❤️Emotional First Aid Dataset, 心理咨询问答、聊天机器人语料库

[EMNLP 2024] 中文领域心理健康对话大模型MeChat

Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels

The official repo for the paper Direct Multi-token Decoding

A unified library of SOTA model optimization techniques like quantization, pruning, distillation, speculative decoding, etc. It compresses deep learning models for downstream deployment frameworks …

🚀 Efficient implementations of state-of-the-art linear attention models

[EMNLP 2025 Main] SpecVLM: Enhancing Speculative Decoding of Video LLMs via Verifier-Guided Token Pruning

MiMo: Unlocking the Reasoning Potential of Language Model – From Pretraining to Posttraining

Official Implementation of "Learning Harmonized Representations for Speculative Sampling" (HASS)

Codes for our paper "Speculative Decoding: Exploiting Speculative Execution for Accelerating Seq2seq Generation" (EMNLP 2023 Findings)

复现大模型相关算法及一些学习记录

Simple and efficient pytorch-native transformer text generation in <1000 LOC of python.

Summary of some awesome work for optimizing LLM inference

[NeurIPS 2025] Scaling Speculative Decoding with Lookahead Reasoning

[ICML 2024] Break the Sequential Dependency of LLM Inference Using Lookahead Decoding

Train speculative decoding models effortlessly and port them smoothly to SGLang serving.

📰 Must-read papers and blogs on Speculative Decoding ⚡️

Official Implementation of EAGLE-1 (ICML'24), EAGLE-2 (EMNLP'24), and EAGLE-3 (NeurIPS'25).

PyTorch Code for Energy-Based Transformers paper -- generalizable reasoning and scalable learning

The best ChatGPT that $100 can buy.

Official Schlably Repository by the Institute for TMDT

PyTorch implementation of paper "Response Length Perception and Sequence Scheduling: An LLM-Empowered LLM Inference Pipeline".

Efficient Interactive LLM Serving with Proxy Model-based Sequence Length Prediction | A tiny BERT model can tell you the verbosity of an LLM (with low latency overhead!)