### Initial MindOCR Setup on Windows

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/frequently_asked_questions.md

This bash snippet outlines the standard procedure for cloning the MindOCR repository and attempting to install it locally using pip. It serves as the starting point before encountering dependency errors.

```bash
git clone git@gitee.com:mindspore-lab/mindocr.git
cd mindocr
pip install -e .
```

--------------------------------

### MindOCR Benchmark Command Example

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/frequently_asked_questions.md

This snippet shows an example of a benchmark command used with MindOCR on an Ascend device. The output logs from such a command might contain the 'acl open device 0 failed' error, indicating an issue with the Ascend environment setup.

```bash
benchmark --modelFile=dbnet_mobilenetv3_lite.mindir --device=Ascend --inputShapes='1,3,736,1280' --loopCount=100 - -wammUpLoopCount=10
ModelPath = dbnet_mobilenetv3_lite.mindir
ModelType = MindIR
InDatapath = 
GroupInfoFile = 
ConfigFilepath = 
InDataType = bin
LoopCount = 100
DeviceType = Ascend
AccuracyThreshold = 0.5
CosineDistanceThreshold = -1.1
WarmUpLoopCount = 10
NumThreads = 2	InterOpParallelNum = 1
Fpl16Priority = 0	EnableparalÍel = 0
calibDataPath = 
EnableGLTexture = 0
cpuBindMode = HIGHER CPU
CalibDataType = FLOAT
Resize Dims: 1 3 736 1280
start unified benchmark run
IERROR] ME (26748,7f6c73867fc0, benchmark) :2023-10-26-09:51 : 54.833.515 Imindspore/lite/src/extend rt/kernel/ascend/model/model_infer.cc:59] Init] Acl open device 0 failed.
[ERROR] ME (26748,7f6c73867fc0,benchmark):2023-10-26-09:51:54.833.573 [mindspore/lite/src/extend rt/kernel/ascend/src/custom_ascend_kernel.cc:141] Init] Model i
nfer init failed.	[ERROR] ME (26748, 7f6c73867fc0, benchmark) :2023-10-26-09:51:54.833.604 [mindspore/lite/src/extendrt/session/single_op_session.cc:198] BuildCustomAscendKernelImpl] kernel init failed CustomAscend
[ERROR] ME (26748,7f6c73867fc0, benchmark) :2023-10-26-09:51 :54.833.669 [mindspore/li te/src/extendrt/session/single_op_sess ion.cc:220] BuildCustomAscendKernel] Build ascend kernel failed for node: custom_0
[ERROR] ME (26748,7f6c73867fc0,benchmark) :2023-10-26-09:51 : 54.833.699 [mindspore/lite/src/extend rt/session/single_op_session.cc:302] CompileGraph] Failed to Build custom ascend kernel
[ERROR] ME (26748,7f6c73867fc0,benchmark) :2023-10-26-09:51:54.833.727 [mindspore/lite/s rc/extendrt/cxx_api/model/model_impl.cc:413] BuildByBufferImpl] compile graph failed.
[ERROR] ME (26748, 7f6c73867fc0, benchmark):2023-10-26-09:51:54.835.590 [mindspore/lite/tools/benchma rk/benchmark_unified_api.cc:1256] CompileGraph] ms_model_.Build failed while running
IERROR] ME (26748,7f6c73867fc0,benchmark) :2023-10-26-09:51:54.835.627 [mindspore/lite/tools/benchma rk/benchmark_unified_api.cc:1325] RunBenchmark] Compile graph failed.
[ERROR] ME(26748,7f6c73867fc0, benchmark):2023-10-26-09: 51:54.835.662 [mindspore/lite/tools/benchmark/ run_benchmark.cc :78] RunBenchmark] Run Benchmark dbnet_mobilenetv3_lite.mindi r Failed : -1
ms_model_.Build failed while running Run Benchmark dbnet mobilenetv3 lite.mindir Failed : -1
```

--------------------------------

### Troubleshoot seqeval pip install error

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/zh/tutorials/frequently_asked_questions.md

