Study Finds Greedy Decoding Superior to Stochastic Sampling for Visual Question Answering Tasks
A new arXiv preprint contends that greedy decoding — selecting the single most probable token at each step — is superior to stochastic sampling strategies for Visual Question Answering (VQA) tasks in Multimodal Large Language Models. The authors provide a theoretical framework linking model calibration to predictive accuracy and derive conditions under which greedy decoding is optimal, supporting their claims with experiments across multiple benchmarks. The findings challenge the common practice of inheriting text-focused LLM decoding defaults in multimodal systems without task-specific justification.
Researchers have published a preprint on arXiv arguing that stochastic sampling strategies — widely used in large language models (LLMs) to balance coherence and diversity — are often suboptimally carried over into Multimodal LLMs (MLLMs) for Visual Question Answering tasks. VQA is characterized as a closed-ended task with head-heavy answer distributions, where uncertainty tends to be epistemic in nature, stemming from missing or ambiguous visual evidence rather than from genuinely plausible alternative continuations. The paper formalizes the relationship between model calibration and predictive accuracy theoretically, deriving sufficient conditions under which greedy decoding is the optimal strategy. Empirical experiments across multiple benchmarks support the claim that greedy decoding consistently outperforms stochastic sampling in this setting. The authors also introduce a variant called Greedy Decoding for Reasoning Models, which they report surpasses both stochastic sampling and standard greedy decoding in multimodal reasoning scenarios. The work cautions the AI research community against uncritically inheriting LLM decoding heuristics when deploying models in multimodal contexts. The paper is a preprint and has not yet undergone formal peer review.
What's missing
The paper is an arXiv preprint and has not yet been peer-reviewed, so its theoretical claims and empirical results have not been independently validated. It is also unclear how the proposed Greedy Decoding for Reasoning Models performs relative to other inference-time optimization techniques beyond those tested.
What different sources said
- arXiv cs.CLCenter
Revisiting Greedy Decoding for Visual Question Answering: A Calibration Perspective
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.