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-0.0-0.0-cosine-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-0.0-0.0-cosine-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: 0.6710
|
| 22 |
+
- Rewards/chosen: -0.1518
|
| 23 |
+
- Rewards/rejected: -0.2380
|
| 24 |
+
- Rewards/accuracies: 0.3155
|
| 25 |
+
- Rewards/margins: 0.0862
|
| 26 |
+
- Logps/rejected: -327.2254
|
| 27 |
+
- Logps/chosen: -231.4086
|
| 28 |
+
- Logits/rejected: 2.3998
|
| 29 |
+
- Logits/chosen: 2.2227
|
| 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: 8.932373406244385e-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: cosine
|
| 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 |
+
| 0.66 | 1.0417 | 100 | 0.6710 | -0.1518 | -0.2380 | 0.3155 | 0.0862 | -327.2254 | -231.4086 | 2.3998 | 2.2227 |
|
| 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": 0.6542075766871372,
|
| 5 |
+
"train_runtime": 2362.1376,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 5.176,
|
| 8 |
+
"train_steps_per_second": 0.081
|
| 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": 0.6542075766871372,
|
| 5 |
+
"train_runtime": 2362.1376,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 5.176,
|
| 8 |
+
"train_steps_per_second": 0.081
|
| 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": 0.6369225577240899,
|
| 14 |
+
"learning_rate": 4.4661867031221926e-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": 0.5699844472823067,
|
| 29 |
+
"learning_rate": 4.4661867031221926e-07,
|
| 30 |
+
"logits/chosen": -2.5559396743774414,
|
| 31 |
+
"logits/rejected": -2.5378711223602295,
|
| 32 |
+
"logps/chosen": -87.90701293945312,
|
| 33 |
+
"logps/rejected": -81.000732421875,
|
| 34 |
+
"loss": 0.6931,
|
| 35 |
+
"rewards/accuracies": 0.1944444477558136,
|
| 36 |
+
"rewards/chosen": 7.878588803578168e-06,
|
| 37 |
+
"rewards/margins": -2.865187343559228e-05,
|
| 38 |
+
"rewards/rejected": 3.653046223917045e-05,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.20833333333333334,
|
| 43 |
+
"grad_norm": 0.5220617909669454,
|
| 44 |
+
"learning_rate": 8.932373406244385e-07,
|
| 45 |
+
"logits/chosen": -2.6017892360687256,
|
| 46 |
+
"logits/rejected": -2.5524039268493652,
|
| 47 |
+
"logps/chosen": -102.29816436767578,
|
| 48 |
+
"logps/rejected": -88.93941497802734,
|
| 49 |
+
"loss": 0.6929,
|
| 50 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 51 |
+
"rewards/chosen": 0.0010181397665292025,
|
| 52 |
+
"rewards/margins": 0.00010430384281789884,
|
| 53 |
+
"rewards/rejected": 0.0009138360619544983,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.3125,
|
| 58 |
+
"grad_norm": 0.8025622814810965,
|
| 59 |
+
"learning_rate": 8.85808141328471e-07,
|
| 60 |
+
"logits/chosen": -2.4462294578552246,
|
| 61 |
+
"logits/rejected": -2.4619638919830322,
|
| 62 |
+
"logps/chosen": -66.71989440917969,
|
| 63 |
+
"logps/rejected": -76.12188720703125,
|
| 64 |
+
"loss": 0.6922,
|
| 65 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 66 |
+
"rewards/chosen": 0.0006937496364116669,
|
| 67 |
+
"rewards/margins": 0.001850384520366788,
|
| 68 |
+
"rewards/rejected": -0.0011566350003704429,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.4166666666666667,
|
| 73 |
+
"grad_norm": 0.8299819796102439,
|
| 74 |
+
"learning_rate": 8.63767702844543e-07,
|
| 75 |
+
"logits/chosen": -2.4784650802612305,
|
| 76 |
+
"logits/rejected": -2.4647819995880127,
|
| 77 |
+
"logps/chosen": -74.95284271240234,
