Model save
Browse files- README.md +75 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +509 -0
README.md
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-align-scan-0.0-0.0-polynomial-3
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-7b-align-scan-0.0-0.0-polynomial-3
|
| 18 |
+
|
| 19 |
+
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.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6749
|
| 22 |
+
- Rewards/chosen: -1.2868
|
| 23 |
+
- Rewards/rejected: -1.7792
|
| 24 |
+
- Rewards/accuracies: 0.3313
|
| 25 |
+
- Rewards/margins: 0.4925
|
| 26 |
+
- Logps/rejected: -233.7709
|
| 27 |
+
- Logps/chosen: -184.8848
|
| 28 |
+
- Logits/rejected: -1.4185
|
| 29 |
+
- Logits/chosen: -1.4498
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 9.787906592614782e-07
|
| 49 |
+
- train_batch_size: 8
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 2
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 32
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: polynomial
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 3
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.5825 | 1.0417 | 100 | 0.6482 | -0.2552 | -0.4894 | 0.3234 | 0.2342 | -123.1158 | -96.3836 | -2.0644 | -2.0903 |
|
| 67 |
+
| 0.3884 | 2.0833 | 200 | 0.6749 | -1.2868 | -1.7792 | 0.3313 | 0.4925 | -233.7709 | -184.8848 | -1.4185 | -1.4498 |
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
### Framework versions
|
| 71 |
+
|
| 72 |
+
- Transformers 4.44.2
|
| 73 |
+
- Pytorch 2.4.0
|
| 74 |
+
- Datasets 2.21.0
|
| 75 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5020060315728188,
|
| 5 |
+
"train_runtime": 3747.4353,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 4.894,
|
| 8 |
+
"train_steps_per_second": 0.077
|
| 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": "4.44.2"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5020060315728188,
|
| 5 |
+
"train_runtime": 3747.4353,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 4.894,
|
| 8 |
+
"train_steps_per_second": 0.077
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,509 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 288,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.010416666666666666,
|
| 13 |
+
"grad_norm": 7.676418798413947,
|
| 14 |
+
"learning_rate": 3.375140204349925e-08,
|
| 15 |
+
"logits/chosen": -2.590585231781006,
|
| 16 |
+
"logits/rejected": -2.5664222240448,
|
| 17 |
+
"logps/chosen": -80.29847717285156,
|
| 18 |
+
"logps/rejected": -53.10200881958008,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.10416666666666667,
|
| 28 |
+
"grad_norm": 6.867859377504262,
|
| 29 |
+
"learning_rate": 3.375140204349925e-07,
|
| 30 |
+
"logits/chosen": -2.5563881397247314,
|
| 31 |
+
"logits/rejected": -2.5385148525238037,
|
| 32 |
+
"logps/chosen": -87.91146850585938,
|
| 33 |
+
"logps/rejected": -81.01969909667969,
|
| 34 |
+
"loss": 0.6932,
|
| 35 |
+
"rewards/accuracies": 0.2291666716337204,
|
| 36 |
+
"rewards/chosen": 4.3005715269828215e-05,
|
| 37 |
+
"rewards/margins": -0.0001762014871928841,
|
| 38 |
+
