This Repo Contains the pre-trained Models using EasyDel Library

What is EasyDel

EasyDeL (Easy Deep Learning) is an open-source library designed to accelerate and optimize the training process of machine learning models. This library is primarily focused on Jax/Flax and plans to offer easy and fine solutions to train Flax/Jax Models on the TPU/GPU

Availalbe on PyPi

To install EasyDeL, you can use pip:

pip install easydel

PyGit

for the latest version (beta)

pip install git+https://github.com/erfanzar/EasyDel

Available Models Are

• Llama (Support FSDP, MP, DP)(Supports gradient checkpointing)

• GPT-J (Support FSDP, MP, DP)(Supports gradient checkpointing)

• LT (Support FSDP, MP, DP)(Supports gradient checkpointing)

• MosaicMPT (Support FSDP, MP, DP)(Supports gradient checkpointing)

• GPTNeoX (Support FSDP, MP, DP)(Supports gradient checkpointing)

• Falcon (Support FSDP, MP, DP)(Supports gradient checkpointing)

• Palm (Support FSDP, MP, DP)(Supports gradient checkpointing)

• T5 (Support FSDP, MP, DP)(Supports gradient checkpointing)

• LLama GPT-J MosaicMPT Falcon supports Flash Attention

you can also tell me the model you want in Flax/Jax version and ill try my best to build it ;)

Tutorials

Tutorials on how to use and train or serve your models with EasyDel are available at examples dir

EasyDel Serving

Easydel also supports serving and using LLMs with less than 50 code with using JAXServer in EasyDel with gradio and FastAPI cool features and its also hackable and opensource to use the library just how you want it