Ask HN: Who is doing distillation or fine tuning?

2 pointsmrburton2 comments

If you're doing fine tuning, how are you curating or creating your training data? If you're doing distillation what tools? Share you're experience.

That includes you PewDiePie; Felix Kjellberg.

View on HN

For me, training data comes from agent runs.

I do fine tuning with pytorch, and RL-forms of distillation with Tinker! I hope that helps. I work with a full stack of models that enable agentic AI: embedders, rerankers, LLMs, decision models.

Curious, why are you asking?

I'm genuinely curious about this topic and love learning. My fundamental belief is that it makes sense to build highly tuned models that specialized in specific areas and using orchestration on top to call out to these specialized models with a "shared memory". This would allow for the ability to improve parts of the system without the fear of regression in other areas and it would be cheaper to create a smart "platform". I know there's models with MoE etc but feel like the "microservices" approach (micromodels) would work well here.

I could be totally wrong.

Reply on news.ycombinator.com