Unverified50% confidenceFactExact time
BEINGS应用早期使用temperature 1.05、top_p 0.95的采样参数，导致模型出现循环输出等异常行为。
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/17/2026
First Seen
Valid until: 1/15/2027
Sources
Related Entities
Related Claims
Unverified动态温度调节策略：正常推理阶段Temperature维持在0.1~0.3，检测到轨迹重复时提升至0.7~1.0，Top-P从0.9提升至0.9575% similarUnverified在LLM层面微量随机因子对应调高采样温度（Temperature）或Top-P参数，工程做法是检测到轨迹重复时将Temperature从默认值（如0.2）动态提升至0.7~1.074% similarUnverifiedWhen testing citation behavior, temperature should be set to 0 or a low value with the top-p parameter fixed, and the same prompt should be run at least 20–50 times to use statistical distributions instead of single results.73% similarUnverified自洽性做法是对同一问题进行多次高温采样（temperature通常设为0.7-1.0），再用多数投票选出频率最高的结果72% similarUnverifiedIn testing scenarios, it is generally necessary to set the Temperature to a low value (e.g., 0.1–0.2) combined with a fixed random seed to ensure output reproducibility.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1051528API
curl https://kongchang.com/api/v1/knowledge/claims/1051528MCP
get_claim(id=1051528)