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-high-bleu
|
| 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-high-bleu
|
| 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: 0.6329
|
| 21 |
+
- Rewards/chosen: -0.2639
|
| 22 |
+
- Rewards/rejected: -0.4887
|
| 23 |
+
- Rewards/accuracies: 0.6875
|
| 24 |
+
- Rewards/margins: 0.2248
|
| 25 |
+
- Logps/rejected: -311.5294
|
| 26 |
+
- Logps/chosen: -289.0153
|
| 27 |
+
- Logits/rejected: -2.2959
|
| 28 |
+
- Logits/chosen: -2.3456
|
| 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 |
+
| 0.66 | 0.3484 | 50 | 0.6513 | -0.1132 | -0.2374 | 0.6641 | 0.1242 | -286.3991 | -273.9513 | -2.4980 | -2.5369 |
|
| 66 |
+
| 0.6376 | 0.6969 | 100 | 0.6329 | -0.2639 | -0.4887 | 0.6875 | 0.2248 | -311.5294 | -289.0153 | -2.2959 | -2.3456 |
|
| 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": 0.6509068312344851,
|
| 5 |
+
"train_runtime": 3880.8302,
|
| 6 |
+
"train_samples": 18340,
|
| 7 |
+
"train_samples_per_second": 4.726,
|
| 8 |
+
"train_steps_per_second": 0.037
|
| 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": 0.6509068312344851,
|
| 5 |
+
"train_runtime": 3880.8302,
|
| 6 |
+
"train_samples": 18340,
|
| 7 |
+
"train_samples_per_second": 4.726,
|
| 8 |
+
"train_steps_per_second": 0.037
|
| 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": 7.974906667324644,
|
| 14 |
+
"learning_rate": 3.333333333333333e-07,
|
| 15 |
+
"logits/chosen": -2.747501850128174,
|
| 16 |
+
"logits/rejected": -2.7370193004608154,
|
| 17 |
+
"logps/chosen": -291.8797302246094,
|
| 18 |
+
"logps/rejected": -281.50189208984375,
|
| 19 |
+
"loss": 0.693,
|
| 20 |
+
"rewards/accuracies": 0.3812499940395355,
|
| 21 |
+
"rewards/chosen": -0.00011452056060079485,
|
| 22 |
+
"rewards/margins": -3.700270826811902e-05,
|
| 23 |
+
"rewards/rejected": -7.751785597065464e-05,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.13937282229965156,
|
| 28 |
+
"grad_norm": 8.1980266894883,
|
| 29 |
+
"learning_rate": 4.981198836496775e-07,
|
| 30 |
+
"logits/chosen": -2.7630066871643066,
|
| 31 |
+
"logits/rejected": -2.748826026916504,
|
| 32 |
+
"logps/chosen": -298.94158935546875,
|
| 33 |
+
"logps/rejected": -320.3427429199219,
|
| 34 |
+
"loss": 0.6902,
|
| 35 |
+
"rewards/accuracies": 0.637499988079071,
|
| 36 |
+
"rewards/chosen": 0.008970660157501698,
|
| 37 |
+
"rewards/margins": 0.006611389108002186,
|
| 38 |
+
"rewards/rejected": 0.002359271515160799,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.20905923344947736,
|
| 43 |
+
"grad_norm": 8.031029605438988,
|
| 44 |
+
"learning_rate": 4.832481997086846e-07,
|
| 45 |
+
"logits/chosen": -2.7137038707733154,
|
| 46 |
+
"logits/rejected": -2.7142727375030518,
|
| 47 |
+
"logps/chosen": -296.2449951171875,
|
| 48 |
+
"logps/rejected": -289.28814697265625,
|
| 49 |
+
"loss": 0.681,
|
| 50 |
+
"rewards/accuracies": 0.6875,
|
| 51 |
+
"rewards/chosen": 0.040586382150650024,
|
| 52 |
+
"rewards/margins": 0.03953887149691582,
|
| 53 |
+
"rewards/rejected": 0.0010475128656253219,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.2787456445993031,
|
| 58 |
+
"grad_norm": 9.014305613217058,
|
| 59 |
+
"learning_rate": 4.543962032878959e-07,
|
| 60 |
+
"logits/chosen": -2.679992198944092,
|
| 61 |
+
"logits/rejected": -2.675081253051758,
|
| 62 |
+
"logps/chosen": -298.83929443359375,
|
| 63 |
+
"logps/rejected": -318.2642822265625,
|
| 64 |
+
"loss": 0.6637,
|
| 65 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 66 |
+
"rewards/chosen": -0.003969565033912659,
|
| 67 |
+
"rewards/margins": 0.06901098042726517,
|
| 68 |
+
"rewards/rejected": -0.07298054546117783,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.34843205574912894,
|
| 73 |
+
"grad_norm": 10.923357206738944,
|
| 74 |
+
"learning_rate": 4.1329321073844413e-07,
|