|
| 78 |
+
"logps/rejected": -78.62578582763672,
|
| 79 |
+
"loss": 0.6904,
|
| 80 |
+
"rewards/accuracies": 0.28125,
|
| 81 |
+
"rewards/chosen": -0.002713928697630763,
|
| 82 |
+
"rewards/margins": 0.004938088357448578,
|
| 83 |
+
"rewards/rejected": -0.007652017287909985,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.5208333333333334,
|
| 88 |
+
"grad_norm": 1.7129675129738506,
|
| 89 |
+
"learning_rate": 8.278492807257787e-07,
|
| 90 |
+
"logits/chosen": -2.275789260864258,
|
| 91 |
+
"logits/rejected": -2.291548252105713,
|
| 92 |
+
"logps/chosen": -55.84526824951172,
|
| 93 |
+
"logps/rejected": -68.4447250366211,
|
| 94 |
+
"loss": 0.6892,
|
| 95 |
+
"rewards/accuracies": 0.20000000298023224,
|
| 96 |
+
"rewards/chosen": -0.004781405441462994,
|
| 97 |
+
"rewards/margins": 0.004440293647348881,
|
| 98 |
+
"rewards/rejected": -0.009221700020134449,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.625,
|
| 103 |
+
"grad_norm": 1.6503123985933317,
|
| 104 |
+
"learning_rate": 7.792478322547763e-07,
|
| 105 |
+
"logits/chosen": -2.007650136947632,
|
| 106 |
+
"logits/rejected": -1.9700886011123657,
|
| 107 |
+
"logps/chosen": -114.91951751708984,
|
| 108 |
+
"logps/rejected": -133.2468719482422,
|
| 109 |
+
"loss": 0.6865,
|
| 110 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 111 |
+
"rewards/chosen": -0.03467016667127609,
|
| 112 |
+
"rewards/margins": 0.017464537173509598,
|
| 113 |
+
"rewards/rejected": -0.05213470384478569,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.7291666666666666,
|
| 118 |
+
"grad_norm": 2.2984298301869166,
|
| 119 |
+
"learning_rate": 7.195802618323811e-07,
|
| 120 |
+
"logits/chosen": -0.7167578935623169,
|
| 121 |
+
"logits/rejected": -0.5914499163627625,
|
| 122 |
+
"logps/chosen": -169.179443359375,
|
| 123 |
+
"logps/rejected": -187.8536376953125,
|
| 124 |
+
"loss": 0.6802,
|
| 125 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 126 |
+
"rewards/chosen": -0.07043690979480743,
|
| 127 |
+
"rewards/margins": 0.02859547734260559,
|
| 128 |
+
"rewards/rejected": -0.09903239458799362,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.8333333333333334,
|
| 133 |
+
"grad_norm": 7.779524343348389,
|
| 134 |
+
"learning_rate": 6.508316287570068e-07,
|
| 135 |
+
"logits/chosen": 0.3923811912536621,
|
| 136 |
+
"logits/rejected": 0.710164487361908,
|
| 137 |
+
"logps/chosen": -222.629638671875,
|
| 138 |
+
"logps/rejected": -283.06634521484375,
|
| 139 |
+
"loss": 0.669,
|
| 140 |
+
"rewards/accuracies": 0.3687500059604645,
|
| 141 |
+
"rewards/chosen": -0.13247501850128174,
|
| 142 |
+
"rewards/margins": 0.06450365483760834,
|
| 143 |
+
"rewards/rejected": -0.19697865843772888,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.9375,
|
| 148 |
+
"grad_norm": 6.759384309065668,
|
| 149 |
+
"learning_rate": 5.752891069888697e-07,
|
| 150 |
+
"logits/chosen": 1.1880756616592407,
|
| 151 |
+
"logits/rejected": 1.3526252508163452,
|
| 152 |
+
"logps/chosen": -208.9159393310547,
|
| 153 |
+
"logps/rejected": -350.1947021484375,
|
| 154 |
+
"loss": 0.6679,
|
| 155 |
+
"rewards/accuracies": 0.26249998807907104,
|
| 156 |
+
"rewards/chosen": -0.14983566105365753,
|
| 157 |
+
"rewards/margins": 0.12745192646980286,
|
| 158 |
+
"rewards/rejected": -0.2772876024246216,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.0416666666666667,
|
| 163 |
+
"grad_norm": 6.64561814006091,
|
| 164 |
+