This snippet captures the error output from a pip installation command failing due to issues with the 'seqeval' package. The failure is attributed to a missing dependency, 'setuptools_scm', during the setup process, leading to a 'metadata-generation-failed' error.

```bash
Collecting seqeval>=1.2.2 (from -r requirements.txt (line 19))
  Downloading http://mirrors.aliyun.com/pypi/packages/9d/2d/233c79d5b4e5ab1dbf111242299153f3caddddbb691219f363ad55ce783d/seqeval-1.2.2.tar.gz (43 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 43.6/43.6 kB 181.0 kB/s eta 0:00:00
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [48 lines of output]
      /home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py:80: _DeprecatedInstaller: setuptools.installer and fetch_build_eggs are deprecated.
      !!

              ********************************************************************************
              Requirements should be satisfied by a PEP 517 installer.
              If you are using pip, you can try `pip install --use-pep517`.
              ********************************************************************************

      !!
        dist.fetch_build_eggs(dist.setup_requires)
      WARNING: The repository located at mirrors.aliyun.com is not a trusted or secure host and is being ignored. If this repository is available via HTTPS we recommend you use HTTPS instead, otherwise you may silence this warning and allow it anyway with '--trusted-host mirrors.aliyun.com'.
      ERROR: Could not find a version that satisfies the requirement setuptools_scm (from versions: none)
      ERROR: No matching distribution found for setuptools_scm
      Traceback (most recent call last):
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/installer.py", line 101, in _fetch_build_egg_no_warn
          subprocess.check_call(cmd)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/subprocess.py", line 373, in check_call
          raise CalledProcessError(retcode, cmd)
      subprocess.CalledProcessError: Command '['/home/ma-user/anaconda3/envs/MindSpore/bin/python3.9', '-m', 'pip', '--disable-pip-version-check', 'wheel', '--no-deps', '-w', '/tmp/tmpusgt0k69', '--quiet', 'setuptools_scm']' returned non-zero exit status 1.

      The above exception was the direct cause of the following exception:

      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "/tmp/pip-install-m2kqztlz/seqeval_da00f708dc0e483b92cd18083513d5e7/setup.py", line 27, in <module>
          setup(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py", line 102, in setup
          _install_setup_requires(attrs)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py", line 75, in _install_setup_requires
          _fetch_build_eggs(dist)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py", line 80, in _fetch_build_eggs
          dist.fetch_build_eggs(dist.setup_requires)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/dist.py", line 636, in fetch_build_eggs
          return _fetch_build_eggs(self, requires)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/installer.py", line 38, in _fetch_build_eggs
          resolved_dists = pkg_resources.working_set.resolve(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 829, in resolve
          dist = self._resolve_dist(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 865, in _resolve_dist
          dist = best[req.key] = env.best_match(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 1135, in best_match
          return self.obtain(req, installer)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 1147, in obtain
          return installer(requirement)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/installer.py", line 103, in _fetch_build_egg_no_warn
          raise DistutilsError(str(e)) from e
      distutils.errors.DistutilsError: Command '['/home/ma-user/anaconda3/envs/MindSpore/bin/python3.9', '-m', 'pip', '--disable-pip-version-check', 'wheel', '--no-deps', '-w', '/tmp/tmpusgt0k69', '--quiet', 'setuptools_scm']' returned non-zero exit status 1.
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.

```

--------------------------------

### Install and Update Packages

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/zh/tutorials/frequently_asked_questions.md

Commands to update setuptools and setuptools_scm, and install the seqeval package with a specific mirror.

```bash
pip3 install --upgrade setuptools
pip3 install --upgrade setuptools_scm
pip3 install seqeval -i https://pypi.tuna.tsinghua.edu.cn/simple
```

--------------------------------

### Setup Local Development Environment

Source: https://github.com/mindspore-lab/mindocr/blob/main/CONTRIBUTING_CN.md

Steps to create a dedicated conda environment for MindOCR development, activate it, and install the project in editable mode. This ensures dependencies are managed correctly.

```shell
conda create -n mindocr python=3.8
conda activate mindocr
cd mindocr
pip install -e .
```

