andy12_'s submissions

1.		From Memorization to Reasoning in the Spectrum of Loss Curvature (arxiv.org)
		65 points by andy12_ 34 days ago \| past \| 14 comments
2.		Concrete "battery" developed at MIT now packs 10 times the power (news.mit.edu)
		3 points by andy12_ 69 days ago \| past \| 2 comments
3.		Gauss, an Agent for Autoformalization (math.inc)
		6 points by andy12_ 3 months ago \| past
4.		Spurious Rewards: Rethinking Training Signals in RLVR (rethink-rlvr.notion.site)
		1 point by andy12_ 6 months ago \| past
5.		VR-CLI: Learning to Reason for Long-Form Story Generation (arxiv.org)
		2 points by andy12_ 7 months ago \| past
6.		Tokenformer: Rethinking transformer scaling with tokenized model parameters (arxiv.org)
		3 points by andy12_ on Oct 31, 2024 \| past \| 1 comment
7.		Selective Attention Improves Transformer (arxiv.org)
		1 point by andy12_ on Oct 7, 2024 \| past \| 1 comment
8.		The AdEMAMix Optimizer: Better, Faster, Older (arxiv.org)
		2 points by andy12_ on Sept 10, 2024 \| past