BuildTaskList fail for 12 not supported, Only TBE, AI_CPU, CUST_AI_CPU kernel are supported

**情况描述** 
使用百川2的原始项目跑web_demo，启动成功进入界面。
模型加载完毕，npu info能看到进程和显存使用
在界面输入文字后回车，等几十秒后出现下面报错。

**报错日志：**

```
Exception in thread Thread-24 (generate):
Traceback (most recent call last):
  File "/root/miniconda3/envs/llf/lib/python3.10/threading.py", line 1016, in _bootstrap_inner
    self.run()
  File "/root/miniconda3/envs/llf/lib/python3.10/threading.py", line 953, in run
    self._target(*self._args, **self._kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/transformers/generation/utils.py", line 1622, in generate
    result = self._sample(
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/transformers/generation/utils.py", line 2791, in _sample
    outputs = self(
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
    return forward_call(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/accelerate/hooks.py", line 166, in new_forward
    output = module._old_forward(*args, **kwargs)
  File "/root/.cache/huggingface/modules/transformers_modules/modeling_baichuan.py", line 686, in forward
    outputs = self.model(
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
    return forward_call(*args, **kwargs)
  File "/root/.cache/huggingface/modules/transformers_modules/modeling_baichuan.py", line 461, in forward
    layer_outputs = decoder_layer(
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
    return forward_call(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/accelerate/hooks.py", line 166, in new_forward
    output = module._old_forward(*args, **kwargs)
  File "/root/.cache/huggingface/modules/transformers_modules/modeling_baichuan.py", line 286, in forward
    hidden_states = self.mlp(hidden_states)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
    return forward_call(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/accelerate/hooks.py", line 166, in new_forward
    output = module._old_forward(*args, **kwargs)
  File "/root/.cache/huggingface/modules/transformers_modules/modeling_baichuan.py", line 169, in forward
    return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
    return forward_call(*args, **kwargs)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/modules/activation.py", line 393, in forward
    return F.silu(input, inplace=self.inplace)
  File "/root/miniconda3/envs/llf/lib/python3.10/site-packages/torch/nn/functional.py", line 2072, in silu
    return torch._C._nn.silu(input)
RuntimeError: InnerRun:torch_npu/csrc/framework/OpParamMaker.cpp:197 NPU error, error code is 145018
[ERROR] 2024-04-25-14:01:58 (PID:2287093, Device:0, RankID:-1) ERR01100 OPS call acl api failed.
E19999: Inner Error!
E19999  BuildTaskList fail for 12 not supported, Only TBE, AI_CPU, CUST_AI_CPU kernel are supported.[FUNC:BuildKernelAndExTask][FILE:single_op_model.cc][LINE:464]
        TraceBack (most recent call last):
        [Load][Op]Load operator failed. model = 281450520691424, ge result = 145018[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
        [Load][Op]LoadSingleOp failed or singleOp is nullptrret = 145018[FUNC:ReportInnerError][FILE:log_inner.cpp][LINE:145]
```

**环境：** 
python：3.10.14
NPU：Atlas 300I DUO  * 2 （两块）
Driver：Ascend-hdk-310p-npu-driver_23.0.1_linux-aarch64
Firmware：Ascend-hdk-310p-npu-firmware_7.1.0.4.220
CANN-toolkit：8.0.RC1
CANN-kernels：310p_8.0.RC1
Torch：2.1.0
Torch-npu:2.1.0.post3

**使用的百川2原始项目地址** 
地址：https://github.com/baichuan-inc/Baichuan2
使用streamlit run web_demo.py启动
模型使用的百川2-7B-chat

Ascend/pytorch
暂停

内容风险标识

评论 (7)

Ascend/pytorch暂停 .gitee-modal { width: 500px !important; }

内容风险标识