--------------------------------

### MindOCR Installation Instruction

Source: https://github.com/mindspore-lab/mindocr/blob/main/configs/rec/crnn/README_CN.md

Provides guidance on how to install the MindOCR environment, directing users to the main installation instructions.

```APIDOC
Installation:
  Refer to MindOCR's [installation instruction](https://github.com/mindspore-lab/mindocr#installation) for environment setup.
```

--------------------------------

### Install MindOCR on Windows (Initial Attempt)

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/zh/tutorials/frequently_asked_questions.md

Shows the standard commands to clone the MindOCR repository and install it using pip. This process often fails on Windows due to the 'lanms' dependency not being compatible.

```bash
git clone git@gitee.com:mindspore-lab/mindocr.git
cd mindocr
pip install -e .
```

--------------------------------

### ABINet Configuration Example

Source: https://github.com/mindspore-lab/mindocr/blob/main/configs/rec/abinet/README_CN.md

Example YAML configuration for ABINet model training and evaluation. It details parameters for distributed training, batch size, dataset paths, and checkpoint saving/loading.

```yaml
system:
  distribute: True                                                    # 分布式训练为True，单卡训练为False
  amp_level: 'O3'
  seed: 42
  val_while_train: True                                               # 边训练边验证
  drop_overflow_update: False
common:
  ... 
  batch_size: &batch_size 96                                          # 训练批大小
...
train:
  ckpt_save_dir: './tmp_rec'                                          # 训练结果（包括checkpoint、每个epoch的性能和曲线图）保存目录
  dataset_sink_mode: False
  dataset:
    type: LMDBDataset
    dataset_root: dir/to/data_lmdb_release/                           # 训练数据集根目录
    data_dir: train/                                               # 训练数据集目录，将与`dataset_root`拼接形成完整训练数据集目录
    # label_files:                                                    # 训练数据集的标签文件路径，将与`dataset_root`拼接形成完整的训练数据的标签文件路径。当数据集为LMDB格式时无需配置
...
eval:
  ckpt_load_path: './tmp_rec/best.ckpt'                               # checkpoint文件路径
  dataset_sink_mode: False
  dataset:
    type: LMDBDataset
    dataset_root: dir/to/data_lmdb_release/                           # 验证或评估数据集根目录
    data_dir: evaluation/                                             # 验证或评估数据集目录，将与`dataset_root`拼接形成完整验证或评估数据集目录
    # label_file:                                                     # 验证或评估数据集的标签文件路径，将与`dataset_root`拼接形成完整的验证或评估数据的标签文件路径。当数据集为LMDB格式时无需配置
  ...
  loader:
      shuffle: False
      batch_size: 96                                                  # 验证或评估批大小
...

```

--------------------------------

### Example 8-Device Training Script (Bash)

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/distribute_train.md

A sample bash script for launching distributed training on 8 Ascend devices. It sets environment variables for device count, rank size, and the HCCL table file, then starts multiple training processes in the background.

```bash
#!/bin/bash
export DEVICE_NUM=8
export RANK_SIZE=8
export RANK_TABLE_FILE="./hccl_8p_01234567_127.0.0.1.json"

for ((i = 0; i < ${RANK_SIZE}; i++)); do
    export DEVICE_ID=$i
    export RANK_ID=$i
    echo "Launching rank: ${RANK_ID}, device: ${DEVICE_ID}"
    if [ $i -eq 0 ]; then
      echo 'i am 0'
      python -u tools/train.py --config configs/rec/crnn/crnn_resnet34_zh.yaml &> ./train.log &
    else
      echo 'not 0'
      python -u tools/train.py --config configs/rec/crnn/crnn_resnet34_zh.yaml &> /dev/null &
    fi
done
```

--------------------------------

### Install Open MPI v4.0.3

Source: https://github.com/mindspore-lab/mindocr/blob/main/examples/license_plate_detection_and_recognition/README_CN.md

