Happy to announce that our paper "IPA: An Information-Reconstructive Input Projection Framework for Efficient Foundation Model Adaptation" got the best paper award at the #CCFM workshop at #NeurIPS2025
Many thanks to the reviewers and organizers.
Kudos to @yuanyinnn.bsky.social & team!
This was a fun project led by @yuanyinnn.bsky.social and awarded as outstanding paper at the NeurIPS #CCFM workshop!
We fermented our thoughts on understanding LoRA & ended up with IPA🍺
We found an asymmetry in LoRA: during training, A changes little & B eats most task-specific adaptation.
So we pre-train A to preserve information before adaptation w/ excellent parameter efficiency #NeurIPS2025 #CCFM 👇
1/Serve your PEFT with a fresh IPA!🍺
Finetuning large models is cheaper thanks to LoRA, but is its random init optimal?🤔
Meet IPA: a feature-aware alternative to random projections
#NeurIPS2025 WS #CCFM Oral+Best Paper
Work w/
S. Venkataramanan @tuanhungvu.bsky.social @abursuc.bsky.social M. Cord
🧵
Chance-Constrained Flow Matching Improves Constraint-Aware Generation
Chance‑Constrained Flow Matching is a training‑free method that works with pre‑trained generative models and guarantees feasibility; tests on PDE systems and molecular docking showed feasible samples. getnews.me/chance-constrained-flow-... #ccfm #ai
L’artiste Solange Roy présente une exposition de céramiques à la galerie d’art du #CCFM #francophonie #Winnipeg #art
www.la-liberte.ca/2025/03/17/u...
