from Hacker News

s-macke

joined 9/17/13, 3:36 PM has 1674 karma

• Reinforcement Learning Finetunes Small Subnetworks in Large Language Models

  by s-macke on 5/21/25, 11:31 AM, with 0 comments

• The simplest, fastest repository for training/finetuning small-sized VLMs

  by s-macke on 5/6/25, 6:37 PM, with 0 comments

• Gemini 2.5 Pro won Pokémon Blue in 106k moves

  by s-macke on 5/3/25, 8:08 AM, with 3 comments

• SWE-Smith: Scaling Data for Software Engineering Agents

  by s-macke on 5/1/25, 12:17 PM, with 0 comments

• Sam Altman: we added one million users in the last hour

  by s-macke on 3/31/25, 6:34 PM, with 0 comments

• Measuring AI Ability to Complete Long Tasks

  by s-macke on 3/19/25, 9:18 PM, with 0 comments

• FrontierMath Was Funded by OpenAI

  by s-macke on 1/19/25, 8:50 AM, with 4 comments

• Grokking at the Edge of Numerical Stability

  by s-macke on 1/15/25, 6:08 PM, with 0 comments

• LLMs struggle with perception, not reasoning, in ARC-AGI

  by s-macke on 12/24/24, 7:32 PM, with 0 comments

• A Llama 70B finetune that has reflection baked into it's weights

  by s-macke on 9/5/24, 7:10 PM, with 0 comments

• OpenAI releases ChatGPT on your desktop for macOS

  by s-macke on 6/25/24, 7:27 PM, with 20 comments

• Examination of Large Language Model Performance on Grade School Arithmetic

  by s-macke on 5/3/24, 6:13 AM, with 0 comments

• Llama3 is the first open model that can also win text adventures

  by s-macke on 4/20/24, 11:10 AM, with 16 comments