jiashu-z

LLM101n

LLM101n: Let's build a Storyteller

lectures

Material for cuda-mode lectures

matplotlib-curly-brace

Plot curly brace with matplotlib

Paper-Picture-Writing-Code

MLNLP: Paper Picture Writing Code

JetMoE

Reaching LLaMA2 Performance with 0.1M Dollars

allo

Allo: A Programming Model for Composable Accelerator Design

llama3-from-scratch

llama3 implementation one matrix multiplication at a time

Awesome-LLM-Long-Context-Modeling

📰 Must-read papers and blogs on LLM based Long Context Modeling 🔥

Open-Sora

Open-Sora: Democratizing Efficient Video Production for All

GPUDB-Prefetch

Source code of our DaMoN@SIGMOD 2024 paper "How Does Software Prefetching Work on GPU Query Processing?"

ragas

Evaluation framework for your Retrieval Augmented Generation (RAG) pipelines

DeepSeek-V2

DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model

torchtitan

A native PyTorch Library for large model training

DeepSeek-LLM

DeepSeek LLM: Let there be answers

BurstGPT

A GPT-3.5 & GPT-4 Workload Trace to Optimize LLM Serving Systems

adaptive-retrieval

Vexless

A code base for Vexless

Adaptive-RAG

FastChat

An open platform for training, serving, and evaluating large language models. Release repo for Vicuna and Chatbot Arena.

GEAR

GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM

retro

Official repo to On the Generalization Ability of Retrieval-Enhanced Transformers

llama_index

LlamaIndex is a data framework for your LLM applications

ollama

Get up and running with Llama 3, Mistral, Gemma 2, and other large language models.

self-rag

This includes the original implementation of SELF-RAG: Learning to Retrieve, Generate and Critique through self-reflection by Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi.

contriever

Contriever: Unsupervised Dense Information Retrieval with Contrastive Learning

Verba

Retrieval Augmented Generation (RAG) chatbot powered by Weaviate

RETRO-pytorch

Implementation of RETRO, Deepmind's Retrieval based Attention net, in Pytorch

pytorch-model-train-template

pytorch单精度、半精度、混合精度、单卡、多卡（DP / DDP）、FSDP、DeepSpeed模型训练代码，并对比不同方法的训练速度以及GPU内存的使用

transformer_vq

AdaQP

Adaptive Message Quantization and Parallelization for Distributed Full-graph GNN Training