Data science, career and other topics
09 June 2025
Only ~20% of tokens actually matter when training LLMs to reason with RL. Updating the low-entropy majority actively hurts performance — a finding that challenges standard RLVR practice.
02 June 2025
My review of the paper SWE-rebench An Automated Pipeline for Task Collection and Decontaminated Evaluation of Software Engineering Agents
26 May 2025
My review of the paper Visual Planning Let's Think Only with Images
15 May 2025
DeepMind's autonomous coding agent that evolves algorithms through LLM-driven iteration — it discovered the first improvement to matrix multiplication over Strassen's algorithm in 56 years.
28 April 2025
My review of the paper AgentA/B Automated and Scalable Web A/BTesting with Interactive LLM Agents
21 April 2025
My review of the paper M1 Towards Scalable Test-Time Compute with Mamba Reasoning Models
Type at least 2 characters to search...