| license: apache-2.0 | |
| library_name: peft | |
| tags: | |
| - trl | |
| - sft | |
| - generated_from_trainer | |
| datasets: | |
| - generator | |
| base_model: mistralai/Mistral-7B-Instruct-v0.2 | |
| pipeline_tag: text-generation | |
| model-index: | |
| - name: mistralFT | |
| results: [] | |
| <!-- This model card has been generated automatically according to the information the Trainer had access to. You | |
| should probably proofread and complete it, then remove this comment. --> | |
| # mistralFT | |
| This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the generator dataset. | |
| ## 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: 4 | |
| - eval_batch_size: 8 | |
| - seed: 42 | |
| - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 | |
| - lr_scheduler_type: constant | |
| - lr_scheduler_warmup_steps: 0.03 | |
| - training_steps: 10 | |
| ### Training results | |
| ### Framework versions | |
| - PEFT 0.10.0 | |
| - Transformers 4.39.3 | |
| - Pytorch 2.2.2+cu121 | |
| - Datasets 2.18.0 | |
| - Tokenizers 0.15.2 |