Tongyi Lab, Alibaba Group

[EMNLP 2025 & AACL 2026] Code for "Detecting Knowledge Boundary of Vision Large Language Models by Sampling-Based Inference" & "Dependency-Aware Trajectory Refinement for Efficient Multi-Turn Agent Fine-Tuning"

qqr is an RL training framework for open-ended agents.

Official repository for the paper CORE: Improving Compositional Reasoning in MLLM Embedding via Reranker Distillation

E2Rank: Your Text Embedding can Also be an Effective and Efficient Listwise Reranker

Multimodal Retrieval-augmented Generation Framework Built by Tongyi Lab, Alibaba Group.

Tongyi Deep Research, the Leading Open-source Deep Research Agent

[EMNLP 2025] ViDoRAG: Visual Document Retrieval-Augmented Generation via Dynamic Iterative Reasoning Agents

A new evaluation paradigm for deep search that identifies specific LLM failure sources, introduces challenging hint-free datasets with holistic evaluation, and offers a strong baseline incorporating memory and verification.