Originally created by: yeqing1a
Originally owned by: igophper
例行检查
问题描述
sealos部署,qwen-long模型,需要稍长时间的回答很容易被截断,报:finish_reason: null
考虑能否自定义改非steam输出,或者是否在qwen-long模型的兼容上存在问题
oneapi日志:[SYS] 2024/06/11 - 10:18:59 | batch update finished
[GIN] 2024/06/11 - 10:19:05 | 2024061110183013490583053930793 | 200 | 35.366610817s | 101.126.13.246 | POST /v1/chat/completions
[INFO] 2024/06/11 - 10:19:05 | 2024061110183013490583053930793 | record consume log: userId=1, channelId=2, promptTokens=86688, completionTokens=392, modelName=qwen-long, tokenName=timai, quota=124403, content=模型倍率 1.43,分组倍率 1.00,补全倍率 1.00
[SYS] 2024/06/11 - 10:19:59 | batch update started
[SYS] 2024/06/11 - 10:19:59 | batch update finished
复现步骤
预期结果
相关截图
Originally posted by: igophper
能否给出one-api输出的响应