Steps to download, configure, compile, and install Open MPI v4.0.3. This is required for distributed training and evaluation. Includes steps for configuring environment variables and testing the installation.

```shell
tar -xvf openmpi-4.0.3.tar.gz
cd openmpi-4.0.0/
./configure --prefix=/安装目录/openmpi
make
make install

vim /etc/profile

##openmpi##
export PATH=$PATH:/安装目录/openmpi/bin
export LD_LIBRARY_PAHT=/安装目录/openmpii/lib

source /etc/profile

cd /安装目录/openmpi/examples
make
./hello_c
```

--------------------------------

### Install MindOCR

Source: https://github.com/mindspore-lab/mindocr/blob/main/examples/license_plate_detection_and_recognition/README_CN.md

Instructions to clone the MindOCR repository, checkout a specific version (v0.3.2), and install its dependencies using pip. This sets up the MindOCR environment.

```shell
git clone https://github.com/mindspore-lab/mindocr.git
git checkout v0.3.2
cd mindocr
pip install -r requirements.txt
pip install -e .
```

--------------------------------

### Rank Table Configuration Example

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/zh/tutorials/distribute_train.md

Example JSON structure for the rank table file. It defines server configurations, including IP addresses and device mappings (device_id, device_ip, rank_id) for distributed training.

```json
{
    "version": "1.0",
    "server_count": "1",
    "server_list": [
        {
            "server_id": "127.0.0.1",
            "device": [
                {
                    "device_id": "4",
                    "device_ip": "192.168.100.100",
                    "rank_id": "0"
                },
                {
                    "device_id": "5",
                    "device_ip": "192.168.101.100",
                    "rank_id": "1"
                },
                {
                    "device_id": "6",
                    "device_ip": "192.168.102.100",
                    "rank_id": "2"
                },
                {
                    "device_id": "7",
                    "device_ip": "192.168.103.100",
                    "rank_id": "3"
                }
            ],
            "host_nic_ip": "reserve"
        }
    ],
    "status": "completed"
}
```

--------------------------------

### Install Visual C++ Build Tools for Windows

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/zh/tutorials/frequently_asked_questions.md

Provides a link to download and install the necessary Microsoft Visual C++ Build Tools, which are required to compile C++ extensions like 'lanms-neo' on Windows.

```bash
https://visualstudio.com/zh-hans/visual-cpp-build-tools/
```

--------------------------------

### Install MindOCR Dependencies

Source: https://github.com/mindspore-lab/mindocr/blob/main/README.md

Installs all required Python packages for MindOCR by reading from the requirements.txt file. This is a prerequisite for building or running the project.

```shell
pip install -r requirements.txt
```

--------------------------------

### Install from Source

Source: https://github.com/mindspore-lab/mindocr/blob/main/README_CN.md

Clones the MindOCR repository from GitHub and installs it in an editable mode, allowing for direct code modifications and testing.

```shell
git clone https://github.com/mindspore-lab/mindocr.git
cd mindocr
pip install -e .
```

--------------------------------

### Install lanms-neo on Windows (Build Error)

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/zh/tutorials/frequently_asked_questions.md

Details the process of attempting to install 'lanms-neo' as a Windows-compatible alternative to 'lanms'. The build process fails with an error requiring Microsoft Visual C++ Build Tools.

```bash
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Collecting lanms-neo==1.0.2
  Downloading https://pypi.tuna.tsinghua.edu.cn/packages/7b/fe/beff7e7e4455cb9f69c5734897ca8552a57f6423b062ec86b2ebc1d79c0d/lanms_neo-1.0.2.tar.gz (39 kB)
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
  Preparing metadata (pyproject.toml) ... done
Building wheels for collected packages: lanms-neo
  Building wheel for lanms-neo (pyproject.toml) ... error
  error: subprocess-exited-with-error

  × Building wheel for lanms-neo (pyproject.toml) did not run successfully.
  │ exit code: 1
  ╰─> [10 lines of output]
      running bdist_wheel
      running build
      running build_py
      creating build
      creating build\lib.win-amd64-cpython-37
      creating build\lib.win-amd64-cpython-37\lanms
      copying lanms\__init__.py -> build\lib.win-amd64-cpython-37\lanms
      running build_ext
      building 'lanms._C' extension
      error: Microsoft Visual C++ 14.0 or greater is required. Get it with "Microsoft C++ Build Tools": https://visualstudio.com/visual-cpp-build-tools/
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
  ERROR: Failed building wheel for lanms-neo
Failed to build lanms-neo
ERROR: Could not build wheels for lanms-neo, which is required to install pyproject.toml-based projects
```

