PewDiePie tries agentic GRPO training; HF points to four free video series
mervenoyann · x · 2026-10-01
PewDiePie is doing agentic training (he mentions GRPO, without naming tools). Hugging Face's merve adds observations: he wanted to distill Sol but got banned a few times, and built a site for donating traces that nobody used (the Hub already covers this). She then points to four HF YouTube video series on GRPO/agentic training for anyone wanting to learn how.
Related event: PewDiePie Trains His Own Local LLM with GRPO, Community Rushes to Help(4 posts)→
More from Fun
- Survival drives and multicellular organisms spontaneously emerge again in ALife primordial soup sim — ctjlewis · 2026-10-06
- 'Lost your job to AI but everything will be fine': Daniel Hentschel meltdown clip circulates — liebebio · 2026-10-06
- Viral post: 'I can't believe Claude made this' as AI-generated work spreads — bigaiguy · 2026-10-06
- Experiment: Claude one-shots an AGI explainer video using any tools it wants — burny_tech · 2026-10-06
- One-line Perl script floods your terminal with 100M random colored characters — doodlestein · 2026-10-06
- Opus-5.5 Builds a Three.js Interactive Silicon Valley City Tribute — CSProfKGD · 2026-10-06