Model save
Browse files- README.md +74 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +358 -0
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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-0.0-0.4-polynomial-2
|
| 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-0.0-0.4-polynomial-2
|
| 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: 0.7712
|
| 22 |
+
- Rewards/chosen: 0.7556
|
| 23 |
+
- Rewards/rejected: 0.1333
|
| 24 |
+
- Rewards/accuracies: 0.3313
|
| 25 |
+
- Rewards/margins: 0.6223
|
| 26 |
+
- Logps/rejected: -80.8200
|
| 27 |
+
- Logps/chosen: -72.7430
|
| 28 |
+
- Logits/rejected: -2.5200
|
| 29 |
+
- Logits/chosen: -2.5356
|
| 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: 8.772160997458124e-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: 2
|
| 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 |
+
| 0.5775 | 1.0417 | 100 | 0.7712 | 0.7556 | 0.1333 | 0.3313 | 0.6223 | -80.8200 | -72.7430 | -2.5200 | -2.5356 |
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
### Framework versions
|
| 70 |
+
|
| 71 |
+
- Transformers 4.44.2
|
| 72 |
+
- Pytorch 2.4.0
|
| 73 |
+
- Datasets 2.21.0
|
| 74 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5438582648833593,
|
| 5 |
+
"train_runtime": 2038.8345,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 5.997,
|
| 8 |
+
"train_steps_per_second": 0.094
|
| 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": 2.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5438582648833593,
|
| 5 |
+
"train_runtime": 2038.8345,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 5.997,
|
| 8 |
+
"train_steps_per_second": 0.094
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,358 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 192,
|
| 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": 284.73684625892474,
|
| 14 |
+
"learning_rate": 4.386080498729062e-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": 269.4644096553184,
|
| 29 |
+
"learning_rate": 4.386080498729062e-07,
|
| 30 |
+
"logits/chosen": -2.5559744834899902,
|
| 31 |
+
"logits/rejected": -2.5380678176879883,
|
| 32 |
+
"logps/chosen": -87.90437316894531,
|
| 33 |
+
"logps/rejected": -80.99159240722656,
|
| 34 |
+
"loss": 0.698,
|
| 35 |
+
"rewards/accuracies": 0.2013888955116272,
|
| 36 |
+
"rewards/chosen": 0.004661789629608393,
|
| 37 |
+
"rewards/margins": -0.01561733242124319,
|
| 38 |
+
"rewards/rejected": 0.02027912251651287,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.20833333333333334,
|
| 43 |
+
"grad_norm": 171.18434231702736,
|
| 44 |
+
"learning_rate": 8.772160997458124e-07,
|
| 45 |
+
"logits/chosen": -2.604264736175537,
|
| 46 |
+
"logits/rejected": -2.5549380779266357,
|
| 47 |
+
"logps/chosen": -102.35679626464844,
|
| 48 |
+
"logps/rejected": -89.08101654052734,
|
| 49 |
+
"loss": 0.6597,
|
| 50 |
+
"rewards/accuracies": 0.35624998807907104,
|
| 51 |
+
"rewards/chosen": 0.4296715259552002,
|
| 52 |
+
"rewards/margins": 0.08247259259223938,
|
| 53 |
+
"rewards/rejected": 0.34719887375831604,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.3125,
|
| 58 |
+
"grad_norm": 307.77121119192236,
|
| 59 |
+
"learning_rate": 8.320291172024512e-07,
|
| 60 |
+
"logits/chosen": -2.482062578201294,
|
| 61 |
+
"logits/rejected": -2.4983456134796143,
|
| 62 |
+
"logps/chosen": -64.65339660644531,
|
| 63 |
+
"logps/rejected": -73.08040618896484,
|
| 64 |
+
"loss": 0.7129,
|
| 65 |
+
"rewards/accuracies": 0.3499999940395355,
|
| 66 |
+
"rewards/chosen": 1.2031724452972412,
|
| 67 |
+
"rewards/margins": 0.40556639432907104,
|
| 68 |
+
"rewards/rejected": 0.7976058721542358,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.4166666666666667,
|
