MASAI: Modular Architecture for Software-engineering AI Agents
A common method to solve complex problems in software engineering, is to divide the problem into multiple sub-problems. Inspired by this, we propose a Modular Architecture for Software-engineering AI (MASAI) agents, where different LLM-powered sub-agents are instantiated with well-defined objectives...
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | Arora, Daman Sonwane, Atharv Wadhwa, Nalin Mehrotra, Abhav Utpala, Saiteja Bairi, Ramakrishna Kanade, Aditya Natarajan, Nagarajan |
| description | A common method to solve complex problems in software engineering, is to
divide the problem into multiple sub-problems. Inspired by this, we propose a
Modular Architecture for Software-engineering AI (MASAI) agents, where
different LLM-powered sub-agents are instantiated with well-defined objectives
and strategies tuned to achieve those objectives. Our modular architecture
offers several advantages: (1) employing and tuning different problem-solving
strategies across sub-agents, (2) enabling sub-agents to gather information
from different sources scattered throughout a repository, and (3) avoiding
unnecessarily long trajectories which inflate costs and add extraneous context.
MASAI enabled us to achieve the highest performance (28.33% resolution rate) on
the popular and highly challenging SWE-bench Lite dataset consisting of 300
GitHub issues from 11 Python repositories. We conduct a comprehensive
evaluation of MASAI relative to other agentic methods and analyze the effects
of our design decisions and their contribution to the success of MASAI. |
| doi_str_mv | 10.48550/arxiv.2406.11638 |
| format | Article |
| fullrecord | <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2406_11638</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2406_11638</sourcerecordid><originalsourceid>FETCH-LOGICAL-a678-20b239ba8cb2743e99d9198cc51a0b31e90451cd00d717818235033f21c3f39d3</originalsourceid><addsrcrecordid>eNotz7tOwzAUgGEvDKjwAEz4BRJ8fOLE7oBkVVwitWJo98ixj1NLJUFuyuXtEaXTv_3Sx9gdiLLSSokHl7_TZykrUZcANepr9rixW9su-WYKp4PL3Ga_TzP5-ZSJxynz7RTnL5epoHFII1FO48Bty-1A43y8YVfRHY50e-mC7Z6fdqvXYv320q7sunB1owspeommd9r3sqmQjAkGjPZegRM9AhlRKfBBiNBAo0FLVAIxSvAY0QRcsPv_7RnQfeT07vJP9wfpzhD8BZBVQRo</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>MASAI: Modular Architecture for Software-engineering AI Agents</title><source>arXiv.org</source><creator>Arora, Daman ; Sonwane, Atharv ; Wadhwa, Nalin ; Mehrotra, Abhav ; Utpala, Saiteja ; Bairi, Ramakrishna ; Kanade, Aditya ; Natarajan, Nagarajan</creator><creatorcontrib>Arora, Daman ; Sonwane, Atharv ; Wadhwa, Nalin ; Mehrotra, Abhav ; Utpala, Saiteja ; Bairi, Ramakrishna ; Kanade, Aditya ; Natarajan, Nagarajan</creatorcontrib><description>A common method to solve complex problems in software engineering, is to
divide the problem into multiple sub-problems. Inspired by this, we propose a
Modular Architecture for Software-engineering AI (MASAI) agents, where
different LLM-powered sub-agents are instantiated with well-defined objectives
and strategies tuned to achieve those objectives. Our modular architecture
offers several advantages: (1) employing and tuning different problem-solving
strategies across sub-agents, (2) enabling sub-agents to gather information
from different sources scattered throughout a repository, and (3) avoiding
unnecessarily long trajectories which inflate costs and add extraneous context.
MASAI enabled us to achieve the highest performance (28.33% resolution rate) on
the popular and highly challenging SWE-bench Lite dataset consisting of 300
GitHub issues from 11 Python repositories. We conduct a comprehensive
evaluation of MASAI relative to other agentic methods and analyze the effects
of our design decisions and their contribution to the success of MASAI.</description><identifier>DOI: 10.48550/arxiv.2406.11638</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Software Engineering</subject><creationdate>2024-06</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2406.11638$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2406.11638$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Arora, Daman</creatorcontrib><creatorcontrib>Sonwane, Atharv</creatorcontrib><creatorcontrib>Wadhwa, Nalin</creatorcontrib><creatorcontrib>Mehrotra, Abhav</creatorcontrib><creatorcontrib>Utpala, Saiteja</creatorcontrib><creatorcontrib>Bairi, Ramakrishna</creatorcontrib><creatorcontrib>Kanade, Aditya</creatorcontrib><creatorcontrib>Natarajan, Nagarajan</creatorcontrib><title>MASAI: Modular Architecture for Software-engineering AI Agents</title><description>A common method to solve complex problems in software engineering, is to
divide the problem into multiple sub-problems. Inspired by this, we propose a
Modular Architecture for Software-engineering AI (MASAI) agents, where
different LLM-powered sub-agents are instantiated with well-defined objectives
and strategies tuned to achieve those objectives. Our modular architecture
offers several advantages: (1) employing and tuning different problem-solving
strategies across sub-agents, (2) enabling sub-agents to gather information
from different sources scattered throughout a repository, and (3) avoiding
unnecessarily long trajectories which inflate costs and add extraneous context.