--------------------------------

### Install MindOCR from Source

Source: https://github.com/mindspore-lab/mindocr/blob/main/README.md

Clones the MindOCR repository from GitHub and installs the package in editable mode. This method is recommended for development as it allows for direct code modifications and immediate reflection.

```shell
git clone https://github.com/mindspore-lab/mindocr.git
cd mindocr
pip install -e .
```

--------------------------------

### FileNotFoundError during MindOCR Installation

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/frequently_asked_questions.md

An example of a `FileNotFoundError` that can occur during the MindOCR installation on Windows. This error typically indicates that a required system file or path cannot be found by the installation process.

```bash
FileNotFoundError: [WinError 2]  the system cannot find the file specified.
```

--------------------------------

### Test Open MPI Installation

Source: https://github.com/mindspore-lab/mindocr/blob/main/examples/license_plate_detection_and_recognition/README.md

Compiles and executes the 'hello_c' example program from the Open MPI installation directory to verify that Open MPI is correctly installed and functional.

```shell
cd /installation_directory/openmpi/examples
make
./hello_c
```

--------------------------------

### Install MindOCR via Docker

Source: https://github.com/mindspore-lab/mindocr/blob/main/README.md

Provides steps to pull MindOCR Docker images for specific configurations (e.g., 910, 910*) and create/run a container with necessary device mappings and volume mounts. It also includes instructions for entering the container and setting up the environment variables.

```bash
docker pull swr.cn-central-221.ovaijisuan.com/mindocr/mindocr_dev_910_ms_2_2_10_cann7_0_py39:v1
docker pull swr.cn-central-221.ovaijisuan.com/mindocr/mindocr_dev_ms_2_2_10_cann7_0_py39:v1
```

```bash
docker_name="temp_mindocr"
# 910
image_name="swr.cn-central-221.ovaijisuan.com/mindocr/mindocr_dev_910_ms_2_2_10_cann7_0_py39:v1"
# 910*
# image_name="swr.cn-central-221.ovaijisuan.com/mindocr/mindocr_dev_ms_2_2_10_cann7_0_py39:v1"

docker run --privileged --name ${docker_name} \
    --tmpfs /tmp \
    --tmpfs /run \
    -v /sys/fs/cgroup:/sys/fs/cgroup:ro \
    --device=/dev/davinci1 \
    --device=/dev/davinci2 \
    --device=/dev/davinci3 \
    --device=/dev/davinci4 \
    --device=/dev/davinci5 \
    --device=/dev/davinci6 \
    --device=/dev/davinci7 \
    --device=/dev/davinci_manager \
    --device=/dev/hisi_hdc \
    --device=/dev/devmm_svm \
    -v /etc/localtime:/etc/localtime \
    -v /usr/local/Ascend/driver:/usr/local/Ascend/driver \
    -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
    --shm-size 800g \
    --cpus 96 \
    --security-opt seccomp=unconfined \
    --network=bridge -itd ${image_name} bash
```

```bash
# set docker id
container_id="your docker id"
docker exec -it --user root $container_id bash
```

```bash
source env_setup.sh
```

--------------------------------

### MindSpore Lite Installation and Environment Setup

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/inference/environment.md

Provides instructions to install MindSpore Lite by unpacking the inference toolkit archive and setting environment variables for runtime and tools. It also includes installing the Python wheel package.

```shell
tar -xvf mindspore-lite-2.2.14-linux-{arch}.tar.gz
cd mindspore-lite-2.2.14-linux-{arch}/
export LITE_HOME=${PWD}    # The actual path after extracting the tar package
export LD_LIBRARY_PATH=$LITE_HOME/runtime/lib:$LITE_HOME/runtime/third_party/dnnl:$LITE_HOME/tools/converter/lib:$LD_LIBRARY_PATH
export PATH=$LITE_HOME/tools/converter/converter:$LITE_HOME/tools/benchmark:$PATH
```

```shell
pip install mindspore_lite-2.2.14-{python_version}-linux_{arch}.whl
```