| 73 |
+
"grad_norm": 247.69654790448098,
|
| 74 |
+
"learning_rate": 7.868421346590901e-07,
|
| 75 |
+
"logits/chosen": -2.5462698936462402,
|
| 76 |
+
"logits/rejected": -2.5348854064941406,
|
| 77 |
+
"logps/chosen": -68.5112075805664,
|
| 78 |
+
"logps/rejected": -67.58045196533203,
|
| 79 |
+
"loss": 0.7089,
|
| 80 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 81 |
+
"rewards/chosen": 1.5711699724197388,
|
| 82 |
+
"rewards/margins": 0.21717211604118347,
|
| 83 |
+
"rewards/rejected": 1.353997826576233,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.5208333333333334,
|
| 88 |
+
"grad_norm": 245.28947413610663,
|
| 89 |
+
"learning_rate": 7.416551521157288e-07,
|
| 90 |
+
"logits/chosen": -2.4909956455230713,
|
| 91 |
+
"logits/rejected": -2.5039918422698975,
|
| 92 |
+
"logps/chosen": -47.20299530029297,
|
| 93 |
+
"logps/rejected": -55.72983932495117,
|
| 94 |
+
"loss": 0.7414,
|
| 95 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 96 |
+
"rewards/chosen": 1.5983082056045532,
|
| 97 |
+
"rewards/margins": 0.22424061596393585,
|
| 98 |
+
"rewards/rejected": 1.3740675449371338,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.625,
|
| 103 |
+
"grad_norm": 185.80480485868813,
|
| 104 |
+
"learning_rate": 6.964681695723677e-07,
|
| 105 |
+
"logits/chosen": -2.6212210655212402,
|
| 106 |
+
"logits/rejected": -2.6057589054107666,
|
| 107 |
+
"logps/chosen": -73.73361206054688,
|
| 108 |
+
"logps/rejected": -75.0650863647461,
|
| 109 |
+
"loss": 0.7278,
|
| 110 |
+
"rewards/accuracies": 0.3125,
|
| 111 |
+
"rewards/chosen": 2.306000232696533,
|
| 112 |
+
"rewards/margins": 0.4594852924346924,
|
| 113 |
+
"rewards/rejected": 1.8465149402618408,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.7291666666666666,
|
| 118 |
+
"grad_norm": 238.44222468726358,
|
| 119 |
+
"learning_rate": 6.512811870290064e-07,
|
| 120 |
+
"logits/chosen": -2.616278886795044,
|
| 121 |
+
"logits/rejected": -2.6074726581573486,
|
| 122 |
+
"logps/chosen": -91.4566421508789,
|
| 123 |
+
"logps/rejected": -81.8633041381836,
|
| 124 |
+
"loss": 0.7043,
|
| 125 |
+
"rewards/accuracies": 0.35624998807907104,
|
| 126 |
+
"rewards/chosen": 2.112656354904175,
|
| 127 |
+
"rewards/margins": 0.5624496936798096,
|
| 128 |
+
"rewards/rejected": 1.5502065420150757,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.8333333333333334,
|
| 133 |
+
"grad_norm": 359.10559072083385,
|
| 134 |
+
"learning_rate": 6.060942044856453e-07,
|
| 135 |
+
"logits/chosen": -2.593449115753174,
|
| 136 |
+
"logits/rejected": -2.54801607131958,
|
| 137 |
+
"logps/chosen": -80.93285369873047,
|
| 138 |
+
"logps/rejected": -76.7210693359375,
|
| 139 |
+
"loss": 0.7058,
|
| 140 |
+
"rewards/accuracies": 0.41874998807907104,
|
| 141 |
+
"rewards/chosen": 2.0366339683532715,
|
| 142 |
+
"rewards/margins": 0.886386513710022,
|
| 143 |
+
"rewards/rejected": 1.1502474546432495,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.9375,
|
| 148 |
+
"grad_norm": 283.4561861873084,
|
| 149 |
+
"learning_rate": 5.609072219422841e-07,
|
| 150 |
+
"logits/chosen": -2.5106282234191895,
|
| 151 |
+
"logits/rejected": -2.506967306137085,
|
| 152 |
+
"logps/chosen": -51.3546028137207,
|
| 153 |
+
"logps/rejected": -61.58504104614258,
|
| 154 |
+
"loss": 0.7583,
|
| 155 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 156 |
+
"rewards/chosen": 1.134612798690796,
|
| 157 |
+
"rewards/margins": 0.32073575258255005,
|
| 158 |
+
"rewards/rejected": 0.8138772249221802,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.0416666666666667,
|
| 163 |
+
"grad_norm": 21.926819298392022,
|
| 164 |
+
"learning_rate": 5.157202393989229e-07,
|
| 165 |
+
"logits/chosen": -2.4917004108428955,
