Unverified50% confidenceFactExact time
The green-list/red-list watermarking scheme is training-free, directly intervening in the sampling logic of the inference stage without modifying model weights.
1
Sources
50%
Confidence
Long-term
Relevance
9/26/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe prompts submitted to the models did not include question-type hints, requiring the models to determine the question type on their own.67% similarUnverifiedWhen parsing model output, the parsing logic should be wrapped with try...except, and an empty list should be returned if parsing fails, to ensure the program does not crash.66% similarUnverifiedDuring the inference phase, only forward propagation needs to be executed; gradient computation and backpropagation are not required.65% similarUnverifiedThe thin adapter layer strategy keeps core task logic unchanged and only adjusts the interface portion that interacts with the model, designing pre-prompts or post-processing logic tailored to the reasoning characteristics of different models.64% similarUnverifiedReAct triggers actual tool calls after each reasoning step, whereas Chain-of-Thought prompting only performs reasoning without calling tools.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/951839API
curl https://kongchang.com/api/v1/knowledge/claims/951839MCP
get_claim(id=951839)