Schedule

Schedule

Time Event
09:10 - 09:20 Opening Remarks
09:20 - 10:00 Invited Talk #1: Graham Neubig (CMU & OpenHands), “Lessons from the Trenches in Building Agents for Software Development”
10:00 - 10:40 Invited Talk #2: Sida Wang (Meta FAIR), “Predicting all the error bars of LLM evaluations”
10:40 - 11:30 Spotlight Sessions
• SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution
• EquiBench: Benchmarking Large Language Models’ Understanding of Program Semantics via Equivalence Checking
• Constrained Decoding of Diffusion LLMs with Context-Free Grammars
• SWE-Perf: Can Language Models Optimize Code Performance on Real-World Repositories?
• Training LLM Agents to Empower Humans
11:30 - 12:10 Invited Talk #3: Dawn Song (Berkeley), “Frontier AI for Secure Code: Challenges and Future Directions”
12:10 - 13:20 Lunch & Social
13:20 - 14:20 Poster Session
14:20 - 15:00 Invited Talk #4: Liangchen Luo (xAI), “How to Develop in the Agentic Era”
15:00 - 15:30 Coffee Break & Social
15:30 - 16:10 Invited Talk #5: Junyang Lin (Alibaba), “Qwen3-Coder: Agentic Coding in the World”
16:10 - 16:50 Panel Discussion: Michele Catasta (Replit), Graham Neubig (OpenHands/CMU), Eiso Kant (poolside), Sida Wang (Meta FAIR), Liangchen Luo (xAI)
16:50 - 17:00 Closing Remarks