|
| 166 |
+
"logits/rejected": -2.476295232772827,
|
| 167 |
+
"logps/chosen": -65.5259780883789,
|
| 168 |
+
"logps/rejected": -74.4820327758789,
|
| 169 |
+
"loss": 0.5775,
|
| 170 |
+
"rewards/accuracies": 0.40625,
|
| 171 |
+
"rewards/chosen": 3.455796718597412,
|
| 172 |
+
"rewards/margins": 4.517679691314697,
|
| 173 |
+
"rewards/rejected": -1.0618823766708374,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.0416666666666667,
|
| 178 |
+
"eval_logits/chosen": -2.5355653762817383,
|
| 179 |
+
"eval_logits/rejected": -2.5199732780456543,
|
| 180 |
+
"eval_logps/chosen": -72.7430191040039,
|
| 181 |
+
"eval_logps/rejected": -80.81999206542969,
|
| 182 |
+
"eval_loss": 0.7711760401725769,
|
| 183 |
+
"eval_rewards/accuracies": 0.3313491940498352,
|
| 184 |
+
"eval_rewards/chosen": 0.7555689811706543,
|
| 185 |
+
"eval_rewards/margins": 0.6222997903823853,
|
| 186 |
+
"eval_rewards/rejected": 0.1332692801952362,
|
| 187 |
+
"eval_runtime": 113.7538,
|
| 188 |
+
"eval_samples_per_second": 17.582,
|
| 189 |
+
"eval_steps_per_second": 0.554,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.1458333333333333,
|
| 194 |
+
"grad_norm": 11.001098983606164,
|
| 195 |
+
"learning_rate": 4.7053325685556164e-07,
|
| 196 |
+
"logits/chosen": -2.5152523517608643,
|
| 197 |
+
"logits/rejected": -2.542959690093994,
|
| 198 |
+
"logps/chosen": -55.308494567871094,
|
| 199 |
+
"logps/rejected": -93.90345001220703,
|
| 200 |
+
"loss": 0.3605,
|
| 201 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 202 |
+
"rewards/chosen": 4.347947120666504,
|
| 203 |
+
"rewards/margins": 9.715516090393066,
|
| 204 |
+
"rewards/rejected": -5.367568492889404,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.25,
|
| 209 |
+
"grad_norm": 5.506771482244887,
|
| 210 |
+
"learning_rate": 4.2534627431220053e-07,
|
| 211 |
+
"logits/chosen": -2.5518078804016113,
|
| 212 |
+
"logits/rejected": -2.5186076164245605,
|
| 213 |
+
"logps/chosen": -95.32207489013672,
|
| 214 |
+
"logps/rejected": -115.25175476074219,
|
| 215 |
+
"loss": 0.3819,
|
| 216 |
+
"rewards/accuracies": 0.581250011920929,
|
| 217 |
+
"rewards/chosen": 5.481812477111816,
|
| 218 |
+
"rewards/margins": 14.953575134277344,
|
| 219 |
+
"rewards/rejected": -9.471762657165527,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.3541666666666667,
|
| 224 |
+
"grad_norm": 80.99035607988066,
|
| 225 |
+
"learning_rate": 3.801592917688393e-07,
|
| 226 |
+
"logits/chosen": -2.4438960552215576,
|
| 227 |
+
"logits/rejected": -2.44427490234375,
|
| 228 |
+
"logps/chosen": -80.55229187011719,
|
| 229 |
+
"logps/rejected": -110.3847885131836,
|
| 230 |
+
"loss": 0.3763,
|
| 231 |
+
"rewards/accuracies": 0.46875,
|
| 232 |
+
"rewards/chosen": 2.599090814590454,
|
| 233 |
+
"rewards/margins": 11.696044921875,
|
| 234 |
+
"rewards/rejected": -9.096954345703125,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.4583333333333333,
|
| 239 |
+
"grad_norm": 81.79049224504786,
|
| 240 |
+
"learning_rate": 3.3497230922547816e-07,
|
| 241 |
+
"logits/chosen": -2.442150831222534,
|
| 242 |
+
"logits/rejected": -2.4587113857269287,
|
| 243 |
+
"logps/chosen": -47.28990936279297,
|
| 244 |
+
"logps/rejected": -80.22529602050781,
|
| 245 |
+
"loss": 0.3766,
|
| 246 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 247 |
+
"rewards/chosen": 0.8634545207023621,
|
| 248 |
+
"rewards/margins": 7.3911895751953125,
|
| 249 |
+
"rewards/rejected": -6.527735233306885,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.5625,
|
| 254 |
+
"grad_norm": 2.5199873819499934,
|
| 255 |
+
"learning_rate": 2.89785326682117e-07,
|
| 256 |
+
"logits/chosen": -2.4821057319641113,
|
| 257 |
+
"logits/rejected": -2.4555060863494873,
