metadata
base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
datasets:
- GaetanMichelet/chat-60_ft_task-2_auto
- GaetanMichelet/chat-120_ft_task-2_auto
library_name: peft
license: llama3.1
tags:
- alignment-handbook
- trl
- sft
- generated_from_trainer
model-index:
- name: Llama-31-8B_task-2_120-samples_config-2_auto
results: []
Llama-31-8B_task-2_120-samples_config-2_auto
This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-2_auto and the GaetanMichelet/chat-120_ft_task-2_auto datasets. It achieves the following results on the evaluation set:
- Loss: 0.5819
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: 0.0001
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 16
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.9245 | 0.9091 | 5 | 0.8871 |
| 0.8834 | 2.0 | 11 | 0.7984 |
| 0.7224 | 2.9091 | 16 | 0.7230 |
| 0.6496 | 4.0 | 22 | 0.6428 |
| 0.5918 | 4.9091 | 27 | 0.6049 |
| 0.5484 | 6.0 | 33 | 0.5821 |
| 0.472 | 6.9091 | 38 | 0.5819 |
| 0.4328 | 8.0 | 44 | 0.5885 |
| 0.3688 | 8.9091 | 49 | 0.6169 |
| 0.2923 | 10.0 | 55 | 0.6638 |
| 0.2063 | 10.9091 | 60 | 0.7679 |
| 0.1206 | 12.0 | 66 | 0.8585 |
| 0.0643 | 12.9091 | 71 | 1.0241 |
| 0.0373 | 14.0 | 77 | 1.0912 |
Framework versions
- PEFT 0.12.0
- Transformers 4.44.0
- Pytorch 2.1.2+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1