metadata
base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
datasets:
- GaetanMichelet/chat-60_ft_task-3_auto
library_name: peft
license: llama3.1
tags:
- alignment-handbook
- trl
- sft
- generated_from_trainer
model-index:
- name: Llama-31-8B_task-3_60-samples_config-2_full_auto
results: []
Llama-31-8B_task-3_60-samples_config-2_full_auto
This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-3_auto dataset. It achieves the following results on the evaluation set:
- Loss: 1.1526
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 |
---|---|---|---|
1.6601 | 0.6957 | 2 | 1.6697 |
1.6649 | 1.7391 | 5 | 1.6260 |
1.591 | 2.7826 | 8 | 1.5468 |
1.4992 | 3.8261 | 11 | 1.4664 |
1.4061 | 4.8696 | 14 | 1.3963 |
1.352 | 5.9130 | 17 | 1.3313 |
1.2367 | 6.9565 | 20 | 1.2646 |
1.2127 | 8.0 | 23 | 1.2216 |
1.1571 | 8.6957 | 25 | 1.2052 |
1.1165 | 9.7391 | 28 | 1.1900 |
1.124 | 10.7826 | 31 | 1.1787 |
1.0947 | 11.8261 | 34 | 1.1694 |
1.0606 | 12.8696 | 37 | 1.1634 |
1.0621 | 13.9130 | 40 | 1.1573 |
1.0235 | 14.9565 | 43 | 1.1550 |
1.0274 | 16.0 | 46 | 1.1531 |
0.9827 | 16.6957 | 48 | 1.1526 |
0.9959 | 17.7391 | 51 | 1.1536 |
0.9813 | 18.7826 | 54 | 1.1576 |
0.9571 | 19.8261 | 57 | 1.1600 |
0.9413 | 20.8696 | 60 | 1.1619 |
0.9355 | 21.9130 | 63 | 1.1652 |
0.9063 | 22.9565 | 66 | 1.1698 |
0.8949 | 24.0 | 69 | 1.1736 |
Framework versions
- PEFT 0.12.0
- Transformers 4.44.0
- Pytorch 2.1.2+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1