"rewards/rejected": 0.0002192071988247335,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.20833333333333334,
|
| 43 |
+
"grad_norm": 6.140252183490494,
|
| 44 |
+
"learning_rate": 6.75028040869985e-07,
|
| 45 |
+
"logits/chosen": -2.6080267429351807,
|
| 46 |
+
"logits/rejected": -2.559004783630371,
|
| 47 |
+
"logps/chosen": -102.70601654052734,
|
| 48 |
+
"logps/rejected": -89.34085083007812,
|
| 49 |
+
"loss": 0.6915,
|
| 50 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 51 |
+
"rewards/chosen": 0.007517552934587002,
|
| 52 |
+
"rewards/margins": 0.0011824130779132247,
|
| 53 |
+
"rewards/rejected": 0.006335140205919743,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.3125,
|
| 58 |
+
"grad_norm": 8.51916101581021,
|
| 59 |
+
"learning_rate": 9.753976451330556e-07,
|
| 60 |
+
"logits/chosen": -2.479109287261963,
|
| 61 |
+
"logits/rejected": -2.495014190673828,
|
| 62 |
+
"logps/chosen": -65.36149597167969,
|
| 63 |
+
"logps/rejected": -74.12493896484375,
|
| 64 |
+
"loss": 0.6863,
|
| 65 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 66 |
+
"rewards/chosen": 0.024195391684770584,
|
| 67 |
+
"rewards/margins": 0.014859529212117195,
|
| 68 |
+
"rewards/rejected": 0.009335865266621113,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.4166666666666667,
|
| 73 |
+
"grad_norm": 7.712483677212969,
|
| 74 |
+
"learning_rate": 9.414675038488285e-07,
|
| 75 |
+
"logits/chosen": -2.49943470954895,
|
| 76 |
+
"logits/rejected": -2.4853296279907227,
|
| 77 |
+
"logps/chosen": -69.53825378417969,
|
| 78 |
+
"logps/rejected": -71.04619598388672,
|
| 79 |
+
"loss": 0.6772,
|
| 80 |
+
"rewards/accuracies": 0.2874999940395355,
|
| 81 |
+
"rewards/chosen": 0.030402392148971558,
|
| 82 |
+
"rewards/margins": 0.03428281098604202,
|
| 83 |
+
"rewards/rejected": -0.0038804211653769016,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.5208333333333334,
|
| 88 |
+
"grad_norm": 8.998261203801041,
|
| 89 |
+
"learning_rate": 9.075373625646016e-07,
|
| 90 |
+
"logits/chosen": -2.406968593597412,
|
| 91 |
+
"logits/rejected": -2.4230380058288574,
|
| 92 |
+
"logps/chosen": -54.66428756713867,
|
| 93 |
+
"logps/rejected": -64.8364486694336,
|
| 94 |
+
"loss": 0.6712,
|
| 95 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 96 |
+
"rewards/chosen": -0.04386414587497711,
|
| 97 |
+
"rewards/margins": 0.02522563934326172,
|
| 98 |
+
"rewards/rejected": -0.06908978521823883,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.625,
|
| 103 |
+
"grad_norm": 8.947771345110965,
|
| 104 |
+
"learning_rate": 8.736072212803746e-07,
|
| 105 |
+
"logits/chosen": -2.486833095550537,
|
| 106 |
+
"logits/rejected": -2.46416974067688,
|
| 107 |
+
"logps/chosen": -103.86640930175781,
|
| 108 |
+
"logps/rejected": -110.64974212646484,
|
| 109 |
+
"loss": 0.6728,
|
| 110 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 111 |
+
"rewards/chosen": -0.2890399098396301,
|
| 112 |
+
"rewards/margins": 0.07593990862369537,
|
| 113 |
+
"rewards/rejected": -0.3649798035621643,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.7291666666666666,
|
| 118 |
+
