Model
MoE-POT
MoE-POT is a mixture-of-experts model for PDEs that uses a homogeneous expert architecture and serves as a comparison baseline for HotPOT.
Timeline (last 90 days)
Oct 30
HotPOT was tested on five PDE families under six settings, covering 24 same-scale comparisons before and after fine-tuning, and outperformed MoE-POT in all comparisons, reducing rollout error by an average of 50.7%.
Unverified50%
Oct 30
The 17M-parameter HotPOT model surpassed the significantly larger MoE-POT model.
Unverified50%
All Facts (2)
Unverified
HotPOT was tested on five PDE families under six settings, covering 24 same-scale comparisons before and after fine-tuning, and outperformed MoE-POT in all comparisons, reducing rollout error by an average of 50.7%.
50%UnverifiedThe 17M-parameter HotPOT model surpassed the significantly larger MoE-POT model.
50%