Commit
•
db2b235
1
Parent(s):
110976e
Update README.md
Browse files
README.md
CHANGED
|
@@ -20,3 +20,43 @@ base_model: unsloth/mistral-7b-instruct-v0.2-bnb-4bit
|
|
| 20 |
This mistral model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
|
| 21 |
|
| 22 |
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
This mistral model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
|
| 21 |
|
| 22 |
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
| 23 |
+
|
| 24 |
+
### Evaluation
|
| 25 |
+
- **ViMMRC test set:** 0.8475 accuracy
|
| 26 |
+
|
| 27 |
+
### Training results
|
| 28 |
+
|
| 29 |
+
| Training Loss | Accuracy | Step | Validation Loss |
|
| 30 |
+
|:-------------:|:---------:|:-----:|:---------------:|
|
| 31 |
+
| 1.033500 | 0.771325 | 240 | 1.478651 |
|
| 32 |
+
| 0.852000 | 0.758621 | 480 | 1.475045 |
|
| 33 |
+
| 0.751200 | 0.751361 | 720 | 1.501176 |
|
| 34 |
+
| 0.668400 | 0.780399 | 960 | 1.543064 |
|
| 35 |
+
| 0.591600 | 0.796733 | 1200 | 1.567212 |
|
| 36 |
+
| 0.498200 | 0.785844 | 1440 | 1.607110 |
|
| 37 |
+
| 0.379600 | 0.796733 | 1680 | 1.643269 |
|
| 38 |
+
| 0.334200 | 0.771324 | 1920 | 1.661141 |
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
### Training hyperparameters
|
| 42 |
+
|
| 43 |
+
The following hyperparameters were used during training:
|
| 44 |
+
- learning_rate: 0.0002
|
| 45 |
+
- train_batch_size: 16
|
| 46 |
+
- eval_batch_size: 8
|
| 47 |
+
- seed: 3407
|
| 48 |
+
- gradient_accumulation_steps: 4
|
| 49 |
+
- eval_accumulation_steps: 4
|
| 50 |
+
- total_train_batch_size: 64
|
| 51 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 52 |
+
- lr_scheduler_type: cosine
|
| 53 |
+
- lr_scheduler_warmup_steps: 5
|
| 54 |
+
- num_epochs: 3
|
| 55 |
+
|
| 56 |
+
### Framework versions
|
| 57 |
+
|
| 58 |
+
- PEFT 0.10.0
|
| 59 |
+
- Transformers 4.40.2
|
| 60 |
+
- Pytorch 2.3.0
|
| 61 |
+
- Datasets 2.19.1
|
| 62 |
+
- Tokenizers 0.19.1
|