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 +284 -0
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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-ultrabin-low-margin
|
| 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-ultrabin-low-margin
|
| 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: 0.6687
|
| 21 |
+
- Rewards/chosen: -0.0876
|
| 22 |
+
- Rewards/rejected: -0.1545
|
| 23 |
+
- Rewards/accuracies: 0.5742
|
| 24 |
+
- Rewards/margins: 0.0669
|
| 25 |
+
- Logps/rejected: -278.1130
|
| 26 |
+
- Logps/chosen: -271.3896
|
| 27 |
+
- Logits/rejected: -2.4595
|
| 28 |
+
- Logits/chosen: -2.4953
|
| 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 |
+
| 0.6816 | 0.3484 | 50 | 0.6725 | -0.0071 | -0.0549 | 0.5898 | 0.0478 | -268.1548 | -263.3413 | -2.5366 | -2.5709 |
|
| 66 |
+
| 0.6723 | 0.6969 | 100 | 0.6687 | -0.0876 | -0.1545 | 0.5742 | 0.0669 | -278.1130 | -271.3896 | -2.4595 | -2.4953 |
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
### Framework versions
|
| 70 |
+
|
| 71 |
+
- Transformers 4.44.0.dev0
|
| 72 |
+
- Pytorch 2.1.2
|
| 73 |
+
- Datasets 2.20.0
|
| 74 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9965156794425087,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6818205426622937,
|
| 5 |
+
"train_runtime": 3740.6507,
|
| 6 |
+
"train_samples": 18340,
|
| 7 |
+
"train_samples_per_second": 4.903,
|
| 8 |
+
"train_steps_per_second": 0.038
|
| 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": 0.9965156794425087,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6818205426622937,
|
| 5 |
+
"train_runtime": 3740.6507,
|
| 6 |
+
"train_samples": 18340,
|
| 7 |
+
"train_samples_per_second": 4.903,
|
| 8 |
+
"train_steps_per_second": 0.038
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,284 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9965156794425087,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 143,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.06968641114982578,
|
| 13 |
+
"grad_norm": 8.18808377346908,
|
| 14 |
+
"learning_rate": 3.333333333333333e-07,
|
| 15 |
+
"logits/chosen": -2.7029919624328613,
|
| 16 |
+
"logits/rejected": -2.729884386062622,
|
| 17 |
+
"logps/chosen": -301.84100341796875,
|
| 18 |
+
"logps/rejected": -331.34771728515625,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 21 |
+
"rewards/chosen": 0.00025743356673046947,
|
| 22 |
+
"rewards/margins": 0.00021156822913326323,
|
| 23 |
+
"rewards/rejected": 4.586534487316385e-05,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.13937282229965156,
|
| 28 |
+
"grad_norm": 7.811499447648029,
|
| 29 |
+
"learning_rate": 4.981198836496775e-07,
|
| 30 |
+
"logits/chosen": -2.7489161491394043,
|
| 31 |
+
"logits/rejected": -2.737814426422119,
|
| 32 |
+
"logps/chosen": -320.05413818359375,
|
| 33 |
+
"logps/rejected": -313.6223449707031,
|
| 34 |
+
"loss": 0.6924,
|
| 35 |
+
"rewards/accuracies": 0.5,
|
| 36 |
+
"rewards/chosen": 0.004671993665397167,
|
| 37 |
+
"rewards/margins": 0.0004203595162834972,
|
| 38 |
+
"rewards/rejected": 0.004251633305102587,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.20905923344947736,
|
| 43 |
