Bake AI Research (BakeLab)

The work behind the ideas.

Technical papers from BakeLab and our research collaborators.

arXiv · 2026

AutoLab: Can Frontier Models Solve Long-Horizon Auto Research and Engineering Tasks?

Autonomous Research
Agent Evaluation
Long-horizon Tasks

Read paper Read the story

arXiv · 2026

JobBench: Aligning Agent Work With Human Will

General Agent
Work Automation
Agent Evaluation

arXiv · 2026

NARRA-Gym for Evaluating Interactive Narrative Agents

Interactive Narratives
Personalization
Agent Evaluation

arXiv · 2025

PersonaMem-v2: Towards Personalized Intelligence via Learning Implicit User Personas and Agentic Memory

Personalization
Agent Memory
Long-context Reasoning

ICLR · 2026

Building a Foundational Guardrail for General Agentic Systems via Synthetic Data

Agent Safety
Guardrails
Synthetic Data

ICLR · 2026

CoDA: Agentic Systems for Collaborative Data Visualization

Multi-agent Systems
Data Visualization
Code Generation

EMNLP · 2026

TOUCAN: Synthesizing 1.5M Tool-Agentic Data from Real-World MCP Environments

Tool Use
MCP Environments
Synthetic Data

NeurIPS · 2025

ChemOrch: Empowering LLMs with Chemical Intelligence via Groundbreaking Synthetic Instructions

Chemistry
Synthetic Data
Tool Use

arXiv · 2025

VisualSphinx: Large-Scale Synthetic Vision Logic Puzzles for RL

Multimodal Reasoning
Visual Puzzles
Reinforcement Learning

ACL · 2025

KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding

Code Generation
Synthetic Data
Unit Testing

NAACL · 2025

Stronger Models are NOT Stronger Teachers for Instruction Tuning

Instruction Tuning
Model Distillation
Model Compatibility

ICLR · 2025

Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing

Instruction Tuning
Synthetic Data
Model Alignment