"grad_norm": 9.915857724019418,
|
| 119 |
+
"learning_rate": 8.396770799961476e-07,
|
| 120 |
+
"logits/chosen": -2.4623074531555176,
|
| 121 |
+
"logits/rejected": -2.4422433376312256,
|
| 122 |
+
"logps/chosen": -121.77342224121094,
|
| 123 |
+
"logps/rejected": -119.22220611572266,
|
| 124 |
+
"loss": 0.6581,
|
| 125 |
+
"rewards/accuracies": 0.33125001192092896,
|
| 126 |
+
"rewards/chosen": -0.2963990569114685,
|
| 127 |
+
"rewards/margins": 0.09725265949964523,
|
| 128 |
+
"rewards/rejected": -0.39365169405937195,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.8333333333333334,
|
| 133 |
+
"grad_norm": 25.03318196876874,
|
| 134 |
+
"learning_rate": 8.057469387119207e-07,
|
| 135 |
+
"logits/chosen": -2.3540024757385254,
|
| 136 |
+
"logits/rejected": -2.2817678451538086,
|
| 137 |
+
"logps/chosen": -105.01051330566406,
|
| 138 |
+
"logps/rejected": -110.41041564941406,
|
| 139 |
+
"loss": 0.6459,
|
| 140 |
+
"rewards/accuracies": 0.39375001192092896,
|
| 141 |
+
"rewards/chosen": -0.22572524845600128,
|
| 142 |
+
"rewards/margins": 0.13594061136245728,
|
| 143 |
+
"rewards/rejected": -0.36166587471961975,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.9375,
|
| 148 |
+
"grad_norm": 16.29058709199533,
|
| 149 |
+
"learning_rate": 7.718167974276937e-07,
|
| 150 |
+
"logits/chosen": -2.132964611053467,
|
| 151 |
+
"logits/rejected": -2.117180824279785,
|
| 152 |
+
"logps/chosen": -66.82723236083984,
|
| 153 |
+
"logps/rejected": -93.7275161743164,
|
| 154 |
+
"loss": 0.6446,
|
| 155 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 156 |
+
"rewards/chosen": -0.14975103735923767,
|
| 157 |
+
"rewards/margins": 0.202955961227417,
|
| 158 |
+
"rewards/rejected": -0.3527069687843323,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.0416666666666667,
|
| 163 |
+
"grad_norm": 16.284386286452357,
|
| 164 |
+
"learning_rate": 7.378866561434668e-07,
|
| 165 |
+
"logits/chosen": -2.041792392730713,
|
| 166 |
+
"logits/rejected": -2.0138609409332275,
|
| 167 |
+
"logps/chosen": -89.52456665039062,
|
| 168 |
+
"logps/rejected": -122.956298828125,
|
| 169 |
+
"loss": 0.5825,
|
| 170 |
+
"rewards/accuracies": 0.40625,
|
| 171 |
+
"rewards/chosen": -0.18652945756912231,
|
| 172 |
+
"rewards/margins": 0.4071316123008728,
|
| 173 |
+
"rewards/rejected": -0.5936610698699951,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.0416666666666667,
|
| 178 |
+
"eval_logits/chosen": -2.0902979373931885,
|
| 179 |
+
"eval_logits/rejected": -2.0643553733825684,
|
| 180 |
+
"eval_logps/chosen": -96.38362121582031,
|
| 181 |
+
"eval_logps/rejected": -123.11580657958984,
|
| 182 |
+
"eval_loss": 0.6482308506965637,
|
| 183 |
+
"eval_rewards/accuracies": 0.32341268658638,
|
| 184 |
+
"eval_rewards/chosen": -0.25518056750297546,
|
| 185 |
+
"eval_rewards/margins": 0.2342308610677719,
|
| 186 |
+
"eval_rewards/rejected": -0.48941144347190857,
|
| 187 |
+
"eval_runtime": 113.6473,
|
| 188 |
+
"eval_samples_per_second": 17.598,
|
| 189 |
+
"eval_steps_per_second": 0.554,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.1458333333333333,
|
| 194 |
