1.8 KiB
1.8 KiB
library_name, license, base_model, tags, model-index
| library_name | license | base_model | tags | model-index | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| transformers | other | meta-llama/Llama-3.2-3B-Instruct |
|
|
Llama3.2-3B_Paper_Impact_SFT
This model is a fine-tuned version of meta-llama/Llama-3.2-3B-Instruct on the paper_impact_sft_train dataset. It achieves the following results on the evaluation set:
- Loss: 0.1446
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: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- total_eval_batch_size: 32
- optimizer: Use adamw_torch 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: 0.1
- num_epochs: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.0607 | 0.7228 | 500 | 0.0733 |
| 0.029 | 1.4452 | 1000 | 0.0819 |
| 0.0058 | 2.1677 | 1500 | 0.1524 |
| 0.005 | 2.8905 | 2000 | 0.1443 |
Framework versions
- Transformers 4.57.1
- Pytorch 2.6.0+cu124
- Datasets 4.0.0
- Tokenizers 0.22.1