+
"grad_norm": 9.855807812241265,
|
| 44 |
+
"learning_rate": 4.832481997086846e-07,
|
| 45 |
+
"logits/chosen": -2.6600353717803955,
|
| 46 |
+
"logits/rejected": -2.6442506313323975,
|
| 47 |
+
"logps/chosen": -291.5880126953125,
|
| 48 |
+
"logps/rejected": -304.8880615234375,
|
| 49 |
+
"loss": 0.6907,
|
| 50 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 51 |
+
"rewards/chosen": 0.046477120369672775,
|
| 52 |
+
"rewards/margins": -0.0016271257773041725,
|
| 53 |
+
"rewards/rejected": 0.04810425639152527,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.2787456445993031,
|
| 58 |
+
"grad_norm": 7.883815856532723,
|
| 59 |
+
"learning_rate": 4.543962032878959e-07,
|
| 60 |
+
"logits/chosen": -2.625170946121216,
|
| 61 |
+
"logits/rejected": -2.6397366523742676,
|
| 62 |
+
"logps/chosen": -286.61895751953125,
|
| 63 |
+
"logps/rejected": -294.7152404785156,
|
| 64 |
+
"loss": 0.6849,
|
| 65 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 66 |
+
"rewards/chosen": 0.036045268177986145,
|
| 67 |
+
"rewards/margins": 0.013393419794738293,
|
| 68 |
+
"rewards/rejected": 0.02265184372663498,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.34843205574912894,
|
| 73 |
+
"grad_norm": 8.729035466800674,
|
| 74 |
+
"learning_rate": 4.1329321073844413e-07,
|
| 75 |
+
"logits/chosen": -2.594357967376709,
|
| 76 |
+
"logits/rejected": -2.6037957668304443,
|
| 77 |
+
"logps/chosen": -331.63134765625,
|
| 78 |
+
"logps/rejected": -313.75775146484375,
|
| 79 |
+
"loss": 0.6816,
|
| 80 |
+
"rewards/accuracies": 0.574999988079071,
|
| 81 |
+
"rewards/chosen": -0.029668737202882767,
|
| 82 |
+
"rewards/margins": 0.02038290724158287,
|
| 83 |
+
"rewards/rejected": -0.05005164071917534,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.34843205574912894,
|
| 88 |
+
"eval_logits/chosen": -2.5709125995635986,
|
| 89 |
+
"eval_logits/rejected": -2.5365915298461914,
|
| 90 |
+
"eval_logps/chosen": -263.3412780761719,
|
| 91 |
+
"eval_logps/rejected": -268.15484619140625,
|
| 92 |
+
"eval_loss": 0.672498881816864,
|
| 93 |
+
"eval_rewards/accuracies": 0.58984375,
|
| 94 |
+
"eval_rewards/chosen": -0.0071137575432658195,
|
| 95 |
+
"eval_rewards/margins": 0.04781381040811539,
|
| 96 |
+
"eval_rewards/rejected": -0.05492757260799408,
|
| 97 |
+
"eval_runtime": 101.2981,
|
| 98 |
+
"eval_samples_per_second": 19.744,
|
| 99 |
+
"eval_steps_per_second": 0.316,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.4181184668989547,
|
| 104 |
+
"grad_norm": 9.255478757143024,
|
| 105 |
+
"learning_rate": 3.624028324136517e-07,
|
| 106 |
+
"logits/chosen": -2.6415677070617676,
|
| 107 |
+
"logits/rejected": -2.614701509475708,
|
| 108 |
+
"logps/chosen": -305.6767883300781,
|
| 109 |
+
"logps/rejected": -307.7297668457031,
|
| 110 |
+
"loss": 0.679,
|
| 111 |
+
"rewards/accuracies": 0.5625,
|
| 112 |
+
"rewards/chosen": -0.0592481791973114,
|
| 113 |
+
"rewards/margins": 0.0437784343957901,
|
| 114 |
+
"rewards/rejected": -0.1030266135931015,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.4878048780487805,
|
| 119 |
+
"grad_norm": 9.856902330832456,
|
| 120 |
+
"learning_rate": 3.047753100392174e-07,
|
| 121 |
