JY-Ren

Neox-LLM

Using the gpt-neox framework to train llama && llama 2 from 7b to 70b, Supports multiple sample splicing, single or multi-round conversation datasets.

labuladong-fucking-algorithm

Labeling-system-for-RLHF

To train the base model of chatgpt, we initially implemented a labeling system based on the actual requirements of SFT and RLHF.

transpeeder

train llama on a single A100 80G node using 🤗 transformers and 🚀 Deepspeed Pipeline Parallelism

pythia

The hub for EleutherAI's work on interpretability and learning dynamics

leeroy

一个基于pytorch-lightning和huggingface的NLP模型训练框架。

bios_entity_classification

BIOS: An Algorithmically Generated Biomedical Knowledge Graph

bios_ner

BIOS: An Algorithmically Generated Biomedical Knowledge Graph