Unverified50% confidenceFactExact time
Decision models are less reliable than LLMs as hard-label detectors of implanted false sources, but their output probabilities can still distinguish implanted sources from clean sources.
1
Sources
50%
Confidence
Long-term
Relevance
10/28/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedUnder domain mismatch conditions, relying solely on confidence-based filtering is insufficient to guarantee pseudo-label quality, because models may also exhibit high confidence on incorrect samples.76% similarUnverifiedLLMs have a strong tendency toward error tolerance and completion, and when faced with misaligned table data they typically do not report errors but instead produce seemingly reasonable inferences based on the erroneous data.75% similarUnverifiedLLM幻觉问题源于模型基于概率预测生成文本而非检索真实数据，在涉及具体统计数字、引文来源和法律条款时尤为危险72% similarUnverifiedThe risk of pseudo-labels lies in confirmation bias—erroneous predictions get reinforced; therefore in practice novelty detection (e.g., Mahalanobis distance) is usually combined to judge whether the distribution has drifted, and pseudo-label updates are enabled only when the drift is significant and prediction confidence is high.72% similarUnverifiedInference methods based on model internal features face a trade-off between false positives and false negatives71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1106320API
curl https://kongchang.com/api/v1/knowledge/claims/1106320MCP
get_claim(id=1106320)