latest / README.md
Shubhang999's picture
End of training
d606f90
metadata
license: mit
base_model: TheBloke/zephyr-7B-alpha-GPTQ
tags:
  - generated_from_trainer
model-index:
  - name: latest
    results: []

latest

This model is a fine-tuned version of TheBloke/zephyr-7B-alpha-GPTQ on the None dataset.

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • training_steps: 50
  • mixed_precision_training: Native AMP

Training results

Framework versions

  • Transformers 4.35.2
  • Pytorch 2.1.0+cu118
  • Datasets 2.15.0
  • Tokenizers 0.15.0