{"id":23972,"date":"2025-02-23T22:00:32","date_gmt":"2025-02-23T19:00:32","guid":{"rendered":"https:\/\/w3.bilkent.edu.tr\/en\/?p=23972"},"modified":"2025-02-25T09:22:53","modified_gmt":"2025-02-25T06:22:53","slug":"math-seminar-mathematical-foundations-of-deepseek-a-reinforcement-learning-perspective-berkay-anahtarci-500pm-february-24-2025-en","status":"publish","type":"post","link":"https:\/\/w3.bilkent.edu.tr\/en\/math-seminar-mathematical-foundations-of-deepseek-a-reinforcement-learning-perspective-berkay-anahtarci-500pm-february-24-2025-en\/","title":{"rendered":"MATH Seminar: \u201cMathematical Foundations of DeepSeek: A Reinforcement Learning Perspective\u201d, Berkay Anahtarc\u0131, 5:00PM February 24 2025 (EN)"},"content":{"rendered":"<p>You are cordially invited to the Analysis Seminar organized by the Department of Mathematics.<\/p>\n<p>Speaker: Berkay Anahtarc\u0131 (\u00d6zye\u011fin University)<\/p>\n<p>\u201cMathematical Foundations of DeepSeek: A Reinforcement Learning Perspective.\u201d<\/p>\n<p>Abstract: This talk explores the mathematical foundations of DeepSeek R1, an RL-only model designed for complex reasoning. Unlike traditional supervised fine-tuning, DeepSeek R1 leverages Group Relative Policy Optimization (GRPO), a novel method that stabilizes Proximal Policy Optimization (PPO) without a critic. GRPO enhances chain-of-thought reasoning by structuring problem-solving into sequential steps. I will analyze its theoretical properties and implications for reasoning-driven reinforcement learning.<\/p>\n<p>Date: Monday, February 24, 2025<br \/>\nTime: 17:00-18:00, GMT+3<br \/>\nPlace: ZOOM<\/p>\n<p>This is an online seminar. To request the zoom link, please send a message to goncha@fen.bilkent.edu.tr<\/p>\n","protected":false},"excerpt":{"rendered":"<p>You are cordially invited to the Analysis Seminar organized by the Department of Mathematics. Speaker: Berkay Anahtarc\u0131 (\u00d6zye\u011fin University) \u201cMathematical Foundations of DeepSeek: A Reinforcement Learning Perspective.\u201d Abstract: This talk explores the mathematical foundations of DeepSeek R1, an RL-only model designed for complex reasoning. Unlike traditional supervised fine-tuning, DeepSeek R1 [\u2026] <\/p>\n<div class=\"clear\"><\/div>\n<p><a class=\"more_link clearfix\" href=\"https:\/\/w3.bilkent.edu.tr\/en\/math-seminar-mathematical-foundations-of-deepseek-a-reinforcement-learning-perspective-berkay-anahtarci-500pm-february-24-2025-en\/\" rel=\"nofollow\">Read More<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[12],"tags":[],"class_list":["post-23972","post","type-post","status-publish","format-standard","hentry","category-oldposts"],"_links":{"self":[{"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/posts\/23972","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/comments?post=23972"}],"version-history":[{"count":1,"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/posts\/23972\/revisions"}],"predecessor-version":[{"id":23973,"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/posts\/23972\/revisions\/23973"}],"wp:attachment":[{"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/media?parent=23972"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/categories?post=23972"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/w3.bilkent.edu.tr\/en\/wp-json\/wp\/v2\/tags?post=23972"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}