Commit
•
7bbb34a
1
Parent(s):
724ba1b
Model save
Browse files- README.md +76 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +269 -0
README.md
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- sft
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
datasets:
|
| 10 |
+
- generator
|
| 11 |
+
model-index:
|
| 12 |
+
- name: zephyr-7b-math-train-test
|
| 13 |
+
results: []
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 17 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 18 |
+
|
| 19 |
+
# zephyr-7b-math-train-test
|
| 20 |
+
|
| 21 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the generator dataset.
|
| 22 |
+
It achieves the following results on the evaluation set:
|
| 23 |
+
- Loss: 0.0130
|
| 24 |
+
|
| 25 |
+
## Model description
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Intended uses & limitations
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Training and evaluation data
|
| 34 |
+
|
| 35 |
+
More information needed
|
| 36 |
+
|
| 37 |
+
## Training procedure
|
| 38 |
+
|
| 39 |
+
### Training hyperparameters
|
| 40 |
+
|
| 41 |
+
The following hyperparameters were used during training:
|
| 42 |
+
- learning_rate: 1e-05
|
| 43 |
+
- train_batch_size: 16
|
| 44 |
+
- eval_batch_size: 8
|
| 45 |
+
- seed: 42
|
| 46 |
+
- distributed_type: multi-GPU
|
| 47 |
+
- num_devices: 4
|
| 48 |
+
- total_train_batch_size: 64
|
| 49 |
+
- total_eval_batch_size: 32
|
| 50 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 51 |
+
- lr_scheduler_type: cosine
|
| 52 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 53 |
+
- num_epochs: 10
|
| 54 |
+
|
| 55 |
+
### Training results
|
| 56 |
+
|
| 57 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 58 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
| 59 |
+
| 0.746 | 1.0 | 10 | 0.6231 |
|
| 60 |
+
| 0.5258 | 2.0 | 20 | 0.3843 |
|
| 61 |
+
| 0.3183 | 3.0 | 30 | 0.1999 |
|
| 62 |
+
| 0.16 | 4.0 | 40 | 0.0864 |
|
| 63 |
+
| 0.0811 | 5.0 | 50 | 0.0496 |
|
| 64 |
+
| 0.0502 | 6.0 | 60 | 0.0345 |
|
| 65 |
+
| 0.035 | 7.0 | 70 | 0.0254 |
|
| 66 |
+
| 0.0241 | 8.0 | 80 | 0.0185 |
|
| 67 |
+
| 0.0165 | 9.0 | 90 | 0.0142 |
|
| 68 |
+
| 0.0129 | 10.0 | 100 | 0.0130 |
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
### Framework versions
|
| 72 |
+
|
| 73 |
+
- Transformers 4.44.2
|
| 74 |
+
- Pytorch 2.4.1+cu124
|
| 75 |
+
- Datasets 2.21.0
|
| 76 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 20937965568000.0,
|
| 4 |
+
"train_loss": 0.21048430703580379,
|
| 5 |
+
"train_runtime": 572.8088,
|
| 6 |
+
"train_samples": 4000,
|
| 7 |
+
"train_samples_per_second": 10.737,
|
| 8 |
+
"train_steps_per_second": 0.175
|
| 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": 10.0,
|
| 3 |
+
"total_flos": 20937965568000.0,
|
| 4 |
+
"train_loss": 0.21048430703580379,
|
| 5 |
+
"train_runtime": 572.8088,
|
| 6 |
+
"train_samples": 4000,
|
| 7 |
+
"train_samples_per_second": 10.737,
|
| 8 |
+
"train_steps_per_second": 0.175
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,269 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 100,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.1,
|
| 13 |
+
"grad_norm": 18.916099095189473,
|
| 14 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 15 |
+
"loss": 0.9574,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.5,
|
| 20 |
+
"grad_norm": 21.53952313643694,
|
| 21 |
+
"learning_rate": 5e-06,
|
| 22 |
+
"loss": 0.8754,
|
| 23 |
+
"step": 5
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 1.0,
|
| 27 |
+
"grad_norm": 5.798825182601948,
|
| 28 |
+
"learning_rate": 1e-05,
|
| 29 |
+
"loss": 0.746,
