Tensor-core CUDA kernels for Nyström attention, linear-time forward and backward with exact autograd gradients. Faster than flash-attention at long sequence length.

Semantic-ID Audit — 生成式检索语义ID冲突诊断

PCTD: Preference-Guided Counterfactual Task Decomposition for Agent Tool Retrieval

这是一个由LangGraph协议主导的因果分析Muti-Agent，结合MCP,RAG等多种工具进行辅助进行因果分析，提供给用户一份完善的因果分析的分析报告和因果图

Official code for the Manning book on structural LLM optimization: depth/width pruning, knowledge distillation, and attention optimization, runnable on free Colab GPUs.

O(N·K) multi-head attention for PyTorch — a sparse drop-in replacement for dense scaled-dot-product attention.

[KDD 2026] LocalSearchBench: Benchmarking Agentic Search in Real-World Local Life Services

「人工智能导论」大作业——超市AI营销系统

[Pytorch] Efficient tokenization for recommendations and generative retrieval. Inspired by STATIC decoding from "Vectorizing the Trie"

Learning-to-rank scheduling for Python job queues. Replaces FIFO with an online-learning LambdaRank ranker that predicts job duration from telemetry and reorders pending work shortest-job-first. Pl…

offline agentic workflow for generating ad creatives

A Collection of BM25 Algorithms in Python

The code of Passing the Turing Test on Screen

OpenClaw-RL: Train any agent simply by talking

[参赛队伍] TAAC 2026 腾讯广告算法大赛 X KDD 2026

机器学习工程师、算法工程师、软件工程师、数据科学家-面试指南 | Interview guide for MLE, SDE, DS

Claude Code Snapshot for Research. All original source code is the property of Anthropic.

Bayesian optimization in PyTorch

Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels.

A unified, efficient, and extensible PyTorch-based recommendation library

Local-first RAG framework for maximizing the value of private and on-device LLM systems.

rl-explainer

从零构建大模型：从预训练到RLHF的完整实践