Skip to content

Comment on DeepSeek: Inference-Time Scaling for Generalist Reward Modelingparent

Comments

Deepseek is super bad at personal advice. I can tell it was trained on an oddly stodgy data set. It gives advice that would suit a hyper conservative world view. Like IBM 1950s middle management training course level advice.

Gemma is by far the best at giving advice and planning ones days and life priorities. Not sure how to benchmark that.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.