[bug report]单机版模型导出,入参不生效

[alexab612]

环境 python3.7
easyrec版本: 0.8.5
tf版本 2.9
正常执行异步训练完成之后
执行导出命令

   python3 -m easy_rec.python.export --pipeline_config_path ${config} --model_dir ${model_path} --export_dir ${export_model_path} --export_done_file EXPORT_DONE

会提示模型路径不存在,

allow_soft_placement: true
, '_keep_checkpoint_max': 10, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': None, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x150495f6a278>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}
[2025-03-12 00:00:00,000][INFO] check_mode: False 
Traceback (most recent call last):
  File "/home/apps/miniconda3/envs/py36_tf12_env/lib/python3.6/runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "/home/apps/miniconda3/envs/py36_tf12_env/lib/python3.6/runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "/home/apps/miniconda3/envs/py36_tf12_env/lib/python3.6/site-packages/easy_rec-0.8.5-py3.6.egg/easy_rec/python/export.py", line 150, in <module>
    tf.app.run()
  File "/home/apps/miniconda3/envs/py36_tf12_env/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 125, in run
    _sys.exit(main(argv))
  File "/home/apps/miniconda3/envs/py36_tf12_env/lib/python3.6/site-packages/easy_rec-0.8.5-py3.6.egg/easy_rec/python/export.py", line 140, in main
    FLAGS.verbose, **extra_params)
  File "/home/apps/miniconda3/envs/py36_tf12_env/lib/python3.6/site-packages/easy_rec-0.8.5-py3.6.egg/easy_rec/python/main.py", line 801, in export
    ckpt_path = _get_ckpt_path(pipeline_config, checkpoint_path)
  File "/home/apps/miniconda3/envs/py36_tf12_env/lib/python3.6/site-packages/easy_rec-0.8.5-py3.6.egg/easy_rec/python/main.py", line 271, in _get_ckpt_path
    % pipeline_config.model_dir
AssertionError: pipeline_config.model_dir(experiments/demo) does not exist
easy_rec version: 0.8.5`

看日志会先执行
export.py中的变量更新,
但是在这里 检查模型目录下pipline.config文件时, 没有这个文件(不清楚为什么2.9版本没有生成, tf1.12就会生成),就导致pipeline_config_path 没有更新,同时model_dir也没有更新

接着 在export.py 这里 更新了一次配置文件
然后在main.py中的 这里 也更新了一次.
如果仅在export.py 里面更新配置,会在main.py这里被重置.

所以需要将手动修改3处,

1. 在export.py里加载 pipeline_config 之后 新增.

  if FLAGS.model_dir:
    pipeline_config.model_dir = FLAGS.model_dir
    logging.info('update model_dir to %s' % pipeline_config.model_dir)

2. 在export.py调用main.py时的pipeline_config_path入参改为 pipeline_config.

  export_out_dir = export(FLAGS.export_dir, pipeline_config,
                          FLAGS.checkpoint_path, FLAGS.asset_files,
                          FLAGS.verbose, **extra_params)

3. 在main.py中注释掉export函数中的

 pipeline_config = config_util.get_configs_from_pipeline_file(pipeline_config)