QwQ-32B输出不完整:

对比:deepseek-32B输出这么多tokens也不会卡死:

config.json配置:
{
"maxSeqLen" : 24576,
"maxInputTokenLen" : 16384,
"truncation" : false,
"ModelConfig" : [
{
"modelInstanceType" : "Standard",
"modelName" : "qwen",
"modelWeightPath" : "/ai/QwQ-32B-w8a8",
"worldSize" : 2,
"cpuMemSize" : 5,
"npuMemSize" : -1,
"backendType" : "atb",
"trustRemoteCode" : false
}
]
},
"ScheduleConfig" :
{
"templateType" : "Standard",
"templateName" : "Standard_LLM",
"cacheBlockSize" : 128,
"maxPrefillBatchSize" : 50,
"maxPrefillTokens" : 24576,
"prefillTimeMsPerReq" : 150,
"prefillPolicyType" : 0,
"decodeTimeMsPerReq" : 50,
"decodePolicyType" : 0,
"maxBatchSize" : 200,
"maxIterTimes" : 8192,
"maxPreemptCount" : 0,
"supportSelectBatch" : false,
"maxQueueDelayMicroseconds" : 5000
}
QwQ-32B输出不完整:
对比:deepseek-32B输出这么多tokens也不会卡死:
config.json配置:
{
"maxSeqLen" : 24576,
"maxInputTokenLen" : 16384,
"truncation" : false,
"ModelConfig" : [
{
"modelInstanceType" : "Standard",
"modelName" : "qwen",
"modelWeightPath" : "/ai/QwQ-32B-w8a8",
"worldSize" : 2,
"cpuMemSize" : 5,
"npuMemSize" : -1,
"backendType" : "atb",
"trustRemoteCode" : false
}
]
},
"ScheduleConfig" :
{
"templateType" : "Standard",
"templateName" : "Standard_LLM",
"cacheBlockSize" : 128,
"maxPrefillBatchSize" : 50,
"maxPrefillTokens" : 24576,
"prefillTimeMsPerReq" : 150,
"prefillPolicyType" : 0,
"decodeTimeMsPerReq" : 50,
"decodePolicyType" : 0,
"maxBatchSize" : 200,
"maxIterTimes" : 8192,
"maxPreemptCount" : 0,
"supportSelectBatch" : false,
"maxQueueDelayMicroseconds" : 5000
}