--------------------------------

### Find liboptiling.so

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/frequently_asked_questions.md

Command to locate the liboptiling.so file within the CANN installation directory. This file is crucial for optimal tiling operations on Ascend devices.

```bash
find ./ -name liboptiling.so
```

--------------------------------

### Install Git Hook Scripts

Source: https://github.com/mindspore-lab/mindocr/blob/main/CONTRIBUTING_CN.md

Installs pre-commit hooks to automatically run linting and code style checks on each commit. This automates code quality enforcement.

```shell
pre-commit install
```

--------------------------------

### YOLOv8 Layout Analysis Configuration Example

Source: https://github.com/mindspore-lab/mindocr/blob/main/configs/layout/yolov8/README_CN.md

Example configuration file for YOLOv8 layout analysis, detailing parameters for system settings, common options, training, and evaluation. Key parameters include distributed training, batch size, and checkpoint paths.

```yaml
system:
  distribute: &distribute True  # 分布式训练为True，单卡训练为False
  amp_level: 'O0'
  amp_level_infer: "O0"
  seed: 42
  val_while_train: False  # 边训练边验证
  drop_overflow_update: False
common:
  ... 
  batch_size: 16  # 训练批大小
  annotations_path: publaynet/val.json
...
train:
  ckpt_save_dir: './tmp_layout'  # 训练结果（包括checkpoint、每个epoch的性能和曲线图）保存目录
  dataset_sink_mode: False
  dataset:
    type: PublayNetDataset
    dataset_path: publaynet/train.txt  # 训练数据集路径
...
eval:
  ckpt_load_path: './tmp_layout/best.ckpt'  # checkpoint文件路径
  dataset_sink_mode: False
  dataset:
    type: PublayNetDataset
    dataset_path: publaynet/val.txt  # 验证数据集路径
  ...
  loader:
    shuffle: False
    batch_size: 16  # 验证批大小
...
```

--------------------------------

### Start Distributed Training Script

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/zh/tutorials/distribute_train.md

Bash script to initiate distributed training on multiple devices. It sets environment variables like RANK_SIZE and RANK_TABLE_FILE, then launches training processes for each rank, mapping them to specific devices.

```bash
#!/bin/bash
export DEVICE_NUM=8
export RANK_SIZE=4
export RANK_TABLE_FILE="./hccl_4p_4567_127.0.0.1.json"

for ((i = 0; i < ${RANK_SIZE}; i++)); do
    export DEVICE_ID=$((i+4))
    export RANK_ID=$i
    echo "Launching rank: ${RANK_ID}, device: ${DEVICE_ID}"
    if [ $i -eq 0 ]; then
      echo 'i am 0'
      python -u tools/train.py --config configs/rec/crnn/crnn_resnet34_zh.yaml &> ./train.log &
    else
      echo 'not 0'
      python -u tools/train.py --config configs/rec/crnn/crnn_resnet34_zh.yaml &> /dev/null &
    fi
done
```

--------------------------------

### Example HCCL Rank Table Configuration

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/distribute_train.md

An example JSON structure for the HCCL rank table file. This configuration details server information, device IDs, IPs, and rank IDs essential for distributed training setup.

```json
{
    "version": "1.0",
    "server_count": "1",
    "server_list": [
        {
            "server_id": "127.0.0.1",
            "device": [
                {
                    "device_id": "4",
                    "device_ip": "192.168.100.100",
                    "rank_id": "0"
                },
                {
                    "device_id": "5",
                    "device_ip": "192.168.101.100",
                    "rank_id": "1"
                },
                {
                    "device_id": "6",
                    "device_ip": "192.168.102.100",
                    "rank_id": "2"
                },
                {
                    "device_id": "7",
                    "device_ip": "192.168.103.100",
                    "rank_id": "3"
                }
            ],
            "host_nic_ip": "reserve"
        }
    ],
    "status": "completed"
}
```

