metadata
base_model: mistralai/Mistral-7B-Instruct-v0.3
library_name: peft
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: pgd_mistral_8bits_lr4e-05_alpha16_rk8_do0.0_wd1.0e-05
results: []
pgd_mistral_8bits_lr4e-05_alpha16_rk8_do0.0_wd1.0e-05
This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.8732
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 4e-05
- train_batch_size: 3
- eval_batch_size: 3
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 12
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 20
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.9721 | 0.9867 | 37 | 2.1189 |
| 1.525 | 2.0 | 75 | 1.0950 |
| 1.0021 | 2.9867 | 112 | 0.9615 |
| 0.9104 | 4.0 | 150 | 0.9231 |
| 0.9014 | 4.9867 | 187 | 0.8953 |
| 0.8562 | 6.0 | 225 | 0.8828 |
| 0.8702 | 6.9867 | 262 | 0.8774 |
| 0.8411 | 8.0 | 300 | 0.8756 |
| 0.8592 | 8.9867 | 337 | 0.8747 |
| 0.8329 | 9.8667 | 370 | 0.8732 |
Framework versions
- PEFT 0.12.0
- Transformers 4.42.4
- Pytorch 2.4.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1