Unverified50% confidenceFactExact time
The study defines factual information that appears only once in the training corpus as a "monofact".
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThrough further experiments, the study confirms that the positive correlation between monofacts and discovery capability cannot be explained by training set diversity alone.70% similarUnverifiedThe study proposes that when constructing training sets, in addition to size, quality, and diversity, the "monofact rate" metric should also be considered.69% similarUnverifiedFacts in pretrained models can be divided into two types, active facts and inert facts; active facts can support generalization beyond the memorized range after fine-tuning, while inert facts can be recalled but cannot produce generalization on the same task.68% similarUnverifiedThe study conducted experiments in materials science and regulatory genomics by directly manipulating the proportion of monofacts in the training set.64% similarUnverifiedThe analysis finds that the true causal evidence in training documents is highly localized, with only a few key tokens carrying the factual association.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1130238API
curl https://kongchang.com/api/v1/knowledge/claims/1130238MCP
get_claim(id=1130238)