|
| 258 |
+
"logps/chosen": -67.4425277709961,
|
| 259 |
+
"logps/rejected": -90.55575561523438,
|
| 260 |
+
"loss": 0.3829,
|
| 261 |
+
"rewards/accuracies": 0.46875,
|
| 262 |
+
"rewards/chosen": 2.033623456954956,
|
| 263 |
+
"rewards/margins": 9.372200012207031,
|
| 264 |
+
"rewards/rejected": -7.338575839996338,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.6666666666666665,
|
| 269 |
+
"grad_norm": 1.8477188844357901,
|
| 270 |
+
"learning_rate": 2.445983441387558e-07,
|
| 271 |
+
"logits/chosen": -2.4257683753967285,
|
| 272 |
+
"logits/rejected": -2.4348573684692383,
|
| 273 |
+
"logps/chosen": -72.42179870605469,
|
| 274 |
+
"logps/rejected": -111.45719146728516,
|
| 275 |
+
"loss": 0.3664,
|
| 276 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 277 |
+
"rewards/chosen": 3.675116777420044,
|
| 278 |
+
"rewards/margins": 13.840252876281738,
|
| 279 |
+
"rewards/rejected": -10.165136337280273,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.7708333333333335,
|
| 284 |
+
"grad_norm": 27.363913250696022,
|
| 285 |
+
"learning_rate": 1.9941136159539462e-07,
|
| 286 |
+
"logits/chosen": -2.3719429969787598,
|
| 287 |
+
"logits/rejected": -2.3617918491363525,
|
| 288 |
+
"logps/chosen": -68.02982330322266,
|
| 289 |
+
"logps/rejected": -105.62237548828125,
|
| 290 |
+
"loss": 0.394,
|
| 291 |
+
"rewards/accuracies": 0.53125,
|
| 292 |
+
"rewards/chosen": 3.107600212097168,
|
| 293 |
+
"rewards/margins": 12.890829086303711,
|
| 294 |
+
"rewards/rejected": -9.783227920532227,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.875,
|
| 299 |
+
"grad_norm": 43.422774421532836,
|
| 300 |
+
"learning_rate": 1.5422437905203344e-07,
|
| 301 |
+
"logits/chosen": -2.4057819843292236,
|
| 302 |
+
"logits/rejected": -2.389813184738159,
|
| 303 |
+
"logps/chosen": -64.71813201904297,
|
| 304 |
+
"logps/rejected": -92.5549545288086,
|
| 305 |
+
"loss": 0.3757,
|
| 306 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 307 |
+
"rewards/chosen": 3.084803581237793,
|
| 308 |
+
"rewards/margins": 11.271722793579102,
|
| 309 |
+
"rewards/rejected": -8.186918258666992,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.9791666666666665,
|
| 314 |
+
"grad_norm": 2.3019013228739755,
|
| 315 |
+
"learning_rate": 1.0903739650867225e-07,
|
| 316 |
+
"logits/chosen": -2.454073429107666,
|
| 317 |
+
"logits/rejected": -2.4500691890716553,
|
| 318 |
+
"logps/chosen": -76.17430877685547,
|
| 319 |
+
"logps/rejected": -113.40323638916016,
|
| 320 |
+
"loss": 0.3748,
|
| 321 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 322 |
+
"rewards/chosen": 5.310562610626221,
|
| 323 |
+
"rewards/margins": 15.156728744506836,
|
| 324 |
+
"rewards/rejected": -9.846165657043457,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.0,
|
| 329 |
+
"step": 192,
|
| 330 |
+
"total_flos": 0.0,
|
| 331 |
+
"train_loss": 0.5438582648833593,
|
| 332 |
+
"train_runtime": 2038.8345,
|
| 333 |
+
"train_samples_per_second": 5.997,
|
| 334 |
+
"train_steps_per_second": 0.094
|
| 335 |
+
}
|
| 336 |
+
],
|
| 337 |
+
"logging_steps": 10,
|
| 338 |
+
"max_steps": 192,
|
| 339 |
+
"num_input_tokens_seen": 0,
|
| 340 |
+
"num_train_epochs": 2,
|
| 341 |
+
"save_steps": 100,
|
| 342 |
+
"stateful_callbacks": {
|
| 343 |
+
"TrainerControl": {
|
| 344 |
+
"args": {
|
| 345 |
+
"should_epoch_stop": false,
|
| 346 |
+
"should_evaluate": false,
|
| 347 |
+
"should_log": false,
|
| 348 |
+
"should_save": true,
|
| 349 |
+
"should_training_stop": true
|
| 350 |
+
},
|
| 351 |
+
"attributes": {}
|
| 352 |
+
}
|
| 353 |
+
},
|
| 354 |
+
"total_flos": 0.0,
|
| 355 |
+
"train_batch_size": 8,
|
| 356 |
+
"trial_name": null,
|
| 357 |
+
"trial_params": null
|
| 358 |
+
}
|