Unverified50% confidenceFactExact time
Existing LLM-based automatic heuristic design (AHD) frameworks either freeze the generator or fine-tune the model using only a scalar fitness signal, ignoring the key information of which problem instances each heuristic covers.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIf LLM judges that are blind to omissions are widely used for scoring, benchmarking, or guiding model training (e.g., RLHF, RLAIF), the entire optimization loop may be misled, causing models to learn to generate content that appears complete but is actually incomplete.73% similarUnverifiedFlyte 支持条件逻辑，可仅当模型评估指标超过预设阈值时才自动触发部署流程72% similarUnverified大模型应用开发（LLM Application）偏向工程化落地，不需要从零推导Transformer数学公式或预训练百亿参数模型，而是调用、编排和微调已有大模型71% similarUnverifiedPrompt injection stems from a fundamental characteristic of current LLM architectures, rather than a vulnerability that can simply be patched.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023789API
curl https://kongchang.com/api/v1/knowledge/claims/1023789MCP
get_claim(id=1023789)