eelxpeng's picture
Create README.md
9eb8bc4 verified
|
raw
history blame
410 Bytes
metadata
license: apache-2.0
datasets:
  - HuggingFaceH4/ultrafeedback_binarized
language:
  - en

Introduction

This model vistagi/Mixtral-8x7b-v0.1-sft is trained with Ultrachat-200K dataset through supervised finetuning using Mixtral-8x7b-v0.1 as the baseline model. The training is done with bfloat16 precision using LoRA.

Details

Used Librarys

  • torch
  • deepspeed
  • pytorch lightning
  • transformers
  • peft