🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

“On SWE-bench Verified, GPT-4o resolves 33.2% of samples, with the best performing open-source scaffold, Agentless.” Wow, our Agentless😺 is used by OpenAI as the default technique to evaluate the performance of GPT-4o on SWE-bench! 🧑‍💻 github.com/OpenAutoCoder/…

Thanks Tanishq Abraham is at ICML for sharing our work! It's been an amazing summer interning at SnowflakeDB AI Research team. I started with no pretraining experience but learned so much. Special thanks to the Arctic team for their invaluable support!! Hojae Han Rajhans Samdani

📢Announcing the 2nd workshop on #LLM4Code, co-located with @ICSEConf 2025 in Ottawa, Canada 🇨🇦! 🎯We are calling for submissions: 🌟Website: llm4code.github.io 🌟Deadline: Nov 18, 2024 🌟8-page research paper / 4-page position paper (including references) 🚀Calling for

How to leverage the white-box info (i.e. source code) for fuzzing compilers? Check out our work “WhiteFox 🦊: White-Box Compiler Fuzzing Empowered by Large Language Models” at OOPSLA 2024! w/ Yinlin Deng, Runyu Lu, JIayi Yao, Jiawei Liu, Reyhan, and Lingming Zhang (1/N)

Human preference has been commonly used for RLHF and eval — but: 🤔How reliable is that for reasoning-heavy areas such as codegen? 🤔Can models effectively predict code preference? We present techniques, evals, & studies to demystify code preference: llm-code-preference.github.io 🧵

💪 I exhausted all my strength to give you the best.

🚨Just one week left to submit to #LLM4Code25!(Co-located with ICSE) We also offer a non-archival option, if selected, your camera-ready will be shared only on our workshop website, not the proceedings (no need to worry about double submission if you have other submission plan)

Interested in code generation and instruction tuning without distillation? Please drop by our #NeurIPS poster (East Axhibit # 2502) this afternoon (Friday / Dec 13)! Sorry to miss NeurIPS this year 🥺 but Jiawei Liu will be there to present the poster. Talk to him!

Find us @ East Exhibit # 2502 (⏰ 4:30PM - ) if you are at #NeurIPS2024 :D

Happy to release SWE Arena, your vibe coding platform! SWE Arena supports real-time code execution and rendering, covering various frontier LLMs & VLMs! We actually had this idea two years ago inside BigCode with Arjun Guha and Daniel Fried. However, there wasn't much tech

Meta just dropped SWE-RL Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution Trained on top of Llama 3, our resulting reasoning model, Llama3-SWE-RL-70B, achieves a 41.0% solve rate on SWE-bench Verified -- a human-verified collection of real-world

SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution arxiv.org/abs/2502.18449 by Yuxiang Wei Sida Wang and the whole team! Get started with your favorite model here github.com/facebookresear…

📢 Excited to share our new paper: Challenges and Paths Towards AI for SWE We discuss: 🛠️ 6 sub-tasks needed for SWE 🤖 9 challenges of today's AI in SWE 🔮 9 future directions to address the challenges w/ collaborators from MIT, Berkeley, Cornell, Stanford, and UPenn ⬇️ (1/n)