LLM
Minimal, clean code for the Byte Pair Encoding (BPE) algorithm commonly used in LLM tokenization.
tiktoken is a fast BPE tokeniser for use with OpenAI's models.
LlamaIndex is a data framework for your LLM applications
lightweight, standalone C++ inference engine for Google's Gemma models.
The official PyTorch implementation of Google's Gemma models
用于从头预训练+SFT一个小参数量的中文LLaMa2的仓库；24G单卡即可运行得到一个具备简单中文问答能力的chat-llama2.
Modeling, training, eval, and inference code for OLMo
基于向量数据库与GPT3.5的通用本地知识库方案(A universal local knowledge base solution based on vector database and GPT3.5)
a fast cross platform AI inference engine 🤖 using Rust 🦀 and WebGPU 🎮
From scratch implementation of a sparse mixture of experts language model inspired by Andrej Karpathy's makemore :)
Get up and running with Llama 3.2, Mistral, Gemma 2, and other large language models.
An unnecessarily tiny implementation of GPT-2 in NumPy.
The simplest, fastest repository for training/finetuning medium-sized GPTs.
Python package for easily interfacing with chat apps, with robust features and minimal code complexity.
📖A curated list of Awesome LLM Inference Paper with codes, TensorRT-LLM, vLLM, streaming-llm, AWQ, SmoothQuant, WINT8/4, Continuous Batching, FlashAttention, PagedAttention etc.
LLM Finetuning with peft
<Beat AI> 又名 <零生万物> , 是一本专属于软件开发工程师的 AI 入门圣经，手把手带你上手写 AI。从神经网络到大模型，从高层设计到微观原理，从工程实现到算法，学完后，你会发现 AI 也并不是想象中那么高不可攀、无法战胜，Just beat it !
中文nlp解决方案(大模型、数据、模型、训练、推理)
Official repo for "Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models"
DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
Doing simple retrieval from LLM models at various context lengths to measure accuracy
Video+code lecture on building nanoGPT from scratch
DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence