Expired50% confidenceBenchmarkExact time
Multimodal large models have extremely low factual recall accuracy in video question answering, which is difficult to improve even with supervised fine-tuning.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/17/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 10/15/2026(expired)
Sources
Related Claims
UnverifiedGPT models perform poorly when recall ability is tested in isolation, correctly recalling only about 50% of facts across all categories including spatial information, transcription content, and physical experiments, with open-source models performing even worse.73% similarUnverified在事实性评估任务上模型评判者准确率仍显著低于人类专家，但在风格、连贯性等主观维度已接近人类评审员间一致性水平71% similarUnverified斯坦福大学研究发现，在推理链关键步骤注入错误信息对最终答案的影响远小于预期70% similarUnverifiedResearch shows that a model's accuracy on self-criticism tasks is significantly lower than in scenarios where criticism is performed by another model.69% similarUnverifiedSimply watching videos without hands-on practice typically results in a knowledge retention rate of no more than 10%.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/543116API
curl https://kongchang.com/api/v1/knowledge/claims/543116MCP
get_claim(id=543116)