mindspore/ccsrc/plugin/device/ascend/hal/device/ascend_device_address.cc:924 CopyDeviceToHost

[zhaonx]

Describe the bug/ 问题描述 (Mandatory / 必填)
A clear and concise description of what the bug is.
使用mindnlp调用bge-m3模型，部分报错
ERROR:root:An error occurred: The pointer[GetDevicePtr()] is null.

---

• Framework Unexpected Exception Raised:

---

This exception is caused by framework's unexpected error. Please create an issue at https://gitee.com/mindspore/mindspore/issues to get help.

---

• C++ Call Stack: (For framework developers)

---

mindspore/ccsrc/plugin/device/ascend/hal/device/ascend_device_address.cc:924 CopyDeviceToHost

from mindspore import ops
from mindnlp.transformers import AutoModel, AutoTokenizer

from mindspore.ops._primitive_cache import _get_cache_prim

import mindspore as ms

print(ms.device_context.ascend.is_available)

ms.context.set_context(device_target="Ascend",device_id=1)
import time
import numpy as np
import logging
from enum import Enum
from fastapi import FastAPI, HTTPException, Query, Request
from fastapi.middleware.cors import CORSMiddleware
from fastapi_offline import FastAPIOffline
from pydantic import BaseModel
from typing import List, Optional

logging.basicConfig(level=logging.ERROR)

app = FastAPIOffline()

app.add_middleware(
CORSMiddleware,
allow_origins=[""], # 允许的源列表，使用 [""] 可以允许所有源
allow_credentials=True, # 是否允许发送 Cookie
allow_methods=[""], # 允许的 HTTP 方法，如 ["GET", "POST"]
allow_headers=[""], # 允许的 HTTP 请求头
)

def normalize_numpy(dense_vecs, axis=-1):
"""
Normalize a numpy array along the specified axis using L2 normalization.

Parameters:
dense_vecs (numpy.ndarray): The input array to be normalized.
axis (int): The axis along which to compute the L2 norm. Default is -1.

Returns:
numpy.ndarray: The L2-normalized array.
"""
# norms = np.linalg.norm(dense_vecs, ord=2, axis=axis, keepdims=True)
return dense_vecs / np.linalg.norm(dense_vecs, ord=2, axis=axis, keepdims=True)

tokenizer = AutoTokenizer.from_pretrained('liuyanyi/bge-m3-hf',cache_dir="./cache")

embedding_model = AutoModel.from_pretrained('liuyanyi/bge-m3-hf',cache_dir="./cache")

tokenizer = AutoTokenizer.from_pretrained('/work/cache/model/liuyanyi/bge-m3-hf')
embedding_model = AutoModel.from_pretrained('/work/cache/model/liuyanyi/bge-m3-hf')
@app.post("/embeddings")
def encode_texts(
texts: Texts,
task: Task = Task.text_matching,
truncate_dim: Optional[TruncateDim] = Query(default=None),
max_str_length: Optional[int] = Query(default=None)
):
try:
input_ids = tokenizer(texts.texts[0], return_tensors="ms", padding=True, truncation=True)
output = embedding_model(**input_ids, return_dict=True)
dense_output = normalize_numpy(output.dense_output[0].asnumpy())
return {"code": 200,"embeddings": [dense_output.tolist()] }
except Exception as e:
logging.error(f"An error occurred: {e}")
return {"code": 500,"message": str(e)}

• Hardware Environment(Ascend/GPU/CPU) / 硬件环境:
• Ascend 910b
  使用 swr.cn-central-221.ovaijisuan.com/mindformers/mindformers1.3_mindspore2.4:20241114镜像

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

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

• Excute Mode / 执行模式 (Mandatory / 必填)(PyNative/Graph):

Please delete the mode not involved / 请删除不涉及的模式:
/mode pynative
/mode graph

To Reproduce / 重现步骤 (Mandatory / 必填)
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

Expected behavior / 预期结果 (Mandatory / 必填)
A clear and concise description of what you expected to happen.

Screenshots/ 日志 / 截图 (Mandatory / 必填)
If applicable, add screenshots to help explain your problem.

Additional context / 备注 (Optional / 选填)
Add any other context about the problem here.