chatglm是否支持x86 cpu+910 gpu的9010服务器

[gptq]

If this is your first time, please read our contributor guidelines: https://github.com/mindspore-lab/mindcv/blob/main/CONTRIBUTING.md

Describe the bug/ 问题描述 (Mandatory / 必填) 提示ReduceSumReduceSum错误

• Hardware Environment(Ascend/GPU/CPU) / 硬件环境:

  Please delete the backend not involved / 请删除不涉及的后端: /device ascend/GPU/CPU/kirin/等其他芯片 9010服务器，ntel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz+910b gpu npu-smi info +-------------------------------------------------------------------------------------------+ | npu-smi 22.0.4 Version: 22.0.4 | +----------------------+---------------+----------------------------------------------------+ | NPU Name | Health | Power(W) Temp(C) Hugepages-Usage(page)| | Chip | Bus-Id | AICore(%) Memory-Usage(MB) HBM-Usage(MB) | +======================+===============+====================================================+ | 3 910B | OK | 69.8 42 0 / 0 | | 0 | 0000:3E:00.0 | 0 2782 / 15039 3 / 32768 | +======================+===============+====================================================+

• Software Environment / 软件环境 (Mandatory / 必填): -- MindSpore version (e.g., 1.7.0.Bxxx) :2.0.0 -- Python version (e.g., Python 3.7.5) :3..75 -- OS platform and distribution (e.g., Linux Ubuntu 16.04):centos 7.6 -- GCC/Compiler version (if compiled from source):

To Reproduce / 重现步骤 (Mandatory / 必填) Steps to reproduce the behavior: python3 -m pytest tests/ut/models/glm/test_modeling_chatglm.py

Screenshots/ 日志 / 截图 (Mandatory / 必填)

E TypeError: E ---------------------------------------------------- E - Kernel select failed: E ---------------------------------------------------- E Select CPU operator[ReduceSum] fail! Unsupported data type! E The supported data types are input[Float32 Int32], output[Float32]; input[Float32 Int64], output[Float32]; input[Float64 Int32], output[Float64]; input[Float64 Int64], output[Float64]; input[Int8 Int32], output[Int8]; input[Int8 Int64], output[Int8]; input[Int16 Int32], output[Int16]; input[Int16 Int64], output[Int16]; input[Int32 Int32], output[Int32]; input[Int32 Int64], output[Int32]; input[Int64 Int32], output[Int64]; input[Int64 Int64], output[Int64]; input[UInt8 Int32], output[UInt8]; input[UInt8 Int64], output[UInt8]; input[UInt16 Int32], output[UInt16]; input[UInt16 Int64], output[UInt16]; input[UInt32 Int32], output[UInt32]; input[UInt32 Int64], output[UInt32]; input[UInt64 Int32], output[UInt64]; input[UInt64 Int64], output[UInt64]; input[Complex64 Int32], output[Complex64]; input[Complex64 Int64], output[Complex64]; input[Complex128 Int32], output[Complex128]; input[Complex128 Int64], output[Complex128]; , but get input[Bool Int64 ] and output[Bool ] E node: @kernel_graph_5:[CNode]1{[0]: ValueNode ReduceSum, [1]: @kernel_graph_5:[Parameter]2, [2]: ValueNode Tensor(shape=[1], dtype=Int64, value=[0])} E
E ---------------------------------------------------- E - C++ Call Stack: (For framework developers) E ---------------------------------------------------- E mindspore/ccsrc/plugin/device/cpu/hal/hardware/cpu_device_context.cc:340 SetOperatorInfo

/usr/local/python3.7.5/lib/python3.7/site-packages/mindspore/common/_stub_tensor.py:150: TypeError

[lvyufeng]

看其实是reduce sum算子的问题，我修一下

[gptq]

是啊。多谢老大。

[lvyufeng]

是啊。多谢老大。

已支持，在启智的1.10环境已测试。