--------------------------------

### Failed seqeval Installation with pip

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/frequently_asked_questions.md

This snippet illustrates the error message received when attempting to install dependencies from a requirements file, specifically failing on the 'seqeval' package. The root cause is identified as the inability to find the 'setuptools_scm' distribution, which is required during the package's setup process.

```bash
pip install -r requirements.txt
```

```python
Collecting seqeval>=1.2.2 (from -r requirements.txt (line 19))
  Downloading http://mirrors.aliyun.com/pypi/packages/9d/2d/233c79d5b4e5ab1dbf111242299153f3caddddbb691219f363ad55ce783d/seqeval-1.2.2.tar.gz (43 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 43.6/43.6 kB 181.0 kB/s eta 0:00:00
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [48 lines of output]
      /home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py:80: _DeprecatedInstaller: setuptools.installer and fetch_build_eggs are deprecated.
      !!

              ********************************************************************************
              Requirements should be satisfied by a PEP 517 installer.
              If you are using pip, you can try `pip install --use-pep517`.
              ********************************************************************************

      !!
        dist.fetch_build_eggs(dist.setup_requires)
      WARNING: The repository located at mirrors.aliyun.com is not a trusted or secure host and is being ignored. If this repository is available via HTTPS we recommend you use HTTPS instead, otherwise you may silence this warning and allow it anyway with '--trusted-host mirrors.aliyun.com'.
      ERROR: Could not find a version that satisfies the requirement setuptools_scm (from versions: none)
      ERROR: No matching distribution found for setuptools_scm
      Traceback (most recent call last):
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/installer.py", line 101, in _fetch_build_egg_no_warn
          subprocess.check_call(cmd)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/subprocess.py", line 373, in check_call
          raise CalledProcessError(retcode, cmd)
      subprocess.CalledProcessError: Command '['/home/ma-user/anaconda3/envs/MindSpore/bin/python3.9', '-m', 'pip', '--disable-pip-version-check', 'wheel', '--no-deps', '-w', '/tmp/tmpusgt0k69', '--quiet', 'setuptools_scm']' returned non-zero exit status 1.

      The above exception was the direct cause of the following exception:

      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "/tmp/pip-install-m2kqztlz/seqeval_da00f708dc0e483b92cd18083513d5e7/setup.py", line 27, in <module>
          setup(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py", line 102, in setup
          _install_setup_requires(attrs)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py", line 75, in _install_setup_requires
          _fetch_build_eggs(dist)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/__init__.py", line 80, in _fetch_build_eggs
          dist.fetch_build_eggs(dist.setup_requires)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/dist.py", line 636, in fetch_build_eggs
          return _fetch_build_eggs(self, requires)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/installer.py", line 38, in _fetch_build_eggs
          resolved_dists = pkg_resources.working_set.resolve(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 829, in resolve
          dist = self._resolve_dist(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 865, in _resolve_dist
          dist = best[req.key] = env.best_match(
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 1135, in best_match
          return self.obtain(req, installer)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/pkg_resources/__init__.py", line 1147, in obtain
          return installer(requirement)
        File "/home/ma-user/anaconda3/envs/MindSpore/lib/python3.9/site-packages/setuptools/installer.py", line 103, in _fetch_build_egg_no_warn
          raise DistutilsError(str(e)) from e
distutils.errors.DistutilsError: Command '['/home/ma-user/anaconda3/envs/MindSpore/bin/python3.9', '-m', 'pip', '--disable-pip-version-check', 'wheel', '--no-deps', '-w', '/tmp/tmpusgt0k69', '--quiet', 'setuptools_scm']' returned non-zero exit status 1.
[end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.