| 75 |
+
"logits/chosen": -2.547337770462036,
|
| 76 |
+
"logits/rejected": -2.5558876991271973,
|
| 77 |
+
"logps/chosen": -272.10382080078125,
|
| 78 |
+
"logps/rejected": -269.55987548828125,
|
| 79 |
+
"loss": 0.66,
|
| 80 |
+
"rewards/accuracies": 0.59375,
|
| 81 |
+
"rewards/chosen": -0.13082024455070496,
|
| 82 |
+
"rewards/margins": 0.06251106411218643,
|
| 83 |
+
"rewards/rejected": -0.1933313012123108,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.34843205574912894,
|
| 88 |
+
"eval_logits/chosen": -2.536876678466797,
|
| 89 |
+
"eval_logits/rejected": -2.4980242252349854,
|
| 90 |
+
"eval_logps/chosen": -273.9512634277344,
|
| 91 |
+
"eval_logps/rejected": -286.3990783691406,
|
| 92 |
+
"eval_loss": 0.6512594819068909,
|
| 93 |
+
"eval_rewards/accuracies": 0.6640625,
|
| 94 |
+
"eval_rewards/chosen": -0.11321306228637695,
|
| 95 |
+
"eval_rewards/margins": 0.12415696680545807,
|
| 96 |
+
"eval_rewards/rejected": -0.23737002909183502,
|
| 97 |
+
"eval_runtime": 103.5252,
|
| 98 |
+
"eval_samples_per_second": 19.319,
|
| 99 |
+
"eval_steps_per_second": 0.309,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.4181184668989547,
|
| 104 |
+
"grad_norm": 11.450114313746024,
|
| 105 |
+
"learning_rate": 3.624028324136517e-07,
|
| 106 |
+
"logits/chosen": -2.5856661796569824,
|
| 107 |
+
"logits/rejected": -2.5894103050231934,
|
| 108 |
+
"logps/chosen": -315.35540771484375,
|
| 109 |
+
"logps/rejected": -320.2020263671875,
|
| 110 |
+
"loss": 0.6507,
|
| 111 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 112 |
+
"rewards/chosen": -0.12617941200733185,
|
| 113 |
+
"rewards/margins": 0.15071702003479004,
|
| 114 |
+
"rewards/rejected": -0.2768964171409607,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.4878048780487805,
|
| 119 |
+
"grad_norm": 10.762410651532795,
|
| 120 |
+
"learning_rate": 3.047753100392174e-07,
|
| 121 |
+
"logits/chosen": -2.531461238861084,
|
| 122 |
+
"logits/rejected": -2.5228896141052246,
|
| 123 |
+
"logps/chosen": -301.96588134765625,
|
| 124 |
+
"logps/rejected": -300.63177490234375,
|
| 125 |
+
"loss": 0.6458,
|
| 126 |
+
"rewards/accuracies": 0.699999988079071,
|
| 127 |
+
"rewards/chosen": -0.23531746864318848,
|
| 128 |
+
"rewards/margins": 0.1557210385799408,
|
| 129 |
+
"rewards/rejected": -0.3910384774208069,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.5574912891986062,
|
| 134 |
+
"grad_norm": 10.243677934319797,
|
| 135 |
+
"learning_rate": 2.4386469286927194e-07,
|
| 136 |
+
"logits/chosen": -2.531421184539795,
|
| 137 |
+
"logits/rejected": -2.5143864154815674,
|
| 138 |
+
"logps/chosen": -289.48651123046875,
|
| 139 |
+
"logps/rejected": -301.8067932128906,
|
| 140 |
+
"loss": 0.6511,
|
| 141 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 142 |
+
"rewards/chosen": -0.187635600566864,
|
| 143 |
+
"rewards/margins": 0.1388576477766037,
|
| 144 |
+
"rewards/rejected": -0.3264932334423065,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.627177700348432,
|
| 149 |
+
"grad_norm": 10.605767289766804,
|
| 150 |
+
"learning_rate": 1.8332181063127542e-07,
|
| 151 |
+
"logits/chosen": -2.370906114578247,
|
| 152 |
+
"logits/rejected": -2.396063804626465,
|
| 153 |
+
"logps/chosen": -266.50262451171875,
|
| 154 |
+
"logps/rejected": -298.68292236328125,
|
| 155 |
+
"loss": 0.634,
|
| 156 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 157 |
+
"rewards/chosen": -0.13985350728034973,
|
| 158 |
+
"rewards/margins": 0.18151375651359558,
|
| 159 |
+
"rewards/rejected": -0.3213672339916229,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.6968641114982579,
|
| 164 |
+
"grad_norm": 11.921015517224783,
|
| 165 |
+
"learning_rate": 1.26775451942554e-07,
|
| 166 |
+
"logits/chosen": -2.410521984100342,
|
| 167 |
+
"logits/rejected": -2.4090819358825684,
|
| 168 |
