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-0.0-1.0-linear-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-0.0-1.0-linear-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: 0.8359
|
| 22 |
+
- Rewards/chosen: 1.9508
|
| 23 |
+
- Rewards/rejected: 1.2167
|
| 24 |
+
- Rewards/accuracies: 0.3333
|
| 25 |
+
- Rewards/margins: 0.7341
|
| 26 |
+
- Logps/rejected: -79.9043
|
| 27 |
+
- Logps/chosen: -72.5286
|
| 28 |
+
- Logits/rejected: -2.5408
|
| 29 |
+
- Logits/chosen: -2.5561
|
| 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: 6.25619875685306e-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: linear
|
| 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 |
+
| 0.6244 | 1.0417 | 100 | 0.8359 | 1.9508 | 1.2167 | 0.3333 | 0.7341 | -79.9043 | -72.5286 | -2.5408 | -2.5561 |
|
| 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": 0.572633953144153,
|
| 5 |
+
"train_runtime": 2040.9175,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 5.99,
|
| 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": 0.572633953144153,
|
| 5 |
+
"train_runtime": 2040.9175,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 5.99,
|
| 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": 654.362555910497,
|
| 14 |
+
"learning_rate": 3.12809937842653e-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": 645.8509361582794,
|
| 29 |
+
"learning_rate": 3.12809937842653e-07,
|
| 30 |
+
"logits/chosen": -2.5562191009521484,
|
| 31 |
+
"logits/rejected": -2.538214683532715,
|
| 32 |
+
"logps/chosen": -87.91580200195312,
|
| 33 |
+
"logps/rejected": -81.03489685058594,
|
| 34 |
+
"loss": 0.7015,
|
| 35 |
+
"rewards/accuracies": 0.2083333283662796,
|
| 36 |
+
"rewards/chosen": -0.0006295897765085101,
|
| 37 |
+
"rewards/margins": -0.004218282178044319,
|
| 38 |
+
"rewards/rejected": 0.0035886918194592,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.20833333333333334,
|
| 43 |
+
"grad_norm": 409.9688749308626,
|
| 44 |
+
"learning_rate": 6.25619875685306e-07,
|
| 45 |
+
"logits/chosen": -2.6103761196136475,
|
| 46 |
+
"logits/rejected": -2.5612125396728516,
|
| 47 |
+
"logps/chosen": -102.897216796875,
|
| 48 |
+
"logps/rejected": -89.5175552368164,
|
| 49 |
+
"loss": 0.6684,
|
| 50 |
+
"rewards/accuracies": 0.34375,
|
| 51 |
+
"rewards/chosen": 0.4510021209716797,
|
| 52 |
+
"rewards/margins": 0.08641182631254196,
|
| 53 |
+
"rewards/rejected": 0.3645903468132019,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.3125,
|
| 58 |
+
"grad_norm": 719.6259537665483,
|
| 59 |
+
"learning_rate": 5.89246627098951e-07,
|
| 60 |
+
"logits/chosen": -2.508330821990967,
|
| 61 |
+
"logits/rejected": -2.5230329036712646,
|
| 62 |
+
"logps/chosen": -67.4363784790039,
|
| 63 |
+
"logps/rejected": -75.49668884277344,
|
| 64 |
+
"loss": 0.7456,
|
| 65 |
+
"rewards/accuracies": 0.34375,
|
| 66 |
+
"rewards/chosen": 0.000880239880643785,
|
| 67 |
+
"rewards/margins": 0.5682423710823059,
|
| 68 |
+
"rewards/rejected": -0.5673621892929077,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.4166666666666667,
|
| 73 |
+
"grad_norm": 505.0085449643877,
|
| 74 |
+
"learning_rate": 5.52873378512596e-07,
|
| 75 |
+
"logits/chosen": -2.5900540351867676,
|
| 76 |
+
"logits/rejected": -2.578765869140625,
|
| 77 |
+
"logps/chosen": -72.26093292236328,
|
| 78 |
+
"logps/rejected": -71.128173828125,
|
| 79 |
+
"loss": 0.7782,
|
| 80 |
+
"rewards/accuracies": 0.25,
|
| 81 |
+
"rewards/chosen": -0.113719142973423,
|
| 82 |
+
"rewards/margins": 0.29866155982017517,
|
| 83 |
+
"rewards/rejected": -0.41238075494766235,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.5208333333333334,
|
| 88 |
+
"grad_norm": 607.5189289574854,
|
| 89 |
+
"learning_rate": 5.16500129926241e-07,
|
| 90 |
+
"logits/chosen": -2.5216305255889893,
|
| 91 |
+
"logits/rejected": -2.533747911453247,
|
| 92 |
+
"logps/chosen": -49.09285354614258,
|
| 93 |
+
"logps/rejected": -57.471824645996094,
|
| 94 |
+
"loss": 0.8133,
|
| 95 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 96 |
+
"rewards/chosen": 1.7973496913909912,
|
| 97 |
+
"rewards/margins": 0.3687282204627991,
|
| 98 |
+
"rewards/rejected": 1.428621530532837,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.625,
|
| 103 |
+
"grad_norm": 445.5346691700546,
|
| 104 |
+
"learning_rate": 4.80126881339886e-07,
|
| 105 |
+
"logits/chosen": -2.6302380561828613,
|
| 106 |
+
"logits/rejected": -2.6127591133117676,
|
| 107 |
+
"logps/chosen": -75.98055267333984,
|
| 108 |
+
"logps/rejected": -76.88851165771484,
|
| 109 |
+
"loss": 0.7692,
|
| 110 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 111 |
+
"rewards/chosen": 3.0699806213378906,
|
| 112 |
+
"rewards/margins": 0.635765016078949,
|
| 113 |
+
"rewards/rejected": 2.434215784072876,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.7291666666666666,
|
| 118 |
+
"grad_norm": 609.003215803196,
|
| 119 |
+
"learning_rate": 4.43753632753531e-07,
|
| 120 |
+
"logits/chosen": -2.626288890838623,
|
| 121 |
+
"logits/rejected": -2.619006633758545,
|
| 122 |
+
"logps/chosen": -93.36027526855469,
|
| 123 |
+
"logps/rejected": -83.28195190429688,
|
| 124 |
+
"loss": 0.7848,
|
| 125 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 126 |
+
"rewards/chosen": 2.9665780067443848,
|
| 127 |
+
"rewards/margins": 0.811475932598114,
|
| 128 |
+
"rewards/rejected": 2.155101776123047,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.8333333333333334,
|
| 133 |
+
"grad_norm": 832.764897592166,
|
| 134 |
+
"learning_rate": 4.07380384167176e-07,
|
| 135 |
+
"logits/chosen": -2.610133647918701,
|
| 136 |
+
"logits/rejected": -2.5608553886413574,
|
| 137 |
+
"logps/chosen": -83.00022888183594,
|
| 138 |
+
"logps/rejected": -77.9175033569336,
|
| 139 |
+
"loss": 0.7948,
|
| 140 |
+
"rewards/accuracies": 0.4375,
|
| 141 |
+
"rewards/chosen": 2.628972053527832,
|
| 142 |
+
"rewards/margins": 1.1728503704071045,
|
| 143 |
+
"rewards/rejected": 1.456121802330017,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.9375,
|
| 148 |
+
"grad_norm": 670.0683208538239,
|
| 149 |
+
"learning_rate": 3.71007135580821e-07,
|
| 150 |
+
"logits/chosen": -2.521639347076416,
|
| 151 |
+
"logits/rejected": -2.5168137550354004,
|
| 152 |
+
"logps/chosen": -52.778839111328125,
|
| 153 |
+
"logps/rejected": -62.82239532470703,
|
| 154 |
+
"loss": 0.8131,
|
| 155 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 156 |
+
"rewards/chosen": 1.1937518119812012,
|
| 157 |
+
"rewards/margins": 0.5518778562545776,
|
| 158 |
+
"rewards/rejected": 0.6418739557266235,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.0416666666666667,
|
| 163 |
+
"grad_norm": 54.34377536034673,
|
| 164 |
+
"learning_rate": 3.3463388699446595e-07,
|
| 165 |
+
"logits/chosen": -2.515855312347412,