+
"logits/chosen": -2.598100185394287,
|
| 122 |
+
"logits/rejected": -2.59047532081604,
|
| 123 |
+
"logps/chosen": -330.283203125,
|
| 124 |
+
"logps/rejected": -311.8639831542969,
|
| 125 |
+
"loss": 0.6799,
|
| 126 |
+
"rewards/accuracies": 0.625,
|
| 127 |
+
"rewards/chosen": -0.04586242884397507,
|
| 128 |
+
"rewards/margins": 0.05955269932746887,
|
| 129 |
+
"rewards/rejected": -0.10541512817144394,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.5574912891986062,
|
| 134 |
+
"grad_norm": 10.552750099759244,
|
| 135 |
+
"learning_rate": 2.4386469286927194e-07,
|
| 136 |
+
"logits/chosen": -2.611276149749756,
|
| 137 |
+
"logits/rejected": -2.5819907188415527,
|
| 138 |
+
"logps/chosen": -285.20318603515625,
|
| 139 |
+
"logps/rejected": -287.8341369628906,
|
| 140 |
+
"loss": 0.6733,
|
| 141 |
+
"rewards/accuracies": 0.65625,
|
| 142 |
+
"rewards/chosen": 0.030143652111291885,
|
| 143 |
+
"rewards/margins": 0.07110844552516937,
|
| 144 |
+
"rewards/rejected": -0.04096478223800659,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.627177700348432,
|
| 149 |
+
"grad_norm": 14.503008300206526,
|
| 150 |
+
"learning_rate": 1.8332181063127542e-07,
|
| 151 |
+
"logits/chosen": -2.5344879627227783,
|
| 152 |
+
"logits/rejected": -2.5114903450012207,
|
| 153 |
+
"logps/chosen": -313.04052734375,
|
| 154 |
+
"logps/rejected": -312.25811767578125,
|
| 155 |
+
"loss": 0.6749,
|
| 156 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 157 |
+
"rewards/chosen": -0.07387819141149521,
|
| 158 |
+
"rewards/margins": 0.06775785237550735,
|
| 159 |
+
"rewards/rejected": -0.14163604378700256,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.6968641114982579,
|
| 164 |
+
"grad_norm": 11.233854788459887,
|
| 165 |
+
"learning_rate": 1.26775451942554e-07,
|
| 166 |
+
"logits/chosen": -2.5010740756988525,
|
| 167 |
+
"logits/rejected": -2.4948906898498535,
|
| 168 |
+
"logps/chosen": -304.60186767578125,
|
| 169 |
+
"logps/rejected": -325.60968017578125,
|
| 170 |
+
"loss": 0.6723,
|
| 171 |
+
"rewards/accuracies": 0.512499988079071,
|
| 172 |
+
"rewards/chosen": -0.10407546907663345,
|
| 173 |
+
"rewards/margins": 0.027523279190063477,
|
| 174 |
+
"rewards/rejected": -0.13159874081611633,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.6968641114982579,
|
| 179 |
+
"eval_logits/chosen": -2.495295286178589,
|
| 180 |
+
"eval_logits/rejected": -2.459512948989868,
|
| 181 |
+
"eval_logps/chosen": -271.3895568847656,
|
| 182 |
+
"eval_logps/rejected": -278.11297607421875,
|
| 183 |
+
"eval_loss": 0.6687415242195129,
|
| 184 |
+
"eval_rewards/accuracies": 0.57421875,
|
| 185 |
+
"eval_rewards/chosen": -0.08759630471467972,
|
| 186 |
+
"eval_rewards/margins": 0.0669122040271759,
|
| 187 |
+
"eval_rewards/rejected": -0.15450850129127502,
|
| 188 |
+
"eval_runtime": 102.2188,
|
| 189 |
+
"eval_samples_per_second": 19.566,
|
| 190 |
+
"eval_steps_per_second": 0.313,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.7665505226480837,
|
| 195 |
+
"grad_norm": 11.106233348415344,
|
| 196 |
+
"learning_rate": 7.761486381573326e-08,
|
| 197 |
+
"logits/chosen": -2.547499179840088,
|
| 198 |
+
"logits/rejected": -2.5436906814575195,
