Research | Tsinghua KEG

Research Areas

Pioneering the next generation of AI by bridging data-driven learning, knowledge modeling, and symbolic reasoning.

Large Language Model

Scaling Laws & Training Infrastructure

Long-Context & Efficient Inference

Multilingual & Knowledge-Enhanced Models

Covers: GLM, GLM-130B, ChatGLM, GLM-4/5

Vision-Language Models

Text-to-Image Generation

Text-to-Video Generation

Unified Multimodal Representation Learning

Covers: CogVLM, GLM-V, CogView, CogVideo

Agent Reasoning & Planning

Tool-Using and Web Agents

Long-Horizon Task Execution

Self-Reflection & Self-Improvement

Covers: AutoGLM, WebGLM, WebRL, AgentBench

Foundation Models for Scientific Domains

Molecular Representation & Drug Discovery

Protein Structure Prediction & Design

AI-Driven Scientific Data

Covers: xTrimoPGLM, ProteinLM

Code Generation & Completion

Automated Program Synthesis

Code Understanding, Review & Verification

Agentic Coding Systems

Covers: CodeGeeX, code agents

Model Robustness & Alignment

Interpretability & Monitoring

Privacy-Preserving Learning

Safety Evaluation for Large Models

Covers: AI safety, alignment, interpretability, controllability