+
"grad_norm": 14.36714525578378,
|
| 195 |
+
"learning_rate": 7.039565148592398e-07,
|
| 196 |
+
"logits/chosen": -2.079672336578369,
|
| 197 |
+
"logits/rejected": -2.1257057189941406,
|
| 198 |
+
"logps/chosen": -75.16223907470703,
|
| 199 |
+
"logps/rejected": -157.490234375,
|
| 200 |
+
"loss": 0.4839,
|
| 201 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 202 |
+
"rewards/chosen": -0.11415556818246841,
|
| 203 |
+
"rewards/margins": 0.7717818021774292,
|
| 204 |
+
"rewards/rejected": -0.8859373331069946,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.25,
|
| 209 |
+
"grad_norm": 10.05291833999338,
|
| 210 |
+
"learning_rate": 6.700263735750129e-07,
|
| 211 |
+
"logits/chosen": -2.1770777702331543,
|
| 212 |
+
"logits/rejected": -2.12736439704895,
|
| 213 |
+
"logps/chosen": -124.73258972167969,
|
| 214 |
+
"logps/rejected": -207.2230224609375,
|
| 215 |
+
"loss": 0.4901,
|
| 216 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 217 |
+
"rewards/chosen": -0.1949707567691803,
|
| 218 |
+
"rewards/margins": 1.1325080394744873,
|
| 219 |
+
"rewards/rejected": -1.3274791240692139,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.3541666666666667,
|
| 224 |
+
"grad_norm": 10.238396987504712,
|
| 225 |
+
"learning_rate": 6.360962322907859e-07,
|
| 226 |
+
"logits/chosen": -1.985735297203064,
|
| 227 |
+
"logits/rejected": -1.9928312301635742,
|
| 228 |
+
"logps/chosen": -108.13710021972656,
|
| 229 |
+
"logps/rejected": -195.22592163085938,
|
| 230 |
+
"loss": 0.4606,
|
| 231 |
+
"rewards/accuracies": 0.4437499940395355,
|
| 232 |
+
"rewards/chosen": -0.2514357268810272,
|
| 233 |
+
"rewards/margins": 0.9828246831893921,
|
| 234 |
+
"rewards/rejected": -1.2342605590820312,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.4583333333333333,
|
| 239 |
+
"grad_norm": 13.992502570332631,
|
| 240 |
+
"learning_rate": 6.02166091006559e-07,
|
| 241 |
+
"logits/chosen": -1.9039230346679688,
|
| 242 |
+
"logits/rejected": -1.9281995296478271,
|
| 243 |
+
"logps/chosen": -81.67506408691406,
|
| 244 |
+
"logps/rejected": -182.8755340576172,
|
| 245 |
+
"loss": 0.4713,
|
| 246 |
+
"rewards/accuracies": 0.40625,
|
| 247 |
+
"rewards/chosen": -0.3775108754634857,
|
| 248 |
+
"rewards/margins": 0.9950442314147949,
|
| 249 |
+
"rewards/rejected": -1.372555136680603,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.5625,
|
| 254 |
+
"grad_norm": 16.317117170378367,
|
| 255 |
+
"learning_rate": 5.682359497223321e-07,
|
| 256 |
+
"logits/chosen": -1.9301204681396484,
|
| 257 |
+
"logits/rejected": -1.8779830932617188,
|
| 258 |
+
"logps/chosen": -101.25444030761719,
|
| 259 |
+
"logps/rejected": -199.9362030029297,
|
| 260 |
+
"loss": 0.4567,
|
| 261 |
+
"rewards/accuracies": 0.4437499940395355,
|
| 262 |
+
"rewards/chosen": -0.3392702639102936,
|
| 263 |
+
"rewards/margins": 1.133601427078247,
|
| 264 |
+
"rewards/rejected": -1.4728715419769287,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.6666666666666665,
|
| 269 |
+
"grad_norm": 13.942547798372193,
|
| 270 |
+
"learning_rate": 5.34305808438105e-07,
|
| 271 |
+