+
"logps/chosen": -325.18603515625,
|
| 169 |
+
"logps/rejected": -335.88739013671875,
|
| 170 |
+
"loss": 0.6376,
|
| 171 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 172 |
+
"rewards/chosen": -0.2180793583393097,
|
| 173 |
+
"rewards/margins": 0.21515274047851562,
|
| 174 |
+
"rewards/rejected": -0.4332321286201477,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.6968641114982579,
|
| 179 |
+
"eval_logits/chosen": -2.345639705657959,
|
| 180 |
+
"eval_logits/rejected": -2.2959439754486084,
|
| 181 |
+
"eval_logps/chosen": -289.0152893066406,
|
| 182 |
+
"eval_logps/rejected": -311.52935791015625,
|
| 183 |
+
"eval_loss": 0.6328787803649902,
|
| 184 |
+
"eval_rewards/accuracies": 0.6875,
|
| 185 |
+
"eval_rewards/chosen": -0.26385369896888733,
|
| 186 |
+
"eval_rewards/margins": 0.22481901943683624,
|
| 187 |
+
"eval_rewards/rejected": -0.4886727035045624,
|
| 188 |
+
"eval_runtime": 105.3794,
|
| 189 |
+
"eval_samples_per_second": 18.979,
|
| 190 |
+
"eval_steps_per_second": 0.304,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.7665505226480837,
|
| 195 |
+
"grad_norm": 10.917825147199611,
|
| 196 |
+
"learning_rate": 7.761486381573326e-08,
|
| 197 |
+
"logits/chosen": -2.425245761871338,
|
| 198 |
+
"logits/rejected": -2.4225878715515137,
|
| 199 |
+
"logps/chosen": -340.6777038574219,
|
| 200 |
+
"logps/rejected": -376.40008544921875,
|
| 201 |
+
"loss": 0.6263,
|
| 202 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 203 |
+
"rewards/chosen": -0.29727238416671753,
|
| 204 |
+
"rewards/margins": 0.22700247168540955,
|
| 205 |
+
"rewards/rejected": -0.5242748856544495,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.8362369337979094,
|
| 210 |
+
"grad_norm": 11.561681770035124,
|
| 211 |
+
"learning_rate": 3.878660868757322e-08,
|
| 212 |
+
"logits/chosen": -2.329202175140381,
|
| 213 |
+
"logits/rejected": -2.3225998878479004,
|
| 214 |
+
"logps/chosen": -323.3353271484375,
|
| 215 |
+
"logps/rejected": -341.76959228515625,
|
| 216 |
+
"loss": 0.6324,
|
| 217 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 218 |
+
"rewards/chosen": -0.2884915769100189,
|
| 219 |
+
"rewards/margins": 0.22677147388458252,
|
| 220 |
+
"rewards/rejected": -0.5152631402015686,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.9059233449477352,
|
| 225 |
+
"grad_norm": 11.514501964616796,
|
| 226 |
+
"learning_rate": 1.261795485174083e-08,
|
| 227 |
+
"logits/chosen": -2.403071165084839,
|
| 228 |
+
"logits/rejected": -2.402184009552002,
|
| 229 |
+
"logps/chosen": -327.9507141113281,
|
| 230 |
+
"logps/rejected": -337.8323059082031,
|
| 231 |
+
"loss": 0.6248,
|
| 232 |
+
"rewards/accuracies": 0.6875,
|
| 233 |
+
"rewards/chosen": -0.27428099513053894,
|
| 234 |
+
"rewards/margins": 0.22150692343711853,
|
| 235 |
+
"rewards/rejected": -0.49578791856765747,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.975609756097561,
|
| 240 |
+
"grad_norm": 11.032500717539952,
|
| 241 |
+
"learning_rate": 6.773858303274482e-10,
|
| 242 |
+
"logits/chosen": -2.3575150966644287,
|
| 243 |
+
"logits/rejected": -2.346174716949463,
|
| 244 |
+
"logps/chosen": -298.3825988769531,
|
| 245 |
+
"logps/rejected": -313.13848876953125,
|
| 246 |
+
"loss": 0.6258,
|
| 247 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 248 |
+
"rewards/chosen": -0.3439212739467621,
|
| 249 |
+
"rewards/margins": 0.12675735354423523,
|
| 250 |
+
"rewards/rejected": -0.4706786274909973,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.9965156794425087,
|
| 255 |
+
"step": 143,
|
| 256 |
+
"total_flos": 0.0,
|
| 257 |
+
"train_loss": 0.6509068312344851,
|
| 258 |
+
"train_runtime": 3880.8302,
|
| 259 |
+
"train_samples_per_second": 4.726,
|
| 260 |
+
"train_steps_per_second": 0.037
|
| 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 |
+
}
|