Starred repositories
An Agent Skill for Generative Engine Optimization (GEO): audit AI search visibility across markets and get evidence-backed recommendations from a URL.
Multi‑agent semantic search system for an e‑commerce dataset, featuring embeddings, Qdrant Vector database, retrieval orchestration, evaluation metrics, and a production‑ready REST API.
LongHarness evaluates how language-model harnesses access and reason over long contexts. Its four task suites contain 200 instances that require selective retrieval, verification, and multi-step re…
Agentic Solution For E-commerce Search Evaluation
Data and code for ACL 2026 Paper "Rethinking Reasoning-Intensive Retrieval: Evaluating and Advancing Retrievers in Agentic Search Systems"
Does a TypeSafe Jev rerank beat embedding search? Graded relevance eval (9,831 pairs, 164 zh/en queries) over the Agent Skills Hub catalog, with the judge-circularity bias measured.
Public artifact for SearchGEO: LLM search-agent endorsement vulnerability evaluation
Local-first Agent RL workbench for Search-R1 rollouts, reward evaluation, GRPO/verl training-data handoff, and an auxiliary private RAG workspace.
[ACL 2026] OPT-BENCH: Evaluating the Iterative Self-Optimization of LLM Agents in Large-Scale Search Spaces
RAG boilerplate with semantic/propositional chunking, hybrid search (BM25 + dense), LLM reranking, query enhancement agents, CrewAI orchestration, Qdrant vector search, Redis/Mongo sessioning, Cele…
Unofficial MCP server for patent search and retrieval via EPO OPS, with analyst workflow skills and an agent evaluation bench.
A unified multimodal deep research agent for images and videos, with open models, datasets, and training and evaluation code.
The official repository of "A Comprehensive Survey on Reinforcement Learning-based Agentic Search: Foundations, Roles, Optimizations, Evaluations, and Applications".
Fine-tuning Qwen2.5-3B with GRPO and LoRA to generate SVGs from text descriptions
AI Model Inference Server template with Gradio UI for Thinkube
RecHarness: A Bandit-Routed Agentic Harness for Self-Evolving Recommender Systems
DeepSeek LLM from scratch in PyTorch
A Skill Library for Automated Machine Learning
Reasoning-enhanced generative recommendation with semantic IDs, behavior-aware decoding, long-tail evaluation, and causal uplift benchmarking.
Public code repository of HEDGE submitted to KDD 2027
Multi-objective, tool-augmented symbolic regression with LLMs. Accepted to KDD 2027 (Oral).
GSM8K evaluation: Qwen-0.6B/1.7B/4B × 3 schemes (Baseline, KV-Reuse, M-Matrix)