--- title: ART: Train Agents That Can Learn From Experience | daily.dev description: ART (Agent Reinforcement Trainer) is an open-source framework that applies GRPO-based reinforcement fine-tuning to multi-step LLM agents. Unlike SFT, which... canonical: https://daily.dev/posts/art-train-agents-that-can-learn-from-experience-1fxmikblg twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: ART: Train Agents That Can Learn From Experience | daily.dev og:description: ART (Agent Reinforcement Trainer) is an open-source framework that applies GRPO-based reinforcement fine-tuning to multi-step LLM agents. Unlike SFT, which... og:url: https://daily.dev/posts/art-train-agents-that-can-learn-from-experience-1fxmikblg og:image: https://api.daily.dev/og/posts/1FxmIkBLG.png og:image:alt: ART: Train Agents That Can Learn From Experience og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # ART: Train Agents That Can Learn From Experience **[Daily Dose of Data Science \| Avi Chawla \| Substack](https://daily.dev/sources/dailydoseofds)** · 9 min read · 1 upvotes · 0 comments ## Summary ART (Agent Reinforcement Trainer) is an open-source framework that applies GRPO-based reinforcement fine-tuning to multi-step LLM agents. Unlike SFT, which teaches imitation, RFT lets agents improve through trial and error with reward signals. ART supports tool calls, multi-turn conversations, and integrates with LangGraph, CrewAI, and ADK. Its RULER component uses an LLM-as-a-judge to automatically generate relative reward scores without labeled data, eliminating the need to hand-craft reward functions. A hands-on walkthrough shows building a text-to-SQL agent connected to an MCP server, where a Qwen2.5-3B model learns to explore schemas, write JOINs, and handle errors through RL training—entirely locally. Benchmarks show a Qwen2.5-14B model trained with ART achieving 96% accuracy on email search, outperforming o3 while being 5x faster and 64x cheaper. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Build Agents That Can Learn Like Humans](https://daily.dev/posts/build-agents-that-can-learn-like-humans-hew5n9evu) · Daily Dose of Data Science \| Avi Chawla \| Substack · 14 upvotes · 0 comments - [How to Fine-Tune LLMs in 2026](https://daily.dev/posts/how-to-fine-tune-llms-in-2026-lx5wel0kf) · Daily Dose of Data Science \| Avi Chawla \| Substack · 1 upvotes · 0 comments - [OpenAI at QCon AI NYC: Fine Tuning the Enterprise](https://daily.dev/posts/openai-at-qcon-ai-nyc-fine-tuning-the-enterprise-gbqxbdmcn) · InfoQ · 1 upvotes · 0 comments - [Mastering Agentic Techniques: AI Agent Reinforcement Learning](https://daily.dev/posts/mastering-agentic-techniques-ai-agent-reinforcement-learning-xsi0iqacc) · NVIDIA Developer · 2 upvotes · 0 comments --- Tags: [#reinforcement-learning](https://daily.dev/tags/reinforcement-learning), [#mcp](https://daily.dev/tags/mcp) [View this post on daily.dev](https://daily.dev/posts/art-train-agents-that-can-learn-from-experience-1fxmikblg) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"ART: Train Agents That Can Learn From Experience","url":"https://daily.dev/posts/art-train-agents-that-can-learn-from-experience-1fxmikblg","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/art-train-agents-that-can-learn-from-experience-1fxmikblg"},"datePublished":"2026-02-26T18:56:36.359Z","dateModified":"2026-02-26T18:57:03.658Z","description":"ART (Agent Reinforcement Trainer) is an open-source framework that applies GRPO-based reinforcement fine-tuning to multi-step LLM agents. Unlike SFT, which...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b22d8c1c5e2b0379afec7d2d1c12ad62?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b22d8c1c5e2b0379afec7d2d1c12ad62?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Daily Dose of Data Science | Avi Chawla | Substack","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Daily Dose of Data Science | Avi Chawla | Substack","logo":"https://media.daily.dev/image/upload/s--4IHQgTOw--/f_auto/v1710503712/logos/dailydoseofds","url":"https://daily.dev/sources/dailydoseofds"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/art-train-agents-that-can-learn-from-experience-1fxmikblg","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"reinforcement-learning,mcp","timeRequired":"PT9M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Daily Dose of Data Science | Avi Chawla | Substack","item":"https://daily.dev/sources/dailydoseofds"},{"@type":"ListItem","position":3,"name":"ART: Train Agents That Can Learn From Experience"}]} ```