|
| 166 |
+
"logits/rejected": -2.4969446659088135,
|
| 167 |
+
"logps/chosen": -68.2948989868164,
|
| 168 |
+
"logps/rejected": -73.43537902832031,
|
| 169 |
+
"loss": 0.6244,
|
| 170 |
+
"rewards/accuracies": 0.38749998807907104,
|
| 171 |
+
"rewards/chosen": 5.195466041564941,
|
| 172 |
+
"rewards/margins": 6.597257137298584,
|
| 173 |
+
"rewards/rejected": -1.401790738105774,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.0416666666666667,
|
| 178 |
+
"eval_logits/chosen": -2.5561327934265137,
|
| 179 |
+
"eval_logits/rejected": -2.540764093399048,
|
| 180 |
+
"eval_logps/chosen": -72.52857208251953,
|
| 181 |
+
"eval_logps/rejected": -79.90425109863281,
|
| 182 |
+
"eval_loss": 0.8358556032180786,
|
| 183 |
+
"eval_rewards/accuracies": 0.3333333432674408,
|
| 184 |
+
"eval_rewards/chosen": 1.9508284330368042,
|
| 185 |
+
"eval_rewards/margins": 0.7341080904006958,
|
| 186 |
+
"eval_rewards/rejected": 1.2167203426361084,
|
| 187 |
+
"eval_runtime": 113.77,
|
| 188 |
+
"eval_samples_per_second": 17.579,
|
| 189 |
+
"eval_steps_per_second": 0.554,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.1458333333333333,
|
| 194 |
+
"grad_norm": 89.01359829903859,
|
| 195 |
+
"learning_rate": 2.9826063840811097e-07,
|
| 196 |
+
"logits/chosen": -2.5290496349334717,
|
| 197 |
+
"logits/rejected": -2.5576484203338623,
|
| 198 |
+
"logps/chosen": -57.967750549316406,
|
| 199 |
+
"logps/rejected": -86.81826782226562,
|
| 200 |
+
"loss": 0.369,
|
| 201 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 202 |
+
"rewards/chosen": 7.356253623962402,
|
| 203 |
+
"rewards/margins": 12.658203125,
|
| 204 |
+
"rewards/rejected": -5.301949501037598,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.25,
|
| 209 |
+
"grad_norm": 36.517350799999726,
|
| 210 |
+
"learning_rate": 2.61887389821756e-07,
|
| 211 |
+
"logits/chosen": -2.585951566696167,
|
| 212 |
+
"logits/rejected": -2.5552000999450684,
|
| 213 |
+
"logps/chosen": -96.2650146484375,
|
| 214 |
+
"logps/rejected": -101.90721130371094,
|
| 215 |
+
"loss": 0.3807,
|
| 216 |
+
"rewards/accuracies": 0.581250011920929,
|
| 217 |
+
"rewards/chosen": 11.669889450073242,
|
| 218 |
+
"rewards/margins": 20.189105987548828,
|
| 219 |
+
"rewards/rejected": -8.519216537475586,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.3541666666666667,
|
| 224 |
+
"grad_norm": 210.1258882273698,
|
| 225 |
+
"learning_rate": 2.2551414123540096e-07,
|
| 226 |
+
"logits/chosen": -2.513514757156372,
|
| 227 |
+
"logits/rejected": -2.5159261226654053,
|
| 228 |
+
"logps/chosen": -78.0460433959961,
|
| 229 |
+
"logps/rejected": -96.77547454833984,
|
| 230 |
+
"loss": 0.3889,
|
| 231 |
+
"rewards/accuracies": 0.46875,
|
| 232 |
+
"rewards/chosen": 8.468559265136719,
|
| 233 |
+
"rewards/margins": 15.862611770629883,
|
| 234 |
+
"rewards/rejected": -7.394052028656006,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.4583333333333333,
|
| 239 |
+
"grad_norm": 328.87414514377355,
|
| 240 |
+
"learning_rate": 1.8914089264904598e-07,
|
| 241 |
+
"logits/chosen": -2.514157772064209,
|
| 242 |
+
"logits/rejected": -2.528646469116211,
|
| 243 |
+
"logps/chosen": -42.35776138305664,
|
| 244 |
+
"logps/rejected": -68.42477416992188,
|
| 245 |
+
"loss": 0.3865,
|
| 246 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 247 |
+
"rewards/chosen": 6.88818883895874,
|
