infrastructure
The Visual Tokens Stayed. Half the Work Disappeared.
Patch-to-Prune bypasses selected visual computations while retaining sequence length, token order and pretrained weights.
Summary
Patch-to-Prune bypasses selected visual computations while retaining sequence length, token order and pretrained weights.
Patch-to-Prune turns activation patching into an inference procedure. Validation-guided sweeps find decoder regions where fixed proxy activations can replace visual-token projection outputs within an accuracy tolerance. The method keeps token positions, attention masks and residual pathways intact. Across four Qwen2.5-VL and LLaVA models and seven multimodal benchmarks, the authors report about 94 percent of dense-model accuracy with 55 percent fewer FLOPs at a three percent tolerance setting. FLOP savings are an evaluated computation measure, not a promised wall-clock speedup on every device.
Why it matters
Patch-to-Prune bypasses selected visual computations while retaining sequence length, token order and pretrained weights.
Limits and context
- FLOP savings are an evaluated computation measure, not a promised wall-clock speedup on every device.
Key claims
Patch-to-Prune bypasses selected visual computations while retaining sequence length, token order and pretrained weights.
Qualification: FLOP savings are an evaluated computation measure, not a promised wall-clock speedup on every device.
Evidence: source-2026-10-04-008
Sources
- arXiv preprint 2610.03389, original v1arXiv / submitting research authors · primary research
Corrections
: Publication note: This October 4 edition was reconstructed on October 5 after the workstation was offline. It uses original October 2 submissions from the standard Sunday-evening arXiv release, rather than claiming those papers were available for the missed 9 AM run. The date and 9 AM publication slot preserve archive chronology; October 5 developments are excluded. The September 27 Invention Desk is carried over because its replacement was generated October 5.