GaetanMichelet's picture
End of training
5746cc3 verified
metadata
base_model: google/gemma-2-2b-it
datasets:
  - GaetanMichelet/chat-60_ft_task-2
library_name: peft
license: gemma
tags:
  - alignment-handbook
  - trl
  - sft
  - generated_from_trainer
model-index:
  - name: Gemma-2-2B_task-2_60-samples_config-2_full
    results: []

Gemma-2-2B_task-2_60-samples_config-2_full

This model is a fine-tuned version of google/gemma-2-2b-it on the GaetanMichelet/chat-60_ft_task-2 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.8957

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
1.2929 0.6957 2 1.3116
1.3056 1.7391 5 1.2842
1.2411 2.7826 8 1.1890
1.1401 3.8261 11 1.1172
1.0766 4.8696 14 1.0738
1.024 5.9130 17 1.0270
0.9524 6.9565 20 0.9851
0.9285 8.0 23 0.9539
0.8884 8.6957 25 0.9410
0.8534 9.7391 28 0.9267
0.8501 10.7826 31 0.9164
0.8295 11.8261 34 0.9091
0.7979 12.8696 37 0.9039
0.796 13.9130 40 0.9001
0.7651 14.9565 43 0.8977
0.769 16.0 46 0.8965
0.733 16.6957 48 0.8957
0.743 17.7391 51 0.8961
0.7346 18.7826 54 0.8963
0.7113 19.8261 57 0.8982
0.7027 20.8696 60 0.9012
0.6939 21.9130 63 0.9043
0.6772 22.9565 66 0.9081
0.6706 24.0 69 0.9127

Framework versions

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