MASAI enabled us to achieve the highest performance (28.33% resolution rate) on
the popular and highly challenging SWE-bench Lite dataset consisting of 300
GitHub issues from 11 Python repositories. We conduct a comprehensive
evaluation of MASAI relative to other agentic methods and analyze the effects
of our design decisions and their contribution to the success of MASAI.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Software Engineering</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz7tOwzAUgGEvDKjwAEz4BRJ8fOLE7oBkVVwitWJo98ixj1NLJUFuyuXtEaXTv_3Sx9gdiLLSSokHl7_TZykrUZcANepr9rixW9su-WYKp4PL3Ga_TzP5-ZSJxynz7RTnL5epoHFII1FO48Bty-1A43y8YVfRHY50e-mC7Z6fdqvXYv320q7sunB1owspeommd9r3sqmQjAkGjPZegRM9AhlRKfBBiNBAo0FLVAIxSvAY0QRcsPv_7RnQfeT07vJP9wfpzhD8BZBVQRo</recordid><startdate>20240617</startdate><enddate>20240617</enddate><creator>Arora, Daman</creator><creator>Sonwane, Atharv</creator><creator>Wadhwa, Nalin</creator><creator>Mehrotra, Abhav</creator><creator>Utpala, Saiteja</creator><creator>Bairi, Ramakrishna</creator><creator>Kanade, Aditya</creator><creator>Natarajan, Nagarajan</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20240617</creationdate><title>MASAI: Modular Architecture for Software-engineering AI Agents</title><author>Arora, Daman ; Sonwane, Atharv ; Wadhwa, Nalin ; Mehrotra, Abhav ; Utpala, Saiteja ; Bairi, Ramakrishna ; Kanade, Aditya ; Natarajan, Nagarajan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a678-20b239ba8cb2743e99d9198cc51a0b31e90451cd00d717818235033f21c3f39d3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Software Engineering</topic><toplevel>online_resources</toplevel><creatorcontrib>Arora, Daman</creatorcontrib><creatorcontrib>Sonwane, Atharv</creatorcontrib><creatorcontrib>Wadhwa, Nalin</creatorcontrib><creatorcontrib>Mehrotra, Abhav</creatorcontrib><creatorcontrib>Utpala, Saiteja</creatorcontrib><creatorcontrib>Bairi, Ramakrishna</creatorcontrib><creatorcontrib>Kanade, Aditya</creatorcontrib><creatorcontrib>Natarajan, Nagarajan</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Arora, Daman</au><au>Sonwane, Atharv</au><au>Wadhwa, Nalin</au><au>Mehrotra, Abhav</au><au>Utpala, Saiteja</au><au>Bairi, Ramakrishna</au><au>Kanade, Aditya</au><au>Natarajan, Nagarajan</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>MASAI: Modular Architecture for Software-engineering AI Agents</atitle><date>2024-06-17</date><risdate>2024</risdate><abstract>A common method to solve complex problems in software engineering, is to
divide the problem into multiple sub-problems. Inspired by this, we propose a
Modular Architecture for Software-engineering AI (MASAI) agents, where
different LLM-powered sub-agents are instantiated with well-defined objectives
and strategies tuned to achieve those objectives. Our modular architecture
offers several advantages: (1) employing and tuning different problem-solving
strategies across sub-agents, (2) enabling sub-agents to gather information
from different sources scattered throughout a repository, and (3) avoiding
unnecessarily long trajectories which inflate costs and add extraneous context.
MASAI enabled us to achieve the highest performance (28.33% resolution rate) on
the popular and highly challenging SWE-bench Lite dataset consisting of 300
GitHub issues from 11 Python repositories. We conduct a comprehensive
evaluation of MASAI relative to other agentic methods and analyze the effects
of our design decisions and their contribution to the success of MASAI.</abstract><doi>10.48550/arxiv.2406.11638</doi><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | DOI: 10.48550/arxiv.2406.11638 |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_arxiv_primary_2406_11638 |
| source | arXiv.org |
| subjects | Computer Science - Artificial Intelligence Computer Science - Software Engineering |
| title | MASAI: Modular Architecture for Software-engineering AI Agents |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-05T21%3A41%3A55IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=MASAI:%20Modular%20Architecture%20for%20Software-engineering%20AI%20Agents&rft.au=Arora,%20Daman&rft.date=2024-06-17&rft_id=info:doi/10.48550/arxiv.2406.11638&rft_dat=%3Carxiv_GOX%3E2406_11638%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |