PinnedPublished inTDS ArchiveA long-term Data Science roadmap which WON’T help you become an expert in only several monthsFrom time to time I’m asked: how does one become Data scientist? What courses are necessary? How long will it take? How did you become DS…Dec 2, 2018A response icon19Dec 2, 2018A response icon19
Paper Review: Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement…Training models on high-entropy tokens drives effective reinforcement learning for LLM reasoning!Jun 9Jun 9
Paper Review: SWE-rebench: An Automated Pipeline for Task Collection and Decontaminated Evaluation…SWE-rebench: an automated pipeline for Task Collection and evaluation of software engineering agents. And a benchmark based on it!Jun 2Jun 2
Paper Review: Visual Planning: Let’s Think Only with ImagesVisual Planning with RL: thinking through images, not texts!May 26May 26
Paper Review: AlphaEvolve: A coding agent for scientific and algorithmic discoveryAlphaEvolve: an LLM-driven coding agent that improves real-world infra (scheduling, TPUs), accelerates its own training, and beats…May 15May 15
Paper Review: AgentA/B: Automated and Scalable Web A/BTesting with Interactive LLM AgentsAutomated A/B tests with LLM agents!Apr 28Apr 28
Paper Review: M1: Towards Scalable Test-Time Compute with Mamba Reasoning ModelsMamba with test-time scaling and reasoning!Apr 21Apr 21
Paper Review: TextCrafter: Accurately Rendering Multiple Texts in Complex Visual ScenesBetter image generation with multiple texts in complex scenes!Apr 7Apr 7
Paper Review: Video-T1: Test-Time Scaling for Video GenerationVideo generation with Test-Time Scaling instead of training more!Mar 31Mar 31
Paper Review: RWKV-7 “Goose” with Expressive Dynamic State EvolutionRNN with a transformer to match SoTA English & multilingual LLMs at a 3B scale using fewer tokens, constant memory/time per token, and a…Mar 24Mar 24
