metadata
license: apache-2.0
library_name: peft
tags:
- generated_from_trainer
base_model: filipealmeida/Mistral-7B-Instruct-v0.1-sharded
model-index:
- name: mistral7b-finetune-10k
results: []
mistral7b-finetune-10k
This model is a fine-tuned version of filipealmeida/Mistral-7B-Instruct-v0.1-sharded on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.0138
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: 0.0002
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- training_steps: 2500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.8342 | 0.08 | 100 | 1.4509 |
| 1.3118 | 0.16 | 200 | 1.2525 |
| 1.2008 | 0.24 | 300 | 1.2086 |
| 1.1544 | 0.33 | 400 | 1.1871 |
| 1.1421 | 0.41 | 500 | 1.1651 |
| 1.1222 | 0.49 | 600 | 1.1497 |
| 1.1234 | 0.57 | 700 | 1.1232 |
| 1.0913 | 0.65 | 800 | 1.1089 |
| 1.0872 | 0.73 | 900 | 1.0906 |
| 1.0396 | 0.82 | 1000 | 1.0784 |
| 1.0634 | 0.9 | 1100 | 1.0701 |
| 1.007 | 0.98 | 1200 | 1.0616 |
| 0.9981 | 1.06 | 1300 | 1.0545 |
| 0.9518 | 1.14 | 1400 | 1.0453 |
| 0.939 | 1.22 | 1500 | 1.0386 |
| 0.9791 | 1.31 | 1600 | 1.0356 |
| 0.977 | 1.39 | 1700 | 1.0302 |
| 0.9287 | 1.47 | 1800 | 1.0233 |
| 0.9393 | 1.55 | 1900 | 1.0209 |
| 0.915 | 1.63 | 2000 | 1.0184 |
| 0.95 | 1.71 | 2100 | 1.0155 |
| 0.9542 | 1.8 | 2200 | 1.0150 |
| 0.9272 | 1.88 | 2300 | 1.0146 |
| 0.9381 | 1.96 | 2400 | 1.0142 |
| 0.9358 | 2.04 | 2500 | 1.0138 |
Framework versions
- PEFT 0.7.1
- Transformers 4.36.0.dev0
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.15.0