yzhangcs

unsloth

Finetune Llama 3, Mistral & Gemma LLMs 2-5x faster with 80% less memory

VMamba

VMamba: Visual State Space Models，code is based on mamba

math-lm

DeepSeek-MoE

awesome-mixture-of-experts

A collection of AWESOME things about mixture-of-experts

mamba.py

A simple and efficient Mamba implementation in PyTorch and MLX.

ml-aim

This repository provides the code and model checkpoints of the research paper: Scalable Pre-training of Large Autoregressive Image Models

review-2023

二〇二三年的年终总结都写好了吗？

inferflow

Inferflow is an efficient and highly configurable inference engine for large language models (LLMs).

einx

Tensor Operations in Einstein-Inspired Notation for Python.

zero-bubble-pipeline-parallelism

Zero Bubble Pipeline Parallelism

SpeculativeDecodingPapers

📰 Must-read papers and blogs on Speculative Decoding ⚡️

lightning-attention

Lightning Attention-2: A Free Lunch for Handling Unlimited Sequence Lengths in Large Language Models

awesome-ssm-ml

accelerated-scan

Accelerated First Order Parallel Associative Scan

moe_attention

Official repository for the paper "SwitchHead: Accelerating Transformers with Mixture-of-Experts Attention"

CLEX

[ICLR 2024] CLEX: Continuous Length Extrapolation for Large Language Models

triton-autodiff

Experiment of using Tangent to autodiff triton

top_k_attention

The accompanying code for "Memory-efficient Transformers via Top-k Attention" (Ankit Gupta, Guy Dar, Shaya Goodman, David Ciprut, Jonathan Berant. SustaiNLP 2021).

llm-misinformation-survey

Paper list for the survey "Combating Misinformation in the Age of LLMs: Opportunities and Challenges" and the initiative "LLMs Meet Misinformation"

Collinear-Constrained-Attention

seq_icl

mamba-triton

why-weight-decay

Why Do We Need Weight Decay in Modern Deep Learning? [arXiv, Oct 2023]

Highway-Transformer

[ACL‘20] Highway Transformer: A Gated Transformer.

ADM-ES

[ICLR 2024] Official code for the paper 'Elucidating the Exposure Bias in Diffusion Models'

NeurIPS-WANT-submission-efficient-parallelization-layouts

tangent

Source-to-Source Debuggable Derivatives in Pure Python

lecture2

Obsolete version of CUDA-mode repo -- use cuda-mode/lectures instead

Awesome-Simultaneous-Translation

Paper list of simultaneous translation / streaming translation, including text-to-text machine translation and speech-to-text translation.