"learning_rate": 4.954658939683863e-07,
|
| 165 |
+
"logits/chosen": 1.6723073720932007,
|
| 166 |
+
"logits/rejected": 2.053896427154541,
|
| 167 |
+
"logps/chosen": -224.61581420898438,
|
| 168 |
+
"logps/rejected": -340.3245849609375,
|
| 169 |
+
"loss": 0.66,
|
| 170 |
+
"rewards/accuracies": 0.35624998807907104,
|
| 171 |
+
"rewards/chosen": -0.14611998200416565,
|
| 172 |
+
"rewards/margins": 0.11334720999002457,
|
| 173 |
+
"rewards/rejected": -0.2594671845436096,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.0416666666666667,
|
| 178 |
+
"eval_logits/chosen": 2.222659111022949,
|
| 179 |
+
"eval_logits/rejected": 2.3998076915740967,
|
| 180 |
+
"eval_logps/chosen": -231.40858459472656,
|
| 181 |
+
"eval_logps/rejected": -327.22540283203125,
|
| 182 |
+
"eval_loss": 0.6709679961204529,
|
| 183 |
+
"eval_rewards/accuracies": 0.3154761791229248,
|
| 184 |
+
"eval_rewards/chosen": -0.15175169706344604,
|
| 185 |
+
"eval_rewards/margins": 0.08624394983053207,
|
| 186 |
+
"eval_rewards/rejected": -0.2379956692457199,
|
| 187 |
+
"eval_runtime": 113.6716,
|
| 188 |
+
"eval_samples_per_second": 17.595,
|
| 189 |
+
"eval_steps_per_second": 0.554,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.1458333333333333,
|
| 194 |
+
"grad_norm": 10.14024135671265,
|
| 195 |
+
"learning_rate": 4.140175999393427e-07,
|
| 196 |
+
"logits/chosen": 2.7885522842407227,
|
| 197 |
+
"logits/rejected": 2.5952858924865723,
|
| 198 |
+
"logps/chosen": -217.1932830810547,
|
| 199 |
+
"logps/rejected": -423.316162109375,
|
| 200 |
+
"loss": 0.6344,
|
| 201 |
+
"rewards/accuracies": 0.375,
|
| 202 |
+
"rewards/chosen": -0.14682665467262268,
|
| 203 |
+
"rewards/margins": 0.18375235795974731,
|
| 204 |
+
"rewards/rejected": -0.3305789828300476,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.25,
|
| 209 |
+
"grad_norm": 5.687581380009954,
|
| 210 |
+
"learning_rate": 3.3365389939087466e-07,
|
| 211 |
+
"logits/chosen": 2.281104564666748,
|
| 212 |
+
"logits/rejected": 2.7267863750457764,
|
| 213 |
+
"logps/chosen": -382.77508544921875,
|
| 214 |
+
"logps/rejected": -542.299560546875,
|
| 215 |
+
"loss": 0.6531,
|
| 216 |
+
"rewards/accuracies": 0.4375,
|
| 217 |
+
"rewards/chosen": -0.2657240629196167,
|
| 218 |
+
"rewards/margins": 0.1684594750404358,
|
| 219 |
+
"rewards/rejected": -0.43418359756469727,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.3541666666666667,
|
| 224 |
+
"grad_norm": 8.59716391066745,
|
| 225 |
+
"learning_rate": 2.5704838385518876e-07,
|
| 226 |
+
"logits/chosen": 2.97074556350708,
|
| 227 |
+
"logits/rejected": 3.06956148147583,
|
| 228 |
+
"logps/chosen": -262.03167724609375,
|
| 229 |
+
"logps/rejected": -430.7930603027344,
|
| 230 |
+
"loss": 0.628,
|
| 231 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 232 |
+
"rewards/chosen": -0.16968943178653717,
|
| 233 |
+
"rewards/margins": 0.16052642464637756,
|
| 234 |
+
"rewards/rejected": -0.3302158713340759,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.4583333333333333,
|
| 239 |
+
"grad_norm": 11.108890574756282,
|
| 240 |
+
"learning_rate": 1.8674961513641657e-07,
|
| 241 |
+
"logits/chosen": 3.3152003288269043,
|
| 242 |
+
"logits/rejected": 3.317988634109497,
|
| 243 |
+
"logps/chosen": -273.95831298828125,
|
| 244 |
+
"logps/rejected": -546.5203247070312,
|
| 245 |
+
"loss": 0.6343,
|
| 246 |
+
"rewards/accuracies": 0.3499999940395355,
|
| 247 |
+
"rewards/chosen": -0.21727445721626282,
|
| 248 |
+
