Unverified50% confidenceBenchmarkExact time
When rewriting with temperature 0 (greedy decoding), 12.7% of the original 5-grams were retained (baseline 10.4%), and the flagging rate rose from 25% to 31%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/14/2026
First Seen
Valid until: 1/12/2027
Sources
Related Entities
Related Claims
UnverifiedOn one task, the proportion of rows that change operator output under temperature 0 reached as high as 24.0%74% similarUnverifiedEven under temperature 0 conditions, 0.07% to 10.2% of rows across five executions still change their operator output, with the exact proportion depending on the model73% similarUnverifiedAmong all 500 rewrites that were scored, the correlation coefficient between 5-gram retention and meaning score was only 0.15.64% similarUnverifiedBecause the experiment used temperature=0 and a fixed seed in a single deterministic run, any field flip moves the score by about 2.8 percentage points out of 12 tasks × 3 fields = 36 units, so the gap between 91.7% and 83.3% cannot be directly generalized into a statistically meaningful general advantage.64% similarUnverifiedWhen responses were frozen at the initial policy, increasing prompt breadth pulled accuracy down from 21.16% to 19.05%.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1029973API
curl https://kongchang.com/api/v1/knowledge/claims/1029973MCP
get_claim(id=1029973)