Unverified50% confidenceFactExact time
Smaller models produce hallucinations when generating journals, such as fabricating information not present in the detection results or misreading 16 audio windows as having heard the bird 16 times or 16 individual birds.
1
Sources
50%
Confidence
Long-term
Relevance
10/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCatastrophic transcription loops are a known defect of autoregressive speech models such as Whisper, where the model may fall into an infinite loop of repeatedly generating the same segment when its confidence is insufficient.62% similarUnverifiedIf the specified duration exceeds the actual amount of content provided, the model will fill the gap with meaningless noise.60% similarVerifiedThe three main causes of hallucinations in large models are training data noise, the probability-first mechanism, and insufficient data timeliness.60% similarUnverifiedThe paper introduces rate-distortion theory into hallucination analysis, pointing out that even with perfect data, insufficient bits can lead to memory confusion.60% similarUnverifiedTypical generation-stage failures include hallucinations caused by ignoring context, middle-content forgetting due to context overflow, insufficient multi-hop reasoning capability, and instruction-following deviation.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/998441API
curl https://kongchang.com/api/v1/knowledge/claims/998441MCP
get_claim(id=998441)