|
| 30 |
+
"step": 10
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 1.0,
|
| 34 |
+
"eval_loss": 0.6230876445770264,
|
| 35 |
+
"eval_runtime": 5.7302,
|
| 36 |
+
"eval_samples_per_second": 53.052,
|
| 37 |
+
"eval_steps_per_second": 1.745,
|
| 38 |
+
"step": 10
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.5,
|
| 42 |
+
"grad_norm": 3.2446769999187137,
|
| 43 |
+
"learning_rate": 9.924038765061042e-06,
|
| 44 |
+
"loss": 0.5866,
|
| 45 |
+
"step": 15
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 2.0,
|
| 49 |
+
"grad_norm": 2.9248994732336526,
|
| 50 |
+
"learning_rate": 9.698463103929542e-06,
|
| 51 |
+
"loss": 0.5258,
|
| 52 |
+
"step": 20
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 2.0,
|
| 56 |
+
"eval_loss": 0.3843041658401489,
|
| 57 |
+
"eval_runtime": 5.7293,
|
| 58 |
+
"eval_samples_per_second": 53.061,
|
| 59 |
+
"eval_steps_per_second": 1.745,
|
| 60 |
+
"step": 20
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 2.5,
|
| 64 |
+
"grad_norm": 3.044511051436622,
|
| 65 |
+
"learning_rate": 9.330127018922195e-06,
|
| 66 |
+
"loss": 0.3601,
|
| 67 |
+
"step": 25
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 3.0,
|
| 71 |
+
"grad_norm": 2.817325596304153,
|
| 72 |
+
"learning_rate": 8.83022221559489e-06,
|
| 73 |
+
"loss": 0.3183,
|
| 74 |
+
"step": 30
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 3.0,
|
| 78 |
+
"eval_loss": 0.1999027132987976,
|
| 79 |
+
"eval_runtime": 5.7226,
|
| 80 |
+
"eval_samples_per_second": 53.123,
|
| 81 |
+
"eval_steps_per_second": 1.747,
|
| 82 |
+
"step": 30
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 3.5,
|
| 86 |
+
"grad_norm": 3.4431057415284267,
|
| 87 |
+
"learning_rate": 8.213938048432697e-06,
|
| 88 |
+
"loss": 0.1832,
|
| 89 |
+
"step": 35
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 4.0,
|
| 93 |
+
"grad_norm": 2.7772199860436415,
|
| 94 |
+
"learning_rate": 7.500000000000001e-06,
|
| 95 |
+
"loss": 0.16,
|
| 96 |
+
"step": 40
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 4.0,
|
| 100 |
+
"eval_loss": 0.0864211916923523,
|
| 101 |
+
"eval_runtime": 5.7261,
|
| 102 |
+
"eval_samples_per_second": 53.09,
|
| 103 |
+
"eval_steps_per_second": 1.746,
|
| 104 |
+
"step": 40
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 4.5,
|
| 108 |
+
"grad_norm": 3.315697037143245,
|
| 109 |
+
"learning_rate": 6.710100716628345e-06,
|
| 110 |
+
"loss": 0.0802,
|
| 111 |
+
"step": 45
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 5.0,
|
| 115 |
+
"grad_norm": 2.0887623280319887,
|
| 116 |
+
"learning_rate": 5.8682408883346535e-06,
|
| 117 |
+
"loss": 0.0811,
|
| 118 |
+
"step": 50
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 5.0,
|
| 122 |
+
"eval_loss": 0.04959693178534508,
|
| 123 |
+
"eval_runtime": 5.7295,
|
| 124 |
+
"eval_samples_per_second": 53.059,
|
| 125 |
+
"eval_steps_per_second": 1.745,
|
| 126 |
+
"step": 50
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 5.5,
|
| 130 |
+
"grad_norm": 2.1432132959969725,
|
| 131 |
+
"learning_rate": 5e-06,
|
| 132 |
+
"loss": 0.0479,
|
| 133 |
+
"step": 55
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 6.0,
|
| 137 |
+
"grad_norm": 1.8407217223386276,
|
| 138 |
+
"learning_rate": 4.131759111665349e-06,
|
| 139 |
+
"loss": 0.0502,
|
| 140 |
+
"step": 60
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 6.0,
|
| 144 |
+
"eval_loss": 0.034496039152145386,
|
| 145 |
+
"eval_runtime": 5.726,
|
| 146 |
+
"eval_samples_per_second": 53.091,
|
| 147 |
+
"eval_steps_per_second": 1.746,
|
| 148 |
+
"step": 60
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 6.5,
