Unverified50% confidenceOpinionExact time
Evaluation in financial scenarios cannot look only at text fluency; it must test the model's performance on number extraction, computational consistency, and the completeness of the logical chain.
1
Sources
50%
Confidence
Long-term
Relevance
11/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe correct way to handle non-numeric salary fields is to parse what can be parsed, flag what cannot be parsed, and let the caller decide, because a string representing salary opacity is itself an important signal for market analysis.65% similarUnverified金融数据的非平稳性意味着消融结论在一个时段成立未必在另一时段同样有效,这是金融AI消融实验的独特挑战65% similarUnverifiedText-level monitors can only analyze a model's explicit outputs and cannot detect cases where the model stays clean at the output level but retains inconsistent objectives internally.64% similarUnverifiedUnder injection bait, weak models treat the injected incorrect numbers as their own calculation results and fabricate verification processes out of thin air, with the explanatory text appearing consistent with genuine verification, making it impossible to detect errors through auditing.64% similarUnverifiedVector retrieval excels at fuzzy semantic matching but is insensitive to precise conditions (such as time ranges, user IDs, and status enum values), and its recall results lack explainable filtering criteria.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1126359API
curl https://kongchang.com/api/v1/knowledge/claims/1126359MCP
get_claim(id=1126359)