Stepsize anything: A unified learning rate schedule for budgeted-iteration training
Paper • 2505.24452 • Published • 5 •
2
Yutao Zeng
Taoer
AI & ML interests
None yet
Recent Activity
upvoted a paper about 1 month ago
InCoder-32B: Code Foundation Model for Industrial Scenarios authored a paper 5 months ago
Virtual Width Networks upvoted a paper 5 months ago
Virtual Width NetworksOrganizations
commented a
paper 11 months ago
commented 4
papers about 1 year ago
HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization
Paper • 2503.04598 • Published • 22 •
8
HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization
Paper • 2503.04598 • Published • 22 •
8
HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization
Paper • 2503.04598 • Published • 22 •
8
Scale-Distribution Decoupling: Enabling Stable and Effective Training of Large Language Models
Paper • 2502.15499 • Published • 15 •
2
commented 2
papers over 1 year ago