"logits/chosen": -1.8111422061920166,
|
| 272 |
+
"logits/rejected": -1.8152790069580078,
|
| 273 |
+
"logps/chosen": -125.3135757446289,
|
| 274 |
+
"logps/rejected": -260.8206787109375,
|
| 275 |
+
"loss": 0.4454,
|
| 276 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 277 |
+
"rewards/chosen": -0.5173971056938171,
|
| 278 |
+
"rewards/margins": 1.497753381729126,
|
| 279 |
+
"rewards/rejected": -2.015150547027588,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.7708333333333335,
|
| 284 |
+
"grad_norm": 20.552273461683416,
|
| 285 |
+
"learning_rate": 5.003756671538781e-07,
|
| 286 |
+
"logits/chosen": -1.6733791828155518,
|
| 287 |
+
"logits/rejected": -1.6446590423583984,
|
| 288 |
+
"logps/chosen": -139.14492797851562,
|
| 289 |
+
"logps/rejected": -261.77056884765625,
|
| 290 |
+
"loss": 0.4492,
|
| 291 |
+
"rewards/accuracies": 0.53125,
|
| 292 |
+
"rewards/chosen": -0.7451165318489075,
|
| 293 |
+
"rewards/margins": 1.3388173580169678,
|
| 294 |
+
"rewards/rejected": -2.0839340686798096,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.875,
|
| 299 |
+
"grad_norm": 16.343999863313055,
|
| 300 |
+
"learning_rate": 4.664455258696512e-07,
|
| 301 |
+
"logits/chosen": -1.697675108909607,
|
| 302 |
+
"logits/rejected": -1.65886652469635,
|
| 303 |
+
"logps/chosen": -117.02315521240234,
|
| 304 |
+
"logps/rejected": -225.981201171875,
|
| 305 |
+
"loss": 0.4335,
|
| 306 |
+
"rewards/accuracies": 0.46875,
|
| 307 |
+
"rewards/chosen": -0.52647864818573,
|
| 308 |
+
"rewards/margins": 1.2495534420013428,
|
| 309 |
+
"rewards/rejected": -1.7760320901870728,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.9791666666666665,
|
| 314 |
+
"grad_norm": 17.543702138725678,
|
| 315 |
+
"learning_rate": 4.325153845854242e-07,
|
| 316 |
+
"logits/chosen": -1.708845853805542,
|
| 317 |
+
"logits/rejected": -1.6851694583892822,
|
| 318 |
+
"logps/chosen": -132.83035278320312,
|
| 319 |
+
"logps/rejected": -266.851318359375,
|
| 320 |
+
"loss": 0.4324,
|
| 321 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 322 |
+
"rewards/chosen": -0.5171667337417603,
|
| 323 |
+
"rewards/margins": 1.5369913578033447,
|
| 324 |
+
"rewards/rejected": -2.0541584491729736,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.0833333333333335,
|
| 329 |
+
"grad_norm": 4.109061869011508,
|
| 330 |
+
"learning_rate": 3.985852433011972e-07,
|
| 331 |
+
"logits/chosen": -1.4759489297866821,
|
| 332 |
+
"logits/rejected": -1.4932861328125,
|
| 333 |
+
"logps/chosen": -147.7212677001953,
|
| 334 |
+
"logps/rejected": -366.11163330078125,
|
| 335 |
+
"loss": 0.3884,
|
| 336 |
+
"rewards/accuracies": 0.543749988079071,
|
| 337 |
+
"rewards/chosen": -0.6844128370285034,
|
| 338 |
+
"rewards/margins": 2.3995113372802734,
|
| 339 |
+
"rewards/rejected": -3.0839240550994873,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 2.0833333333333335,
|
| 344 |
+
"eval_logits/chosen": -1.449843168258667,
|
| 345 |
+
"eval_logits/rejected": -1.4185280799865723,
|
| 346 |
+
"eval_logps/chosen": -184.884765625,
|
| 347 |
+
"eval_logps/rejected": -233.7708740234375,
