Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
model-index:
|
| 7 |
+
- name: mistral-alpaca2k-3e
|
| 8 |
+
results: []
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
|
| 12 |
+
# mistral-alpaca2k-3e
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the mhenrichsen/alpaca_2k_test dataset.
|
| 15 |
+
It achieves the following results on the evaluation set:
|
| 16 |
+
- Loss: 0.8850
|
| 17 |
+
|
| 18 |
+
## Training procedure
|
| 19 |
+
accelerate launch -m axolotl.cli.train examples/mistral/qlora.yml
|
| 20 |
+
|
| 21 |
+
### Training hyperparameters
|
| 22 |
+
|
| 23 |
+
The following hyperparameters were used during training:
|
| 24 |
+
- learning_rate: 0.0002
|
| 25 |
+
- train_batch_size: 2
|
| 26 |
+
- eval_batch_size: 2
|
| 27 |
+
- seed: 42
|
| 28 |
+
- gradient_accumulation_steps: 4
|
| 29 |
+
- total_train_batch_size: 8
|
| 30 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 31 |
+
- lr_scheduler_type: cosine
|
| 32 |
+
- lr_scheduler_warmup_steps: 10
|
| 33 |
+
- num_epochs: 3
|
| 34 |
+
|
| 35 |
+
### Training results
|
| 36 |
+
|
| 37 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 38 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
| 39 |
+
| 1.392 | 0.0 | 1 | 1.2581 |
|
| 40 |
+
| 0.912 | 0.15 | 36 | 0.7686 |
|
| 41 |
+
| 0.7114 | 0.3 | 72 | 0.7590 |
|
| 42 |
+
| 0.7849 | 0.45 | 108 | 0.7561 |
|
| 43 |
+
| 0.693 | 0.61 | 144 | 0.7546 |
|
| 44 |
+
| 0.686 | 0.76 | 180 | 0.7538 |
|
| 45 |
+
| 0.782 | 0.91 | 216 | 0.7524 |
|
| 46 |
+
| 0.5691 | 1.06 | 252 | 0.7700 |
|
| 47 |
+
| 0.5295 | 1.21 | 288 | 0.7883 |
|
| 48 |
+
| 0.5313 | 1.36 | 324 | 0.7876 |
|
| 49 |
+
| 0.4994 | 1.52 | 360 | 0.7971 |
|
| 50 |
+
| 0.6007 | 1.67 | 396 | 0.7881 |
|
| 51 |
+
| 0.5459 | 1.82 | 432 | 0.7911 |
|
| 52 |
+
| 0.5194 | 1.97 | 468 | 0.7924 |
|
| 53 |
+
| 0.3376 | 2.12 | 504 | 0.8711 |
|
| 54 |
+
| 0.2983 | 2.27 | 540 | 0.8916 |
|
| 55 |
+
| 0.341 | 2.43 | 576 | 0.8891 |
|
| 56 |
+
| 0.2961 | 2.58 | 612 | 0.8861 |
|
| 57 |
+
| 0.2469 | 2.73 | 648 | 0.8860 |
|
| 58 |
+
| 0.3535 | 2.88 | 684 | 0.8850 |
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
### Framework versions
|
| 62 |
+
|
| 63 |
+
- Transformers 4.35.2
|
| 64 |
+
- Pytorch 2.0.1+cu118
|
| 65 |
+
- Datasets 2.15.0
|
| 66 |
+
- Tokenizers 0.15.0
|