Youtu-LLM
Lightweight, high-performance Large Language Models (2B parameters) for edge deployment.
Di Yin尹迪
Principal Researcher & Team Leader
I lead research at Tencent Youtu Lab, exploring large language models, multimodal intelligence, and agentic systems. My work bridges foundation models and real-world applications through open-source tools and benchmarks.
Research interestsLLM & MultimodalAgentic IntelligenceRetrieval-Augmented Generation
A closer look
A little about me.
My research spans model architecture optimization, native multimodal pre-training, and evaluation of large language models, with a focus on connecting advances in foundation models to practical applications.
My work centers on three core pillars: LLM pre-training and architecture optimization, Agent, and RAG. Through robust, open-source tools and benchmarks, I aim to bridge the gap between foundation models and real-world applications.
- Papers published
- 50+
- Google Scholar citations
- 1,400+
- Personal GitHub stars
- 6,000+
Current research focusMy current work focuses on the training of native multimodal pre-trained models and the exploration of unified model architectures for both understanding and generation.
Open-source projects.
From foundation models to practical intelligence.
LLM & MLLM
Models & multimodal understandingYoutu-VL
Open-source Vision-Language models including training recipes and inference code (e.g., Youtu-VL-4B).
Youtu-Parsing-Omni
One model that parses documents, images, charts, geometry, audio and video into structured JSON.
RAG
Retrieval, reasoning & knowledgeYoutu-RAG
A next-generation agentic retrieval-augmented generation system built on the “Local Deployment · Autonomous Decision · Memory-Driven” paradigm.
Youtu-GraphRAG
A vertically unified agentic paradigm that jointly connects the entire framework as an intricate integration based on graph schema.
Youtu-Embedding
An industry-leading, general-purpose text representation model developed by Tencent Youtu Lab.
Selected publications.
Research in language, vision, and agentic systems · 2025–2026
14publications
-
Youtu-GraphRAG: Vertically Unified Agents for Graph Retrieval-Augmented Complex Reasoning
ICLR2026 -
Collision to Cognition: Hash-Driven Graph Construction for Efficient RAG
ACL2026 -
Query-Aware Knowledge Retrieval via Hyperbolic Structuring
ACL2026 -
PoLi-RL: A Point-to-List Reinforcement Learning Framework for Conditional Semantic Textual Similarity
ICLR2026 -
Enhancing Refusal Capabilities of Role-Playing Agents via Representation Space Analysis and Editing
ACL2025 -
Mac-SQL: A Multi-Agent Collaborative Framework for Text-to-SQL
COLING2025 -
Youtu-LLM: Unlocking the Native Agentic Potential for Lightweight Large Language Models
arXiv2026 -
TPLA: Tensor Parallel Latent Attention for Efficient Disaggregated Prefill and Decode Inference
ASPLOS2026 -
SSA: Sparse Sparse Attention by Aligning Full and Sparse Attention Outputs in Feature Space
ICML2026 -
HISA: Efficient Hierarchical Indexing for Fine-Grained Sparse Attention
CoLM2026 -
A Curriculum Learning Approach to Automated Reasoning with Large Language Models
ACL2025 -
Youtu-VL: Unleashing Visual Potential via Unified Vision-Language Supervision
arXiv2026 -
Toward Native Multimodal Modeling: A Roadmap
EMNLP2026 -
TangramPuzzle: Evaluating Multimodal Large Language Models with Compositional Spatial Reasoning
EMNLP2026
No matching publications. Try another keyword or research area.
Showing 14 of 14 publicationsFull list on Google Scholar
Let's connect.
Research, open source, and ideas worth sharing.