|
| 152 |
+
"grad_norm": 1.5928033449038717,
|
| 153 |
+
"learning_rate": 3.289899283371657e-06,
|
| 154 |
+
"loss": 0.0339,
|
| 155 |
+
"step": 65
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 7.0,
|
| 159 |
+
"grad_norm": 1.359878925254289,
|
| 160 |
+
"learning_rate": 2.5000000000000015e-06,
|
| 161 |
+
"loss": 0.035,
|
| 162 |
+
"step": 70
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 7.0,
|
| 166 |
+
"eval_loss": 0.025400497019290924,
|
| 167 |
+
"eval_runtime": 5.7281,
|
| 168 |
+
"eval_samples_per_second": 53.072,
|
| 169 |
+
"eval_steps_per_second": 1.746,
|
| 170 |
+
"step": 70
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 7.5,
|
| 174 |
+
"grad_norm": 1.1463040760182688,
|
| 175 |
+
"learning_rate": 1.7860619515673034e-06,
|
| 176 |
+
"loss": 0.0248,
|
| 177 |
+
"step": 75
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 8.0,
|
| 181 |
+
"grad_norm": 1.0125936690851214,
|
| 182 |
+
"learning_rate": 1.1697777844051105e-06,
|
| 183 |
+
"loss": 0.0241,
|
| 184 |
+
"step": 80
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 8.0,
|
| 188 |
+
"eval_loss": 0.0185158122330904,
|
| 189 |
+
"eval_runtime": 5.7304,
|
| 190 |
+
"eval_samples_per_second": 53.051,
|
| 191 |
+
"eval_steps_per_second": 1.745,
|
| 192 |
+
"step": 80
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 8.5,
|
| 196 |
+
"grad_norm": 0.8554989663331637,
|
| 197 |
+
"learning_rate": 6.698729810778065e-07,
|
| 198 |
+
"loss": 0.0177,
|
| 199 |
+
"step": 85
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 9.0,
|
| 203 |
+
"grad_norm": 0.6771852125259162,
|
| 204 |
+
"learning_rate": 3.015368960704584e-07,
|
| 205 |
+
"loss": 0.0165,
|
| 206 |
+
"step": 90
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 9.0,
|
| 210 |
+
"eval_loss": 0.014229783788323402,
|
| 211 |
+
"eval_runtime": 5.7297,
|
| 212 |
+
"eval_samples_per_second": 53.057,
|
| 213 |
+
"eval_steps_per_second": 1.745,
|
| 214 |
+
"step": 90
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 9.5,
|
| 218 |
+
"grad_norm": 0.5405112996842772,
|
| 219 |
+
"learning_rate": 7.59612349389599e-08,
|
| 220 |
+
"loss": 0.0135,
|
| 221 |
+
"step": 95
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 10.0,
|
| 225 |
+
"grad_norm": 0.5192308458018222,
|
| 226 |
+
"learning_rate": 0.0,
|
| 227 |
+
"loss": 0.0129,
|
| 228 |
+
"step": 100
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 10.0,
|
| 232 |
+
"eval_loss": 0.01297028735280037,
|
| 233 |
+
"eval_runtime": 5.7028,
|
| 234 |
+
"eval_samples_per_second": 53.307,
|
| 235 |
+
"eval_steps_per_second": 1.754,
|
| 236 |
+
"step": 100
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 10.0,
|
| 240 |
+
"step": 100,
|
| 241 |
+
"total_flos": 20937965568000.0,
|
| 242 |
+
"train_loss": 0.21048430703580379,
|
| 243 |
+
"train_runtime": 572.8088,
|
| 244 |
+
"train_samples_per_second": 10.737,
|
| 245 |
+
"train_steps_per_second": 0.175
|
| 246 |
+
}
|
| 247 |
+
],
|
| 248 |
+
"logging_steps": 5,
|
| 249 |
+
"max_steps": 100,
|
| 250 |
+
"num_input_tokens_seen": 0,
|
| 251 |
+
"num_train_epochs": 10,
|
| 252 |
+
"save_steps": 100,
|
| 253 |
+
"stateful_callbacks": {
|
| 254 |
+
"TrainerControl": {
|
| 255 |
+
"args": {
|
| 256 |
+
"should_epoch_stop": false,
|
| 257 |
+
"should_evaluate": false,
|
| 258 |
+
"should_log": false,
|
| 259 |
+
"should_save": true,
|
| 260 |
+
"should_training_stop": true
|
| 261 |
+
},
|
| 262 |
+
"attributes": {}
|
| 263 |
+
}
|
| 264 |
+
},
|
| 265 |
+
"total_flos": 20937965568000.0,
|
| 266 |
+
"train_batch_size": 16,
|
| 267 |
+
"trial_name": null,
|
| 268 |
+
"trial_params": null
|
| 269 |
+
}
|