| 248 |
+
"rewards/margins": 10.171409606933594,
|
| 249 |
+
"rewards/rejected": -3.283221483230591,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.5625,
|
| 254 |
+
"grad_norm": 40.158855663364754,
|
| 255 |
+
"learning_rate": 1.52767644062691e-07,
|
| 256 |
+
"logits/chosen": -2.5543441772460938,
|
| 257 |
+
"logits/rejected": -2.5303092002868652,
|
| 258 |
+
"logps/chosen": -63.319725036621094,
|
| 259 |
+
"logps/rejected": -77.66620635986328,
|
| 260 |
+
"loss": 0.3882,
|
| 261 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 262 |
+
"rewards/chosen": 8.774893760681152,
|
| 263 |
+
"rewards/margins": 12.840434074401855,
|
| 264 |
+
"rewards/rejected": -4.065541744232178,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.6666666666666665,
|
| 269 |
+
"grad_norm": 0.8217104876546546,
|
| 270 |
+
"learning_rate": 1.1639439547633599e-07,
|
| 271 |
+
"logits/chosen": -2.5138983726501465,
|
| 272 |
+
"logits/rejected": -2.521850109100342,
|
| 273 |
+
"logps/chosen": -69.83772277832031,
|
| 274 |
+
"logps/rejected": -95.20994567871094,
|
| 275 |
+
"loss": 0.3683,
|
| 276 |
+
"rewards/accuracies": 0.543749988079071,
|
| 277 |
+
"rewards/chosen": 11.020576477050781,
|
| 278 |
+
"rewards/margins": 18.24923324584961,
|
| 279 |
+
"rewards/rejected": -7.228658199310303,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.7708333333333335,
|
| 284 |
+
"grad_norm": 171.24133690438418,
|
| 285 |
+
"learning_rate": 8.0021146889981e-08,
|
| 286 |
+
"logits/chosen": -2.4773125648498535,
|
| 287 |
+
"logits/rejected": -2.4666218757629395,
|
| 288 |
+
"logps/chosen": -66.26407623291016,
|
| 289 |
+
"logps/rejected": -90.37199401855469,
|
| 290 |
+
"loss": 0.3983,
|
| 291 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 292 |
+
"rewards/chosen": 8.90200138092041,
|
| 293 |
+
"rewards/margins": 16.24318504333496,
|
| 294 |
+
"rewards/rejected": -7.341184139251709,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.875,
|
| 299 |
+
"grad_norm": 186.68325753918893,
|
| 300 |
+
"learning_rate": 4.3647898303625996e-08,
|
| 301 |
+
"logits/chosen": -2.5022826194763184,
|
| 302 |
+
"logits/rejected": -2.4847512245178223,
|
| 303 |
+
"logps/chosen": -63.6507453918457,
|
| 304 |
+
"logps/rejected": -80.68909454345703,
|
| 305 |
+
"loss": 0.3745,
|
| 306 |
+
"rewards/accuracies": 0.5,
|
| 307 |
+
"rewards/chosen": 8.155413627624512,
|
| 308 |
+
"rewards/margins": 15.189491271972656,
|
| 309 |
+
"rewards/rejected": -7.034076690673828,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.9791666666666665,
|
| 314 |
+
"grad_norm": 1.1959777156996747,
|
| 315 |
+
"learning_rate": 7.274649717270999e-09,
|
| 316 |
+
"logits/chosen": -2.539518356323242,
|
| 317 |
+
"logits/rejected": -2.5355026721954346,
|
| 318 |
+
"logps/chosen": -77.53422546386719,
|
| 319 |
+
"logps/rejected": -100.30330657958984,
|
| 320 |
+
"loss": 0.373,
|
| 321 |
+
"rewards/accuracies": 0.53125,
|
| 322 |
+
"rewards/chosen": 10.861578941345215,
|
| 323 |
+
"rewards/margins": 20.484987258911133,
|
| 324 |
+
"rewards/rejected": -9.62341022491455,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.0,
|
| 329 |
+
"step": 192,
|
| 330 |
+
"total_flos": 0.0,
|
| 331 |
+
"train_loss": 0.572633953144153,
|
| 332 |
+
"train_runtime": 2040.9175,
|
| 333 |
+
"train_samples_per_second": 5.99,
|
| 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 |
+
}
|