参考ACL_PyTorch/built-in/audio/CosyVoice/CosyVoice2/README.md · Ascend/ModelZoo-PyTorch - Gitee.com部署
300I Duo卡
CANN版本:8.0.RC3
执行推理报错信息如下:
Removing weight norm...
warm up start
第1/2轮 warmup:↓↓↓
curr prompt text:CosyVoice是一款基于语音量化编码的语音生成大模型,能够深度融合文本理解和语音生成,实现自然流畅的语音体验。它通过离散化编码和依托大模型技术,能够精准解析并诠释各类文本内容,将其转化为宛如真人般的自然语音。CosyVoice2在原始1的基础上,把QWEN2模型接入CosyVoice的LLM部分,实现了推理加速
.Traceback (most recent call last):
File "/usr/local/lib64/python3.11/site-packages/torch_npu/dynamo/torchair/_utils/error_code.py", line 43, in wapper
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib64/python3.11/site-packages/torch_npu/dynamo/torchair/core/_backend.py", line 133, in run
return super(TorchNpuGraph, self).run((inputs, assigned_outputs, stream))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: EZ1001: [PID: 133038] 2025-07-02-16:45:34.342.301 PromptFlashAttention LaunchAicore failed.
TraceBack (most recent call last):
attention mask must be NULL, when Qs,Kvs is unAlign or Qs is not equal to Kvs, Qs = 6, Kvs = 6[FUNC:RunBigKernelTilingWithParams][FILE:prompt_flash_attention_tiling.cpp][LINE:3299]
Tiling failed
Tiling Failed.
Kernel GetWorkspace failed. opType: 15
PromptFlashAttention LaunchAicore failed.
call aclnnPromptFlashAttentionV3 failed:[FUNC:PromptHostExecuteFunc][FILE:fallback_prompt_flash_attention.cpp][LINE:116]
api_ret faild:4294967295[FUNC:PromptHostExecuteFunc][FILE:fallback_prompt_flash_attention.cpp][LINE:119]
Assert ((op_execute_func(op_execute_context)) == ge::SUCCESS) failed[FUNC:ExecuteOpFunc][FILE:aclnn_op_execute_kernel.cc][LINE:47]
Failed to execute rt v2 model for graph graph_1_0, model_id 4.[FUNC:ExecuteWithStreamAsync][FILE:hybrid_model_rt_v2_executor.cc][LINE:1072]
GraphManager ExecuteGrapWithStreamhAsync failed,session id = 1, graph id = 0, stream = 0xaaaefa265660.[FUNC:ExecuteGraphWithStreamAsync][FILE:inner_session.cc][LINE:576]
Execute graph with stream async failed, error code:1343225857, session_id:1, graph_id:0, stream:0xaaaefa265660.[FUNC:ExecuteGraphWithStreamAsync][FILE:ge_api.cc][LINE:844]
参考ACL_PyTorch/built-in/audio/CosyVoice/CosyVoice2/README.md · Ascend/ModelZoo-PyTorch - Gitee.com部署
300I Duo卡
CANN版本:8.0.RC3
执行推理报错信息如下:
Removing weight norm...
warm up start
第1/2轮 warmup:↓↓↓
curr prompt text:CosyVoice是一款基于语音量化编码的语音生成大模型,能够深度融合文本理解和语音生成,实现自然流畅的语音体验。它通过离散化编码和依托大模型技术,能够精准解析并诠释各类文本内容,将其转化为宛如真人般的自然语音。CosyVoice2在原始1的基础上,把QWEN2模型接入CosyVoice的LLM部分,实现了推理加速
.Traceback (most recent call last):
File "/usr/local/lib64/python3.11/site-packages/torch_npu/dynamo/torchair/_utils/error_code.py", line 43, in wapper
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib64/python3.11/site-packages/torch_npu/dynamo/torchair/core/_backend.py", line 133, in run
return super(TorchNpuGraph, self).run((inputs, assigned_outputs, stream))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: EZ1001: [PID: 133038] 2025-07-02-16:45:34.342.301 PromptFlashAttention LaunchAicore failed.
TraceBack (most recent call last):
attention mask must be NULL, when Qs,Kvs is unAlign or Qs is not equal to Kvs, Qs = 6, Kvs = 6[FUNC:RunBigKernelTilingWithParams][FILE:prompt_flash_attention_tiling.cpp][LINE:3299]
Tiling failed
Tiling Failed.
Kernel GetWorkspace failed. opType: 15
PromptFlashAttention LaunchAicore failed.
call aclnnPromptFlashAttentionV3 failed:[FUNC:PromptHostExecuteFunc][FILE:fallback_prompt_flash_attention.cpp][LINE:116]
api_ret faild:4294967295[FUNC:PromptHostExecuteFunc][FILE:fallback_prompt_flash_attention.cpp][LINE:119]
Assert ((op_execute_func(op_execute_context)) == ge::SUCCESS) failed[FUNC:ExecuteOpFunc][FILE:aclnn_op_execute_kernel.cc][LINE:47]
Failed to execute rt v2 model for graph graph_1_0, model_id 4.[FUNC:ExecuteWithStreamAsync][FILE:hybrid_model_rt_v2_executor.cc][LINE:1072]
GraphManager ExecuteGrapWithStreamhAsync failed,session id = 1, graph id = 0, stream = 0xaaaefa265660.[FUNC:ExecuteGraphWithStreamAsync][FILE:inner_session.cc][LINE:576]
Execute graph with stream async failed, error code:1343225857, session_id:1, graph_id:0, stream:0xaaaefa265660.[FUNC:ExecuteGraphWithStreamAsync][FILE:ge_api.cc][LINE:844]