metadata
license: apache-2.0
library_name: peft
tags:
- generated_from_trainer
datasets:
- viggo
base_model: mistralai/Mistral-7B-v0.1
model-index:
- name: mixtral-viggo-finetune
results: []
mixtral-viggo-finetune
This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the viggo dataset. It achieves the following results on the evaluation set:
- Loss: 0.1785
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: 2.5e-05
- train_batch_size: 1
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 5
- training_steps: 1000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.6845 | 0.04 | 50 | 0.3694 |
| 0.2931 | 0.08 | 100 | 0.2794 |
| 0.2689 | 0.12 | 150 | 0.2516 |
| 0.2524 | 0.16 | 200 | 0.2398 |
| 0.2226 | 0.2 | 250 | 0.2268 |
| 0.2195 | 0.24 | 300 | 0.2211 |
| 0.2081 | 0.27 | 350 | 0.2141 |
| 0.2052 | 0.31 | 400 | 0.2085 |
| 0.1982 | 0.35 | 450 | 0.2040 |
| 0.2004 | 0.39 | 500 | 0.1974 |
| 0.1889 | 0.43 | 550 | 0.1931 |
| 0.1929 | 0.47 | 600 | 0.1891 |
| 0.185 | 0.51 | 650 | 0.1872 |
| 0.1925 | 0.55 | 700 | 0.1846 |
| 0.1778 | 0.59 | 750 | 0.1885 |
| 0.1741 | 0.63 | 800 | 0.1809 |
| 0.1746 | 0.67 | 850 | 0.1802 |
| 0.1769 | 0.71 | 900 | 0.1792 |
| 0.1795 | 0.74 | 950 | 0.1785 |
| 0.158 | 0.78 | 1000 | 0.1785 |
Framework versions
- PEFT 0.7.2.dev0
- Transformers 4.37.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0