Researchers Release ExtremeWhenBench, First Hour-Scale Video Temporal Grounding Benchmark
Researchers have released ExtremeWhenBench, a benchmark of 2,273 natural-language queries over 194 hour-scale videos, finding that current Video-LLMs largely fail at temporal grounding in long-form video. The study argues the core bottleneck is search—finding the relevant region of a long video—rather than recognition of nearby events. The findings suggest that retrieve-then-ground hybrid approaches, analogous to retrieve-then-read in text QA, offer a 6.7x performance improvement over monolithic Video-LLMs.
A new arXiv preprint introduces ExtremeWhenBench, the first open benchmark specifically designed for hour-scale temporal grounding in video, comprising 2,273 queries over 194 videos with a mean length of 75.7 minutes and a maximum of 9 hours. Temporal grounding—identifying the precise time interval corresponding to a natural-language query—is a fundamental capability for interacting with long-form video, but prior research has focused predominantly on short clips. The authors argue that at hour-scale, the dominant challenge shifts from event recognition to search: models must locate the relevant segment within a vast temporal space before they can localize it precisely. Experiments show that every tested open Video-LLM collapses on this benchmark, while a simple frame-level retrieval baseline outperforms them all. A failure taxonomy attributes 85% of model errors specifically to the search phase rather than recognition. A retrieve-then-ground hybrid system recovers 6.7x performance over monolithic Video-LLMs, drawing a direct parallel to retrieve-then-read pipelines in open-domain question answering. The benchmark and code are publicly released to facilitate further research.
What's missing
It is unclear how the retrieve-then-ground hybrid performs relative to human baselines, and whether the benchmark queries were validated for ambiguity or annotator agreement. As a preprint, the work has not yet undergone peer review.
What different sources said
- arXiv cs.AICenter
Natural-Language Temporal Grounding in Hour-Long Videos is a Search Problem: A Benchmark and Empirical Decomposition
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.