A new open-source software engineering agent called SWE-agent has been developed, achieving impressive results on SWE-bench. The agent, led by jyangballin and _carlosejimenez, offers a 12.3% resolve rate and competes with the closed-source Devin. SWE-agent utilizes an Agent-Computer Interface (ACI) design to enhance its functionality and efficiency, allowing it to edit and run code autonomously on GitHub repositories. The agent, based on ReAct and ACI, demonstrates state-of-the-art performance and is available for public use, contrasting with the closed-source and waitlisted Devin.
SWE-Agent is a fully autonomous coding agent WITH TOOLS. Simply drop a GitHub issue URL and it'll replicate the issue, attempt to solve it, and prepare a fix. Here's a full review and tutorial ๐ฅ๐ https://t.co/WBCsbnapcn
SWE-agent turns LMs (e.g. GPT-4) into software engineering agents that can fix bugs and issues in real GitHub repositories. https://t.co/ABINx7387H https://t.co/oXfykxyZ6E
SWE-agent, from Princetonโs NLP lab, is the best open-source alternative for Devin, Demo, easy setup guide, and everything you need to know in this ๐งต https://t.co/bcWSjYFHKB
SWE-agent is finally out. A few highlights: 1. Agent-Computer Interface (ACI) design will be critical for the success of AI agents, much like HCI is critical for how effective humans are with computers. 2. You can use SWE-agent out of the box on any github issue. (1/2) https://t.co/Cbh7qUR6Ei https://t.co/5LdbsVkbye
Another nice project - SWE-agent โจ ๐ SWE-agent turns LMs (e.g. GPT-4) into software engineering agents that can fix bugs and issues in real GitHub repositories. ๐ There are two steps to the SWE-agent pipeline. First SWE-agent takes an input GitHub issue and returns a pullโฆ https://t.co/p7reEDHogM
Seems like the creators of SWEBench (benchmark used by Devin), just updated their leaderboard and released an open source competitor to Devin, SWE-Agent, that has a 12.3% resolve rate. Devin's was 13.86% and is not public or open source. They also published a paper to go withโฆ https://t.co/S5Q8W0nTfm
New OPEN SOURCE Software ENGINEER Agent Outperforms ALL! (New SWE AGENT!) https://t.co/hxmerRj8AA
Impressive results from @jyangballin at Princeton releasing an OSS SWE agent at near-parity with Devin (reported). Not only is Devin closed source but you can't even use it due to the waitlist! SWE Agent is open-source and free for everyone. https://t.co/JtN6LTZKSS https://t.co/JTPJ5R4nB5
SWE-agent led by amazing @jyangballin @_carlosejimenez , first authors of SWE-bench Besides code base, also check out our discord https://t.co/BogeKNYmjP https://t.co/tcktoanGch
Extremely excited to open-source our SWE-agent that achieves SoTA on SWE-bench๐ Turns out ReAct + Agent-Computer Interface (ACI) can go a long way, very excited about the implications for SWE and beyond! https://t.co/wnlk2HhYo2
Open source Devin, with very impressive numbers. From the thread: 'letting SWE-agent only view 100 lines at a time was better than letting it view 200 or 300 lines and much better than letting it view the entire file'. https://t.co/mQsplpewWw
SWE-Agent is an open-source software engineering agent with a 12.3% resolve rate on SWE-Bench! Check out SWE-agent in action at https://t.co/1NNL526gMy Repo: https://t.co/LsgeVvD1UC https://t.co/KZDmFw67l3
SWE-agent is our new system for autonomously solving issues in GitHub repos. It gets similar accuracy to Devin on SWE-bench, takes 93 seconds on avg + it's open source! We designed a new agent-computer interface to make it easy for GPT-4 to edit+run code https://t.co/CTzMxDiouH https://t.co/VW9FuZGIUf