GaetanMichelet's picture
End of training
dc3f0f2 verified
metadata
base_model: mistralai/Mistral-7B-Instruct-v0.3
datasets:
  - GaetanMichelet/chat-60_ft_task-2
library_name: peft
license: apache-2.0
tags:
  - alignment-handbook
  - trl
  - sft
  - generated_from_trainer
model-index:
  - name: Mistral-7B_task-2_60-samples_config-1
    results: []

Mistral-7B_task-2_60-samples_config-1

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the GaetanMichelet/chat-60_ft_task-2 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5375

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
0.9616 0.8696 5 0.8768
0.7463 1.9130 11 0.6409
0.6311 2.9565 17 0.5873
0.5306 4.0 23 0.5467
0.4454 4.8696 28 0.5375
0.3475 5.9130 34 0.5507
0.2291 6.9565 40 0.6329
0.1245 8.0 46 0.7403
0.0548 8.8696 51 0.8630
0.0302 9.9130 57 0.8626
0.014 10.9565 63 0.9664
0.0152 12.0 69 0.9425

Framework versions

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