Creating a new transformer architecture making LLM have controllable reasoning.

Code and data for the paper "Bridging Online and Offline RL: Contextual Bandit Learning for Multi-Turn Code Generation"

DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents

Generative Agents: Interactive Simulacra of Human Behavior

Pocket Flow: 100-line LLM framework. Let Agents build Agents!

This is the code of M-ASK.

This repository includes the official implementation of OpenScholar: Synthesizing Scientific Literature with Retrieval-augmented LMs.

⚔️ [ICLR 2026] Official code of "Search Arena: Analyzing Search-Augmented LLMs".

Probing Scientific General Intelligence of LLMs with Scientist-Aligned Workflows

Official repository for DR Tulu: Reinforcement Learning with Evolving Rubrics for Deep Research

We introduce DreamPRM-1.5, an instance-reweighted framework that adaptively adjusts the importance of each training example via bi-level optimization. We design two complementary strategies: Instan…

The Station, an open-world multi-agent environment that models a miniature scientific ecosystem.

A Systematic Survey of Deep Research

[WWW 2026] 🛠️ DeepAgent: A General Reasoning Agent with Scalable Toolsets

Tutorial on PhD Application

A novel framework named Cognitive-Structured for Relation Extraction (CogRE) that that jointly optimizes task accuracy and explainability.

[ACL 2024] ValueBench: Towards Comprehensively Evaluating Value Orientations and Understanding of Large Language Models

A large-scale, fine-grained, diverse preference dataset (and models).

Codes for paper: Evaluating the Utilities of Large Language Models in Single-cell Data Analysis.

To achieve rapid spatial analysis of large-scale vector data by leveraging the high concurrency of the Go language and the vector analysis capabilities of the GDAL library.

从 0 开始的Java架构资料库

Official repo for 'Large Multimodal Models Evaluation: A Survey'

A version of verl to support diverse tool use

System Level Intelligent Router for Mixture-of-Models at Cloud, Data Center and Edge

基于 Dart 实现的 UI 组件库，支持 Android、iOS、Web、Windows、macOS、Linux 、 HarmonyOS（SDK ≥ 3.29）等多平台编译，持续维护更新

The code of paper "DeFillet: Detection and Removal of Fillet Regions in Polygonal CAD Models" , ACM Transactions on Graphics (SIGGRAPH 2025)

青稞Talk

RxNet 是一款专为 Flutter 开发的跨平台网络请求工具，贴合原生开发习惯，几乎零学习成本即可上手。它不仅让网络通信更丝滑，还支持丰富的功能组合，助你构建高性能、可维护的移动应用，已经支持Android、ios、windows、linux、macos、Web、HarmonyOS

本项目是一个基于 Java 开发的简易工具，主要功能为通过 AI 对代码进行检测分析，并将检测结果自动推送至企业微信，帮助团队快速获取代码质量反馈。支持通过配置与 GitLab 集成，可响应代码相关事件触发检测流程。