|
| 348 |
+
"eval_loss": 0.6749430298805237,
|
| 349 |
+
"eval_rewards/accuracies": 0.3313491940498352,
|
| 350 |
+
"eval_rewards/chosen": -1.2867618799209595,
|
| 351 |
+
"eval_rewards/margins": 0.49245980381965637,
|
| 352 |
+
"eval_rewards/rejected": -1.7792216539382935,
|
| 353 |
+
"eval_runtime": 113.4025,
|
| 354 |
+
"eval_samples_per_second": 17.636,
|
| 355 |
+
"eval_steps_per_second": 0.556,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.1875,
|
| 360 |
+
"grad_norm": 3.8226987442037843,
|
| 361 |
+
"learning_rate": 3.646551020169702e-07,
|
| 362 |
+
"logits/chosen": -1.3593485355377197,
|
| 363 |
+
"logits/rejected": -1.32414972782135,
|
| 364 |
+
"logps/chosen": -109.7019271850586,
|
| 365 |
+
"logps/rejected": -280.071044921875,
|
| 366 |
+
"loss": 0.4007,
|
| 367 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 368 |
+
"rewards/chosen": -0.5374400019645691,
|
| 369 |
+
"rewards/margins": 1.9264366626739502,
|
| 370 |
+
"rewards/rejected": -2.463876485824585,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 2.2916666666666665,
|
| 375 |
+
"grad_norm": 9.430823798801429,
|
| 376 |
+
"learning_rate": 3.307249607327433e-07,
|
| 377 |
+
"logits/chosen": -0.4775095582008362,
|
| 378 |
+
"logits/rejected": -0.3128586709499359,
|
| 379 |
+
"logps/chosen": -199.72067260742188,
|
| 380 |
+
"logps/rejected": -454.793212890625,
|
| 381 |
+
"loss": 0.3453,
|
| 382 |
+
"rewards/accuracies": 0.574999988079071,
|
| 383 |
+
"rewards/chosen": -1.1384669542312622,
|
| 384 |
+
"rewards/margins": 3.0831480026245117,
|
| 385 |
+
"rewards/rejected": -4.221614837646484,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.3958333333333335,
|
| 390 |
+
"grad_norm": 9.78181725384326,
|
| 391 |
+
"learning_rate": 2.9679481944851634e-07,
|
| 392 |
+
"logits/chosen": 0.34354180097579956,
|
| 393 |
+
"logits/rejected": 0.4988431930541992,
|
| 394 |
+
"logps/chosen": -136.43673706054688,
|
| 395 |
+
"logps/rejected": -317.75787353515625,
|
| 396 |
+
"loss": 0.396,
|
| 397 |
+
"rewards/accuracies": 0.4312500059604645,
|
| 398 |
+
"rewards/chosen": -0.8929149508476257,
|
| 399 |
+
"rewards/margins": 2.1588006019592285,
|
| 400 |
+
"rewards/rejected": -3.05171537399292,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 2.5,
|
| 405 |
+
"grad_norm": 7.751369494008921,
|
| 406 |
+
"learning_rate": 2.628646781642894e-07,
|
| 407 |
+
"logits/chosen": 0.8226224184036255,
|
| 408 |
+
"logits/rejected": 1.1203088760375977,
|
| 409 |
+
"logps/chosen": -195.84498596191406,
|
| 410 |
+
"logps/rejected": -419.219482421875,
|
| 411 |
+
"loss": 0.3849,
|
| 412 |
+
"rewards/accuracies": 0.5,
|
| 413 |
+
"rewards/chosen": -1.258272647857666,
|
| 414 |
+
"rewards/margins": 2.7024192810058594,
|
| 415 |
+
"rewards/rejected": -3.9606919288635254,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.6041666666666665,
|
| 420 |
+
"grad_norm": 8.853348347253043,
|
| 421 |
+
"learning_rate": 2.2893453688006246e-07,
|
| 422 |
+
"logits/chosen": 1.439245581626892,
|
| 423 |
+
"logits/rejected": 1.607928991317749,
|
| 424 |
+
"logps/chosen": -183.87660217285156,
