--- title: SFT Memorizes, RL Generalizes | daily.dev description: Research comparing Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) for post-training foundation models reveals that SFT causes models to memorize... canonical: https://daily.dev/posts/sft-memorizes-rl-generalizes-3vju22atp twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: SFT Memorizes, RL Generalizes | daily.dev og:description: Research comparing Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) for post-training foundation models reveals that SFT causes models to memorize... og:url: https://daily.dev/posts/sft-memorizes-rl-generalizes-3vju22atp og:image: https://api.daily.dev/og/posts/3vJu22ATP.png og:image:alt: SFT Memorizes, RL Generalizes 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. # SFT Memorizes, RL Generalizes **[Build With GenAI](https://daily.dev/sources/buildwithgenai)** · [@javierblanco](https://daily.dev/javierblanco) · 0 upvotes · 0 comments ## Summary Research comparing Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) for post-training foundation models reveals that SFT causes models to memorize training data and fail on out-of-distribution examples, while RL enables genuine generalization. Using Llama-3.2-Vision-11B on arithmetic and navigation tasks, experiments show SFT performance collapses from 80.8% to 1.3% on new instructions, while RL improves from 80.8% to 91.8%. RL also enhances visual perception in multimodal models, whereas increased SFT compute degrades it. However, SFT remains necessary as a "format teacher" to prepare models for RL training, and RL requires verifiable objectives, making it unsuitable for subjective tasks. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Researchers propose a self-distillation fix for ‘catastrophic forgetting’ in LLMs](https://daily.dev/posts/researchers-propose-a-self-distillation-fix-for-catastrophic-forgetting-in-llms-cfeweepzh) · InfoWorld · 0 upvotes · 0 comments - [Post-Training Generative Recommenders with Advantage-Weighted Supervised Finetuning](https://daily.dev/posts/post-training-generative-recommenders-with-advantage-weighted-supervised-finetuning-aaj5ljika) · Netflix TechBlog · 1 upvotes · 0 comments - [OSFT explained: Prevent catastrophic forgetting in LLM fine-tuning](https://daily.dev/posts/osft-explained-prevent-catastrophic-forgetting-in-llm-fine-tuning-w5nzecsi8) · Red Hat Developer · 0 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#llm](https://daily.dev/tags/llm), [#computer-vision](https://daily.dev/tags/computer-vision), [#reinforcement-learning](https://daily.dev/tags/reinforcement-learning) [View this post on daily.dev](https://daily.dev/posts/sft-memorizes-rl-generalizes-3vju22atp) ```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":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/sft-memorizes-rl-generalizes-3vju22atp","headline":"SFT Memorizes, RL Generalizes","text":"Shared: SFT Memorizes, RL Generalizes","url":"https://daily.dev/posts/sft-memorizes-rl-generalizes-3vju22atp","datePublished":"2026-01-23T21:03:40.554Z","dateModified":"2026-01-23T21:04:26.516Z","author":{"@type":"Person","name":"Javier Blanco","url":"https://daily.dev/javierblanco","image":"https://lh3.googleusercontent.com/a/ACg8ocKpusKcJupXatWj8NuiQYPZEsay7nVHCbe6X4uICErB_y2Ccfb5=s96-c","description":"A Chemical Engineer fallen for Computation Maths ^^!","interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"EndorseAction"},"userInteractionCount":3060}},"image":"https://media.daily.dev/image/upload/s--CxzD6vbw--/f_auto/v1722860399/public/Placeholder%2005","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"sharedContent":{"@type":"WebPage","url":"https://api.daily.dev/r/wgsOTGW4I"},"isPartOf":{"@type":"WebPage","url":"https://daily.dev/squads/buildwithgenai","name":"Build With GenAI"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Build With GenAI","item":"https://daily.dev/squads/buildwithgenai"},{"@type":"ListItem","position":3,"name":"SFT Memorizes, RL Generalizes"}]} ```