GaetanMichelet's picture
End of training
6d575c1 verified
metadata
base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
datasets:
  - GaetanMichelet/chat-60_ft_task-2
library_name: peft
license: llama3.1
tags:
  - alignment-handbook
  - trl
  - sft
  - generated_from_trainer
model-index:
  - name: Llama-31-8B_task-2_60-samples_config-1_full
    results: []

Llama-31-8B_task-2_60-samples_config-1_full

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-2 dataset. It achieves the following results on the evaluation set:

  • Loss: 1.0622

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: 8
  • total_train_batch_size: 8
  • 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
1.5616 0.8696 5 1.5646
1.5234 1.9130 11 1.4617
1.4186 2.9565 17 1.3542
1.2508 4.0 23 1.2442
1.1191 4.8696 28 1.1498
1.0371 5.9130 34 1.1010
0.9695 6.9565 40 1.0839
0.968 8.0 46 1.0704
0.9532 8.8696 51 1.0646
0.9439 9.9130 57 1.0622
0.8579 10.9565 63 1.0686
0.8117 12.0 69 1.0771
0.8138 12.8696 74 1.0899
0.7059 13.9130 80 1.1196
0.6519 14.9565 86 1.1535
0.65 16.0 92 1.2045
0.405 16.8696 97 1.2544

Framework versions

  • PEFT 0.12.0
  • Transformers 4.44.0
  • Pytorch 2.1.2+cu121
  • Datasets 2.20.0
  • Tokenizers 0.19.1