|
| 425 |
+
"logps/rejected": -415.3749084472656,
|
| 426 |
+
"loss": 0.3591,
|
| 427 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 428 |
+
"rewards/chosen": -1.2962507009506226,
|
| 429 |
+
"rewards/margins": 2.7243804931640625,
|
| 430 |
+
"rewards/rejected": -4.020631313323975,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 2.7083333333333335,
|
| 435 |
+
"grad_norm": 6.5857498778744805,
|
| 436 |
+
"learning_rate": 1.9500439559583547e-07,
|
| 437 |
+
"logits/chosen": 1.5265406370162964,
|
| 438 |
+
"logits/rejected": 1.5034195184707642,
|
| 439 |
+
"logps/chosen": -151.31393432617188,
|
| 440 |
+
"logps/rejected": -378.006103515625,
|
| 441 |
+
"loss": 0.3878,
|
| 442 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 443 |
+
"rewards/chosen": -1.0861542224884033,
|
| 444 |
+
"rewards/margins": 2.4694628715515137,
|
| 445 |
+
"rewards/rejected": -3.555617094039917,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.8125,
|
| 450 |
+
"grad_norm": 8.328330751046998,
|
| 451 |
+
"learning_rate": 1.610742543116085e-07,
|
| 452 |
+
"logits/chosen": 1.2100938558578491,
|
| 453 |
+
"logits/rejected": 1.192025065422058,
|
| 454 |
+
"logps/chosen": -182.81759643554688,
|
| 455 |
+
"logps/rejected": -417.20623779296875,
|
| 456 |
+
"loss": 0.365,
|
| 457 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 458 |
+
"rewards/chosen": -1.2372219562530518,
|
| 459 |
+
"rewards/margins": 2.7132503986358643,
|
| 460 |
+
"rewards/rejected": -3.950472354888916,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 2.9166666666666665,
|
| 465 |
+
"grad_norm": 9.271318342539349,
|
| 466 |
+
"learning_rate": 1.2714411302738161e-07,
|
| 467 |
+
"logits/chosen": 1.5337226390838623,
|
| 468 |
+
"logits/rejected": 1.8394181728363037,
|
| 469 |
+
"logps/chosen": -204.70779418945312,
|
| 470 |
+
"logps/rejected": -450.3301696777344,
|
| 471 |
+
"loss": 0.3737,
|
| 472 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 473 |
+
"rewards/chosen": -1.3911634683609009,
|
| 474 |
+
"rewards/margins": 2.897785186767578,
|
| 475 |
+
"rewards/rejected": -4.288949012756348,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 3.0,
|
| 480 |
+
"step": 288,
|
| 481 |
+
"total_flos": 0.0,
|
| 482 |
+
"train_loss": 0.5020060315728188,
|
| 483 |
+
"train_runtime": 3747.4353,
|
| 484 |
+
"train_samples_per_second": 4.894,
|
| 485 |
+
"train_steps_per_second": 0.077
|
| 486 |
+
}
|
| 487 |
+
],
|
| 488 |
+
"logging_steps": 10,
|
| 489 |
+
"max_steps": 288,
|
| 490 |
+
"num_input_tokens_seen": 0,
|
| 491 |
+
"num_train_epochs": 3,
|
| 492 |
+
"save_steps": 100,
|
| 493 |
+
"stateful_callbacks": {
|
| 494 |
+
"TrainerControl": {
|
| 495 |
+
"args": {
|
| 496 |
+
"should_epoch_stop": false,
|
| 497 |
+
"should_evaluate": false,
|
| 498 |
+
"should_log": false,
|
| 499 |
+
"should_save": true,
|
| 500 |
+
"should_training_stop": true
|
| 501 |
+
},
|
| 502 |
+
"attributes": {}
|
| 503 |
+
}
|
| 504 |
+
},
|
| 505 |
+
"total_flos": 0.0,
|
| 506 |
+
"train_batch_size": 8,
|
| 507 |
+
"trial_name": null,
|
| 508 |
+
"trial_params": null
|
| 509 |
+
}
|