```

--------------------------------

### Install via Docker

Source: https://github.com/mindspore-lab/mindocr/blob/main/README_CN.md

Provides instructions for pulling a pre-configured Docker image, running a container with necessary device mappings and volume mounts, and entering the container to set up the environment.

```shell
# Pull Docker image (example for 910)
docker pull swr.cn-central-221.ovaijisuan.com/mindocr/mindocr_dev_910_ms_2_2_10_cann7_0_py39:v1

# Run container (example for 910)
docker_name="temp_mindocr"
image_name="swr.cn-central-221.ovaijisuan.com/mindocr/mindocr_dev_910_ms_2_2_10_cann7_0_py39:v1"
docker run --privileged --name ${docker_name} \
    --tmpfs /tmp \
    --tmpfs /run \
    -v /sys/fs/cgroup:/sys/fs/cgroup:ro \
    --device=/dev/davinci1 \
    --device=/dev/davinci2 \
    --device=/dev/davinci3 \
    --device=/dev/davinci4 \
    --device=/dev/davinci5 \
    --device=/dev/davinci6 \
    --device=/dev/davinci7 \
    --device=/dev/davinci_manager \
    --device=/dev/hisi_hdc \
    --device=/dev/devmm_svm \
    -v /etc/localtime:/etc/localtime \
    -v /usr/local/Ascend/driver:/usr/local/Ascend/driver \
    -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
    --shm-size 800g \
    --cpus 96 \
    --security-opt seccomp=unconfined \
    --network=bridge -itd ${image_name} bash

# Enter container
# Replace 'your docker id' with the actual container ID
docker exec -it --user root your_docker_id bash

# Set environment variables inside container
source env_setup.sh
```

--------------------------------

### Distributed Training Command for EAST

Source: https://github.com/mindspore-lab/mindocr/blob/main/configs/det/east/README.md

Command to start distributed training for the EAST model, specifying worker numbers for multi-NPU or multi-machine setups.

```shell
# worker_num is the total number of Worker processes participating in the distributed task.\
# local_worker_num is the number of Worker processes pulled up on the current node.\
# The number of processes is equal to the number of NPUs used for training. In the case of single-machine multi-card worker_num and local_worker_num must be the same.\
msrun --worker_num=8 --local_worker_num=8 python tools/train.py --config configs/det/east/east_r50_icdar15.yaml
```

--------------------------------

### YAML Model Configuration (with Neck)

Source: https://github.com/mindspore-lab/mindocr/blob/main/mindocr/models/README_CN.md

Example YAML configuration for defining a model architecture that includes a backbone, neck, and head. This format allows for quick modification of the base architecture.

```yaml
model: 	# R
  type: det
  backbone: 	# R
    name: det_resnet50 	# R, backbone specification function name
    pretrained: False
  neck: 	# R
    name: FPN 	# R, neck class name
    out_channels: 256	# D, neck class __init__ arg
    #use_asf: True
  head: 	# R, head class name
    name: ConvHead 	# D, head class __init__ arg
    out_channels: 2
    k: 50
```

--------------------------------

### Precision Mode Configuration (config.txt)

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/inference/convert_tutorial.md

Configuration file example for setting the model's precision mode. Options include enforce_fp16, enforce_fp32, preferred_fp32, and enforce_origin. Defaults to enforce_fp16 if not specified.

```text
[ascend_context]
input_format=NCHW
input_shape=x:[1,3,736,1280]
precision_mode=enforce_fp32
```

--------------------------------

### YAML Model Configuration (without Neck)

Source: https://github.com/mindspore-lab/mindocr/blob/main/mindocr/models/README_CN.md

Example YAML configuration for defining a model architecture that omits the neck component, directly connecting the backbone to the head. This is common for certain model types.

```yaml
model:				# R
  type: rec
  backbone:			# R
    name: resnet50 		# R
    pretrained: False
  head:				# R
    name: ConvHead 		# R
    out_channels: 30		# D
```

--------------------------------

### Start Distributed Training

Source: https://github.com/mindspore-lab/mindocr/blob/main/docs/en/tutorials/training_recognition_custom_dataset.md

Initiate distributed training across multiple devices by setting `system.distribute` to True in the configuration and using `mpirun` to launch the training script.

```shell
# To perform distributed training on 4 Ascend devices
mpirun -n 4 python tools/train.py --config configs/rec/crnn/crnn_resnet34_ch.yaml
```