"rewards/margins": 0.2482767552137375,
|
| 249 |
+
"rewards/rejected": -0.4655512273311615,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.5625,
|
| 254 |
+
"grad_norm": 6.0584348916117925,
|
| 255 |
+
"learning_rate": 1.250963381092756e-07,
|
| 256 |
+
"logits/chosen": 3.313689708709717,
|
| 257 |
+
"logits/rejected": 3.6598522663116455,
|
| 258 |
+
"logps/chosen": -273.2975769042969,
|
| 259 |
+
"logps/rejected": -551.4356689453125,
|
| 260 |
+
"loss": 0.6207,
|
| 261 |
+
"rewards/accuracies": 0.38749998807907104,
|
| 262 |
+
"rewards/chosen": -0.19452792406082153,
|
| 263 |
+
"rewards/margins": 0.2676007151603699,
|
| 264 |
+
"rewards/rejected": -0.4621286988258362,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.6666666666666665,
|
| 269 |
+
"grad_norm": 11.564568090628018,
|
| 270 |
+
"learning_rate": 7.413967384294965e-08,
|
| 271 |
+
"logits/chosen": 3.5584030151367188,
|
| 272 |
+
"logits/rejected": 3.723942279815674,
|
| 273 |
+
"logps/chosen": -357.8267517089844,
|
| 274 |
+
"logps/rejected": -687.1414794921875,
|
| 275 |
+
"loss": 0.6181,
|
| 276 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 277 |
+
"rewards/chosen": -0.2677861154079437,
|
| 278 |
+
"rewards/margins": 0.31169238686561584,
|
| 279 |
+
"rewards/rejected": -0.5794785022735596,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.7708333333333335,
|
| 284 |
+
"grad_norm": 12.348100258027504,
|
| 285 |
+
"learning_rate": 3.557488157963567e-08,
|
| 286 |
+
"logits/chosen": 4.085668087005615,
|
| 287 |
+
"logits/rejected": 4.25767183303833,
|
| 288 |
+
"logps/chosen": -461.731689453125,
|
| 289 |
+
"logps/rejected": -780.9080810546875,
|
| 290 |
+
"loss": 0.6175,
|
| 291 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 292 |
+
"rewards/chosen": -0.3737877607345581,
|
| 293 |
+
"rewards/margins": 0.30115869641304016,
|
| 294 |
+
"rewards/rejected": -0.6749464273452759,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.875,
|
| 299 |
+
"grad_norm": 14.884569591942372,
|
| 300 |
+
"learning_rate": 1.0684959754358576e-08,
|
| 301 |
+
"logits/chosen": 3.8572006225585938,
|
| 302 |
+
"logits/rejected": 4.063995361328125,
|
| 303 |
+
"logps/chosen": -347.04864501953125,
|
| 304 |
+
"logps/rejected": -606.2640380859375,
|
| 305 |
+
"loss": 0.617,
|
| 306 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 307 |
+
"rewards/chosen": -0.26613375544548035,
|
| 308 |
+
"rewards/margins": 0.248983234167099,
|
| 309 |
+
"rewards/rejected": -0.5151170492172241,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.9791666666666665,
|
| 314 |
+
"grad_norm": 7.915820342970624,
|
| 315 |
+
"learning_rate": 2.979623737803134e-10,
|
| 316 |
+
"logits/chosen": 3.7023074626922607,
|
| 317 |
+
"logits/rejected": 3.886380434036255,
|
| 318 |
+
"logps/chosen": -341.0790710449219,
|
| 319 |
+
"logps/rejected": -703.4373779296875,
|
| 320 |
+
"loss": 0.6085,
|
| 321 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 322 |
+
"rewards/chosen": -0.24430128931999207,
|
| 323 |
+
"rewards/margins": 0.3483408987522125,
|
| 324 |
+
"rewards/rejected": -0.5926421880722046,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.0,
|
| 329 |
+
"step": 192,
|
| 330 |
+
"total_flos": 0.0,
|
| 331 |
+
"train_loss": 0.6542075766871372,
|
| 332 |
+
"train_runtime": 2362.1376,
|
| 333 |
+
"train_samples_per_second": 5.176,
|
| 334 |
+
"train_steps_per_second": 0.081
|
| 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 |
+
}
|