Benchmark Almanac

Shenzhen

We track 8 benchmarks that Shenzhen made or helped make.

0 in use 5 with public code 3 with a leaderboard here

Benchmarks

Name Kind Released Best score Status
MatCha Models answer multiple-choice questions about material structure, shape and properties using images from materials characterization instruments. Lab and data work Sep 2025 68.1% Not in cards
Bank Question Corpus Models determine whether two Chinese customer service questions ask for the same thing. The authors reserved 10,000 labeled pairs for testing. Finance Sep 2018 0.76 Not in cards
InfiHall-Bench Models generate images from 411 prompts that combine visual requirements. The evaluator checks details such as object parts, actions, and spatial relations separately. Image generation and editing Nov 2026 67.1 Not in cards
SPA-Bench Agents use Android apps such as Airbnb, Spotify, and WeChat. The grader matches screenshots against task requirements and uses a vision model to resolve uncertain cases. Phones Oct 2024 1.35% Not in cards
Chinese Classical Poetry Appreciation 2025 Models explain words and lines from classical Chinese poems and identify the emotions they express. The test covers 327 poems. Reading in many languages Jan 2025 Not in cards
Chinese Classical Poetry Appreciation 2026 Models explain words, lines, and allusions in classical Chinese poems. They also complete analogies and choose among interpretations. Reading in many languages Feb 2026 Not in cards
Cross-lingual Sentiment Consistency Analysis for Classical Chinese Literature Models predict emotions in classical Chinese passages and their English translations. The test checks how accurately each prediction captures the passage and how closely the predictions agree. Reading in many languages Feb 2026 Not in cards
LP-TTS-Bench Models synthesize speech from extended recordings of a speaker. The benchmark measures how well they reproduce timing, emphasis, and intonation on 2,728 held-out utterances. Speech recognition and translation Oct 2026 Not in cards