Benchmark Almanac

Hong Kong University of Science and Technology

We track 10 benchmarks that Hong Kong University of Science and Technology made or helped make.

0 in use 8 with public code 6 with a leaderboard here

Benchmarks

Name Kind Released Best score Status
UGMathBench Models solve 5,062 undergraduate math problems across 16 subjects. Each problem has three randomized versions, and effective accuracy counts problems solved in all versions. Word problems Jan 2025 81.9% System cards only
MirageBench Models personalize six kinds of responses from three disclosed facts about each of 150 personas. A model judge classifies claims by evidential support. The primary over-inference rate counts stereotyped and fabricated claims. Faithful to sources Aug 2026 35.1% Not in cards
MarineEval Models answer 2,000 questions about marine images and documents. The authors combine answer checks with model judgments. They report total accuracy and mean accuracy across seven tasks. General visual questions Nov 2025 50.4% Not in cards
MarineEVT Models answer 2,000 held-out questions about underwater events, including species interactions, movement, and timing. The authors report accuracy across five reasoning categories. Video understanding Jun 2026 43.7% Not in cards
AssertLLM2 Models write assertions for hardware specifications. A formal verifier checks which assertions hold, which logic they cover, and which deliberately introduced faults they detect. Electronics May 2026 95% Not in cards
RTL-OPT Models rewrite Verilog to improve a circuit while preserving its behavior. The authors compare synthesized power, timing, and area against human reference designs. Electronics 72.5% Not in cards
EvalVerse Video models generate cinematic clips from prompts and reference material derived from professional videos. Film experts compare outputs across visual design, acting, cinematography, aesthetics, emotion, shot sequences, and sound. A calibrated vision-language evaluator also estimates these judgments. Video generation May 2026 0.73 Not in cards
MultiHop-RAG Retrieval-augmented systems answer 2,556 queries whose evidence is spread across two to four news articles. The queries also use document metadata, such as source and date. The benchmark scores both retrieval quality and answer accuracy. Questions that need search Jan 2024 56% Not in cards
PCB-Bench Models answer placement and routing questions and describe circuit boards from editor screenshots. The authors report choice accuracy and separate measures for free-form answers. Electronics Not in cards
WorldOlympiad Models generate long videos for gaming, robotics, and everyday scenes. The evaluator checks physical rules, reconstructed geometry, and instruction following across video segments. Video world models Jun 2026 Not in cards