|
| 199 |
+
"logps/chosen": -311.0939025878906,
|
| 200 |
+
"logps/rejected": -317.546875,
|
| 201 |
+
"loss": 0.6789,
|
| 202 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 203 |
+
"rewards/chosen": -0.07100532948970795,
|
| 204 |
+
"rewards/margins": 0.048421461135149,
|
| 205 |
+
"rewards/rejected": -0.11942678689956665,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.8362369337979094,
|
| 210 |
+
"grad_norm": 8.671494044278699,
|
| 211 |
+
"learning_rate": 3.878660868757322e-08,
|
| 212 |
+
"logits/chosen": -2.4711318016052246,
|
| 213 |
+
"logits/rejected": -2.4641852378845215,
|
| 214 |
+
"logps/chosen": -250.763916015625,
|
| 215 |
+
"logps/rejected": -268.88031005859375,
|
| 216 |
+
"loss": 0.6795,
|
| 217 |
+
"rewards/accuracies": 0.5,
|
| 218 |
+
"rewards/chosen": -0.021220296621322632,
|
| 219 |
+
"rewards/margins": 0.010579629801213741,
|
| 220 |
+
"rewards/rejected": -0.03179992362856865,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.9059233449477352,
|
| 225 |
+
"grad_norm": 16.493475346448555,
|
| 226 |
+
"learning_rate": 1.261795485174083e-08,
|
| 227 |
+
"logits/chosen": -2.546675205230713,
|
| 228 |
+
"logits/rejected": -2.548250675201416,
|
| 229 |
+
"logps/chosen": -298.49188232421875,
|
| 230 |
+
"logps/rejected": -296.81341552734375,
|
| 231 |
+
"loss": 0.68,
|
| 232 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 233 |
+
"rewards/chosen": 0.026115726679563522,
|
| 234 |
+
"rewards/margins": 0.03052101656794548,
|
| 235 |
+
"rewards/rejected": -0.004405293148010969,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.975609756097561,
|
| 240 |
+
"grad_norm": 10.645354039301385,
|
| 241 |
+
"learning_rate": 6.773858303274482e-10,
|
| 242 |
+
"logits/chosen": -2.5520262718200684,
|
| 243 |
+
"logits/rejected": -2.539773941040039,
|
| 244 |
+
"logps/chosen": -326.08477783203125,
|
| 245 |
+
"logps/rejected": -305.88226318359375,
|
| 246 |
+
"loss": 0.6766,
|
| 247 |
+
"rewards/accuracies": 0.574999988079071,
|
| 248 |
+
"rewards/chosen": -0.00047526060370728374,
|
| 249 |
+
"rewards/margins": 0.04427434504032135,
|
| 250 |
+
"rewards/rejected": -0.044749610126018524,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.9965156794425087,
|
| 255 |
+
"step": 143,
|
| 256 |
+
"total_flos": 0.0,
|
| 257 |
+
"train_loss": 0.6818205426622937,
|
| 258 |
+
"train_runtime": 3740.6507,
|
| 259 |
+
"train_samples_per_second": 4.903,
|
| 260 |
+
"train_steps_per_second": 0.038
|
| 261 |
+
}
|
| 262 |
+
],
|
| 263 |
+
"logging_steps": 10,
|
| 264 |
+
"max_steps": 143,
|
| 265 |
+
"num_input_tokens_seen": 0,
|
| 266 |
+
"num_train_epochs": 1,
|
| 267 |
+
"save_steps": 100,
|
| 268 |
+
"stateful_callbacks": {
|
| 269 |
+
"TrainerControl": {
|
| 270 |
+
"args": {
|
| 271 |
+
"should_epoch_stop": false,
|
| 272 |
+
"should_evaluate": false,
|
| 273 |
+
"should_log": false,
|
| 274 |
+
"should_save": true,
|
| 275 |
+
"should_training_stop": true
|
| 276 |
+
},
|
| 277 |
+
"attributes": {}
|
| 278 |
+
}
|
| 279 |
+
},
|
| 280 |
+
"total_flos": 0.0,
|
| 281 |
+
"train_batch_size": 8,
|
| 282 |
+
"trial_name": null,
|
| 283 |
+
"trial_params": null
|
| 284 |
+
}
|