Model Card for GT-J (Generative Transformer)

Model Details

for manual uses you should prompt model like

<|prompter|>{question}<|endoftext|><|assistant|>

Model Description

GT-J (Generative Transformer) is a fine tuned version of GPT-J on custom dataset that make it available for conversations between model and user

• Developed by: Erfan Zare Chavoshi
• Model type: Generative Model
• Language(s) (NLP): ENG
• License: Apache 2.0
• Finetuned from model: GPT-J

Uses

The easiest way to use this model is to use this with OST-UI from OpenSourceTransformers

Install Dependencies

pip install datasets sentencepiece accelerate bitsandbytes wandb peft tensorboard fjutils transformers[deepspeed] gradio  git+https://github.com/openai/whisper -q  --upgrade

Clone OpenSourceTransformers

git clone https://github.com/erfanzar/OST-OpenSourceTransformers
cd OST-OpenSourceTransformers/

Run Model

python3 OST_UI/app.py --model_id="erfanzar/GT-J"

Note

if you think this card lack information know that ill update this soon