metadata
base_model: mistralai/Mistral-Nemo-Instruct-2407
library_name: peft
tags:
- lora
- adapter
lora
This model is a fine-tuned version of mistralai/Mistral-Nemo-Instruct-2407 on the ft_01KSWQ2Z_d0, the ft_01KSWQ2Z_d1, the ft_01KSWQ2Z_d2, the ft_01KSWQ2Z_d3, the ft_01KSWQ2Z_d4 and the ft_01KSWQ2Z_d5 datasets. It achieves the following results on the evaluation set:
- Loss: 0.7760
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: 5e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 2.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.8569 | 0.2903 | 100 | 0.8779 |
| 0.8435 | 0.5806 | 200 | 0.8248 |
| 0.7267 | 0.8708 | 300 | 0.8032 |
| 0.7409 | 1.1597 | 400 | 0.7901 |
| 0.663 | 1.4499 | 500 | 0.7802 |
| 0.7083 | 1.7402 | 600 | 0.7767 |
Framework versions
- PEFT 0.19.1
- Transformers 4.57.1
- Pytorch 2.10.0+rocm7.0
- Datasets 4.0.0
- Tokenizers 0.22.2