Temperature, Not Knowledge: What Recursive Self-Training and Its Remedies Do to a Language Model's Confidence
arXiv:2611.01059v1 Announce Type: new Abstract: Recursive self-training can collapse language models, and many proposed remedies target overconfidence, yet their effects on confidence are rarely measured. We study a five-round self-training loop on…
Read the full story at arXiv cs.AI ↗
ImpactNotable
Published1 d ago (Tue, 03 Nov 2026 05:00:00 GMT)
AuthorDipto Sumit, Syeda Fabiayee Jahan, Farig Sadeque
MoreLess
Impact score40/100
RatingRule-based estimate: event keywords (+12); trust 6/10.
RegionsGlobal
Retrievedvia rss
ClassifiedTue, 03 Nov 2026 13:09:02 GMT by heuristic