Hugging Face's logo

Holarissun
/
RM-HH-Gemma_helpful_human_20000_gemma2b_shuffleFalse_extractchosenTrue

PEFT
Safetensors
trl
reward-trainer
Generated from Trainer
Model card Files Community
RM-HH-Gemma_helpful_human_20000_gemma2b_shuffleFalse_extractchosenTrue / README.md
Holarissun's picture
Holarissun
End of training
5d5cf0b verified
preview code
|
raw
history blame contribute delete
3.67 kB
metadata
license: gemma
library_name: peft
tags:
  - trl
  - reward-trainer
  - generated_from_trainer
base_model: google/gemma-2b
metrics:
  - accuracy
model-index:
  - name: RM-HH-Gemma_helpful_human_20000_gemma2b_shuffleFalse_extractchosenTrue
    results: []

RM-HH-Gemma_helpful_human_20000_gemma2b_shuffleFalse_extractchosenTrue

This model is a fine-tuned version of google/gemma-2b on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6203
  • Accuracy: 0.6600

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: 1.41e-05
  • train_batch_size: 1
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 4
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 2.0

Training results

Training Loss Epoch Step Validation Loss Accuracy
0.733 0.06 250 0.7295 0.5123
0.7034 0.11 500 0.7061 0.5498
0.7165 0.17 750 0.6913 0.5729
0.6821 0.22 1000 0.6809 0.5884
0.6707 0.28 1250 0.6744 0.6024
0.6551 0.33 1500 0.6692 0.6114
0.6744 0.39 1750 0.6617 0.6144
0.6418 0.45 2000 0.6591 0.6335
0.6546 0.5 2250 0.6567 0.6355
0.6465 0.56 2500 0.6537 0.6375
0.6489 0.61 2750 0.6471 0.6390
0.6555 0.67 3000 0.6399 0.6390
0.647 0.72 3250 0.6374 0.6460
0.6555 0.78 3500 0.6365 0.6490
0.6165 0.83 3750 0.6347 0.6465
0.6385 0.89 4000 0.6338 0.6485
0.6202 0.95 4250 0.6317 0.6490
0.6198 1.0 4500 0.6316 0.6520
0.6092 1.06 4750 0.6325 0.6515
0.6091 1.11 5000 0.6339 0.6510
0.605 1.17 5250 0.6338 0.6540
0.673 1.22 5500 0.6263 0.6550
0.6119 1.28 5750 0.6267 0.6565
0.6153 1.34 6000 0.6267 0.6580
0.6048 1.39 6250 0.6249 0.6560
0.62 1.45 6500 0.6228 0.6540
0.6213 1.5 6750 0.6234 0.6595
0.6107 1.56 7000 0.6228 0.6605
0.6266 1.61 7250 0.6212 0.6580
0.6088 1.67 7500 0.6211 0.6595
0.6282 1.72 7750 0.6210 0.6615
0.6384 1.78 8000 0.6197 0.6610
0.5987 1.84 8250 0.6198 0.6580
0.5911 1.89 8500 0.6201 0.6600
0.5981 1.95 8750 0.6203 0.6600

Framework versions

  • PEFT 0.9.0
  • Transformers 4.38.2
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.15.2