+
"rewards/chosen": -1.7404506206512451,
|
| 613 |
+
"rewards/margins": 67.73704528808594,
|
| 614 |
+
"rewards/rejected": -69.47749328613281,
|
| 615 |
+
"step": 340
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.9943181818181818,
|
| 619 |
+
"grad_norm": 1.0466321976770805,
|
| 620 |
+
"learning_rate": 4.9417557483610875e-11,
|
| 621 |
+
"logits/chosen": -0.9777113199234009,
|
| 622 |
+
"logits/rejected": 2.094377040863037,
|
| 623 |
+
"logps/chosen": -576.5343017578125,
|
| 624 |
+
"logps/rejected": -8160.74853515625,
|
| 625 |
+
"loss": 0.0029,
|
| 626 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 627 |
+
"rewards/chosen": -2.0226657390594482,
|
| 628 |
+
"rewards/margins": 73.10923767089844,
|
| 629 |
+
"rewards/rejected": -75.13190460205078,
|
| 630 |
+
"step": 350
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.9943181818181818,
|
| 634 |
+
"eval_logits/chosen": -0.908790111541748,
|
| 635 |
+
"eval_logits/rejected": 2.3223326206207275,
|
| 636 |
+
"eval_logps/chosen": -561.191650390625,
|
| 637 |
+
"eval_logps/rejected": -8146.36962890625,
|
| 638 |
+
"eval_loss": 0.0005959240952506661,
|
| 639 |
+
"eval_rewards/accuracies": 1.0,
|
| 640 |
+
"eval_rewards/chosen": -1.9420990943908691,
|
| 641 |
+
"eval_rewards/margins": 73.11370849609375,
|
| 642 |
+
"eval_rewards/rejected": -75.0558090209961,
|
| 643 |
+
"eval_runtime": 196.8349,
|
| 644 |
+
"eval_samples_per_second": 19.839,
|
| 645 |
+
"eval_steps_per_second": 0.315,
|
| 646 |
+
"step": 350
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 1.0,
|
| 650 |
+
"step": 352,
|
| 651 |
+
"total_flos": 0.0,
|
| 652 |
+
"train_loss": 0.0465552191166022,
|
| 653 |
+
"train_runtime": 10297.8682,
|
| 654 |
+
"train_samples_per_second": 4.37,
|
| 655 |
+
"train_steps_per_second": 0.034
|
| 656 |
+
}
|
| 657 |
+
],
|
| 658 |
+
"logging_steps": 10,
|
| 659 |
+
"max_steps": 352,
|
| 660 |
+
"num_input_tokens_seen": 0,
|
| 661 |
+
"num_train_epochs": 1,
|
| 662 |
+
"save_steps": 100,
|
| 663 |
+
"stateful_callbacks": {
|
| 664 |
+
"TrainerControl": {
|
| 665 |
+
"args": {
|
| 666 |
+
"should_epoch_stop": false,
|
| 667 |
+
"should_evaluate": false,
|
| 668 |
+
"should_log": false,
|
| 669 |
+
"should_save": true,
|
| 670 |
+
"should_training_stop": true
|
| 671 |
+
},
|
| 672 |
+
"attributes": {}
|
| 673 |
+
}
|
| 674 |
+
},
|
| 675 |
+
"total_flos": 0.0,
|
| 676 |
+
"train_batch_size": 8,
|
| 677 |
+
"trial_name": null,
|
| 678 |
+
"trial_params": null
|
| 679 |
+
}
|