SF20K competition results show top AI models reach 65.7% accuracy on story-level video QA using amateur films, with information selection and reasoning structure as the main bottlenecks rather than model size, well below human performance of 91.7%.
Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities.arXiv, 2025
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SF20K Competition 2025: Summary and findings
SF20K competition results show top AI models reach 65.7% accuracy on story-level video QA using amateur films, with information selection and reasoning structure as the main bottlenecks rather than model size, well below human performance of 91.7%.