Papers/2611.25069
🧪 Test?View on arXiv

VisAlloc: Allocating Visual Supervision for On-Policy Self-Distillation

Kaelyn, Author2, Author3, Author4, Author5

self-distillationvisual reasoningmultimodalsupervision
2611.25069
Builder Relevance
70%
2d ago

Abstract

VisAlloc introduces a dual-level framework for improving visual reasoning in multimodal large language models through optimized visual supervision.

Reality Card

Core Claim

VisAlloc outperforms existing methods by achieving macro-averages of 81.44 and 84.92 on eight benchmarks with its 4B and 9B models, respectively.

Method / Result

Achieved a performance improvement of 0.97 and 0.52 points over Vision-OPD reproductions.

Limitations

The reliance on specific paired-teacher filters may limit generalizability to other datasets or tasks.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers