update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
tags:
|
| 3 |
+
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- generator
|
| 6 |
+
model-index:
|
| 7 |
+
- name: bert-trainer-8b
|
| 8 |
+
results: []
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 12 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 13 |
+
|
| 14 |
+
# bert-trainer-8b
|
| 15 |
+
|
| 16 |
+
This model is a fine-tuned version of [](https://huggingface.co/) on the generator dataset.
|
| 17 |
+
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 3.1639
|
| 19 |
+
|
| 20 |
+
## Model description
|
| 21 |
+
|
| 22 |
+
More information needed
|
| 23 |
+
|
| 24 |
+
## Intended uses & limitations
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Training and evaluation data
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training procedure
|
| 33 |
+
|
| 34 |
+
### Training hyperparameters
|
| 35 |
+
|
| 36 |
+
The following hyperparameters were used during training:
|
| 37 |
+
- learning_rate: 0.0005
|
| 38 |
+
- train_batch_size: 64
|
| 39 |
+
- eval_batch_size: 64
|
| 40 |
+
- seed: 42
|
| 41 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 42 |
+
- lr_scheduler_type: cosine
|
| 43 |
+
- lr_scheduler_warmup_steps: 1000
|
| 44 |
+
- num_epochs: 32
|
| 45 |
+
- mixed_precision_training: Native AMP
|
| 46 |
+
|
| 47 |
+
### Training results
|
| 48 |
+
|
| 49 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 50 |
+
|:-------------:|:-----:|:-----:|:---------------:|
|
| 51 |
+
| 6.5416 | 1.0 | 500 | 6.5207 |
|
| 52 |
+
| 6.393 | 1.99 | 1000 | 6.3903 |
|
| 53 |
+
| 6.2817 | 2.99 | 1500 | 6.3033 |
|
| 54 |
+
| 6.2274 | 3.98 | 2000 | 6.2671 |
|
| 55 |
+
| 6.179 | 4.98 | 2500 | 6.2431 |
|
| 56 |
+
| 6.1684 | 5.98 | 3000 | 6.2309 |
|
| 57 |
+
| 6.1244 | 6.97 | 3500 | 6.2114 |
|
| 58 |
+
| 6.0879 | 7.97 | 4000 | 6.1932 |
|
| 59 |
+
| 6.0643 | 8.96 | 4500 | 6.1791 |
|
| 60 |
+
| 6.0481 | 9.96 | 5000 | 6.1638 |
|
| 61 |
+
| 6.0231 | 10.96 | 5500 | 6.1581 |
|
| 62 |
+
| 5.9987 | 11.95 | 6000 | 6.1365 |
|
| 63 |
+
| 5.9989 | 12.95 | 6500 | 6.1194 |
|
| 64 |
+
| 5.9535 | 13.94 | 7000 | 6.1095 |
|
| 65 |
+
| 5.9139 | 14.94 | 7500 | 6.0890 |
|
| 66 |
+
| 5.8462 | 15.94 | 8000 | 6.0224 |
|
| 67 |
+
| 5.7689 | 16.93 | 8500 | 5.9266 |
|
| 68 |
+
| 5.6137 | 17.93 | 9000 | 5.7195 |
|
| 69 |
+
| 4.7163 | 18.92 | 9500 | 4.6131 |
|
| 70 |
+
| 4.0877 | 19.92 | 10000 | 4.0903 |
|
| 71 |
+
| 3.7832 | 20.92 | 10500 | 3.8340 |
|
| 72 |
+
| 3.6104 | 21.91 | 11000 | 3.6572 |
|
| 73 |
+
| 3.4615 | 22.91 | 11500 | 3.5278 |
|
| 74 |
+
| 3.3661 | 23.9 | 12000 | 3.4201 |
|
| 75 |
+
| 3.271 | 24.9 | 12500 | 3.3333 |
|
| 76 |
+
| 3.2179 | 25.9 | 13000 | 3.2720 |
|
| 77 |
+
| 3.1759 | 26.89 | 13500 | 3.2317 |
|
| 78 |
+
| 3.1419 | 27.89 | 14000 | 3.2006 |
|
| 79 |
+
| 3.1041 | 28.88 | 14500 | 3.1806 |
|
| 80 |
+
| 3.0836 | 29.88 | 15000 | 3.1693 |
|
| 81 |
+
| 3.0998 | 30.88 | 15500 | 3.1679 |
|
| 82 |
+
| 3.08 | 31.87 | 16000 | 3.1639 |
|
| 83 |
+
|
| 84 |
+
|
| 85 |
+
### Framework versions
|
| 86 |
+
|
| 87 |
+
- Transformers 4.26.1
|
| 88 |
+
- Pytorch 1.13.1
|
| 89 |
+
- Datasets 2.9.0
|
| 90 |
+
- Tokenizers 0.13.2
|