Flash-Muon: An Efficient Implementation of Muon Optimizer

Muon is Scalable for LLM Training

🔬 A Researcher-Friendly Framework for Time Series Analysis. Train Any Model on Any Dataset!

This is the official code and supplementary materials for our AAAI-2024 paper: MASTER: Market-Guided Stock Transformer for Stock Price Forecasting. MASTER is a stock transformer for stock price for…

股票AI操盘手：从学习、模拟到实盘，一站式平台。包含股票知识、策略实例、大模型、因子挖掘、传统策略、机器学习、深度学习、强化学习、图网络、高频交易、C++部署和聚宽实例代码等，可以方便学习、模拟及实盘交易

efinance 是一个可以快速获取基金、股票、债券、期货数据的 Python 库，回测以及量化交易的好帮手！🚀🚀🚀

Code release for DynamicTanh (DyT)

LongRoPE is a novel method that can extends the context window of pre-trained LLMs to an impressive 2048k tokens.

Python library designed to integrate Kolmogorov Arnold Networks with recurrent mechanisms.

An offical implementation of "TimeKAN: KAN-based Frequency Decomposition Learning Architecture for Long-term Time Series Forecasting" (ICLR 2025)

A Comprehensive Survey of Deep Learning for Multivariate Time Series Forecasting: A Channel Strategy Perspective

PyTorch Implementation of FinMamba

Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs

[ICML 2025] Fourier Position Embedding: Enhancing Attention’s Periodic Extension for Length Generalization

PyTorch implementation of FractalGen https://arxiv.org/abs/2502.17437

FlashMLA: Efficient MLA decoding kernels

Muon: An optimizer for hidden layers in neural networks

🚀 Efficient implementations of state-of-the-art linear attention models in Torch and Triton

🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"

MoBA: Mixture of Block Attention for Long-Context LLMs

[ICLR 2024] Official Implementation of "Diffusion-TS: Interpretable Diffusion for General Time Series Generation"

Official PyTorch implementation for "Large Language Diffusion Models"

This is a personal reimplementation of Google's Infini-transformer, utilizing a small 2b model. The project includes both model and training code.

Efficient Infinite Context Transformers with Infini-attention Pytorch Implementation + QwenMoE Implementation + Training Script + 1M context keypass retrieval