beamaia's picture
Model save
01c7387 verified
|
raw
history blame
2.08 kB
metadata
library_name: peft
tags:
  - trl
  - kto
  - generated_from_trainer
base_model: HuggingFaceH4/zephyr-7b-beta
model-index:
  - name: WeniGPT-Agents-Zephyr-1.0.22-KTO
    results: []

WeniGPT-Agents-Zephyr-1.0.22-KTO

This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-beta on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5
  • Rewards/chosen: -201.5416
  • Rewards/rejected: -167.1877
  • Rewards/margins: -34.3538
  • Kl: 0.0
  • Logps/chosen: -2299.1667
  • Logps/rejected: -1911.5133

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: 0.0002
  • train_batch_size: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.03
  • training_steps: 147
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Rewards/chosen Rewards/rejected Rewards/margins Kl Logps/chosen Logps/rejected
0.7678 0.32 50 0.5000 -200.7396 -166.6349 -34.1048 0.0 -2291.1479 -1905.9844
0.6427 0.65 100 0.5 -201.5416 -167.1877 -34.3538 0.0 -2299.1667 -1911.5133

Framework versions

  • PEFT 0.10.0
  • Transformers 4.38.2
  • Pytorch 2.1.0+cu118
  • Datasets 2.18.0
  • Tokenizers 0.15.2