Supervised Fine Tuning on Curated Data is Reinforcement Learning (and can be improved) · HackerPapers discussions

Supervised Fine Tuning on Curated Data is Reinforcement Learning (and can be improved) — mentioned by andy99

Sun, 17 Aug 2025 12:32:44 GMT

A couple things I've seen go by that make the connection. I haven't looked at them closely enough to have an opinion. > https://arxiv.org/abs/2507.12856 > https://justinchiu.netlify.app/blog/sftrl/

Paper and discussion history. A mention is not an endorsement.

Supervised Fine Tuning on Curated Data is Reinforcement Learning (and can be improved) — mentioned by roadside_picnic

Wed, 30 Jul 2025 17:14:16 GMT

That original model collapse paper has largely been misunderstood and in practice, this is only true if you're not curating the generated data at all. The original paper even specifies (emphasis mine): > We find that indiscriminate use of model-generated content in training causes irreversible defects in the resulting models, in which tails of the original content distribution disappear. [0] In practice nobody is "indiscriminately" using model output to fine-tune models since that doesn't even make sense. Even if you're harvesting web data generated by LLMs, that data has in fact been curated…

Paper and discussion history. A mention is not an endorsement.

Supervised Fine Tuning on Curated Data is Reinforcement Learning (and can be improved) — HN submission

Tue, 29 Jul 2025 20:18:23 GMT

Supervised fine tuning on curated data is reinforcement learning

71 points; 19 discussion replies on HN. Paper and source history. A submission is not an endorsement.