Unverified50% confidenceOpinionExact time
When evaluating the scientific capabilities of LLMs, strict contamination screening must be performed, and the two mechanisms of "learning" and "memorization" must be distinguished.
1
Sources
50%
Confidence
Long-term
Relevance
11/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn the experimental design, having the LLM decide 'whether to give a hint' would introduce nondeterminism and directly contaminate the independent variable, making the timing of the help learners receive irreproducible.72% similarVerifiedData contamination refers to test problems having already appeared in pretraining corpora, causing a model's high scores to more closely reflect memorization rather than reasoning.66% similarUnverifiedThe dependency relationship questions that an LLM memory system needs to answer are formally highly consistent with data flow analysis and dependency analysis in program analysis.65% similarUnverifiedThe first step in evaluating an LLM should be to clarify what is to be measured, rather than running the model directly64% similarUnverifiedA study systematically evaluated the ability of 7 frontier LLMs to predict molecular properties in in-context learning and paid particular attention to whether the models' self-reported uncertainty correlates with prediction errors.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1124964API
curl https://kongchang.com/api/v1/knowledge/claims/1124964MCP
get_claim(id=1124964)