Skip to content

jincan333/Dtopp.github.io

Repository files navigation

DTop-p MoE — ICML 2026 Project Page

Sparsity-Controlled Dynamic Top-p MoE for Foundation Model Pre-training Can Jin*, Hongwu Peng*, Mingcan Xiang, Qixin Zhang, Xiangchi Yuan, Amit Hasan, Ohi Dibua, Yifan Gong, Yan Kang†, Dimitris N. Metaxas† — ICML 2026 (PMLR 306).

🔗 Live page: https://jincan333.github.io/Dtopp.github.io/

This repository is the project page for the paper. It is a static site (no build step) — all figures, tables, and equations are clipped directly from the paper PDF.

Contents

File What
index.html Project page (responsive landing page)
slides.html / slides.pdf 16:9 talk deck (interactive + PDF)
poster.html / poster_print_48x36in.pdf Conference poster (web + 48″×36″ print PDF)
poster_highres.png 4320×3240 poster raster
thumbnail.png / thumbnail_1200.png Social / preview thumbnails
paper.pdf Full manuscript
figures/ Figures, tables, and equations clipped from the paper

Deploy (GitHub Pages)

# from this folder
git init && git add -A && git commit -m "DTop-p MoE project page"
git branch -M main
git remote add origin https://github.com/jincan333/Dtopp.github.io.git
git push -u origin main

Then in the repo: Settings → Pages → Build and deployment → Source: main / root. The site will be served at https://jincan333.github.io/Dtopp.github.io/.

Tip: a repo named simply Dtopp serves a cleaner URL https://jincan333.github.io/Dtopp/ — just change the remote and the og:image/og:url meta in index.html.

About

[ICML2026] DTop-p MoE: Sparsity-Controlled Dynamic Top-p MoE for Foundation Model Pre-training

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages