o W h@sjddlZddlZddlZddlZddlZddlmZddlmZde fddZ dd Z d d Z d d Z dS)N)USER_CONFIG_DIR) TORCH_1_9returncCsJttjtj}|d|dWdS1swYdS)a Find a free port on localhost. It is useful in single-node training when we don't want to connect to a real main node but have to set the `MASTER_PORT` environment variable. Returns: (int): The available network port number. )z 127.0.0.1rrN)socketAF_INET SOCK_STREAMbind getsockname)sr J/var/www/vscode/kcb/lib/python3.10/site-packages/ultralytics/utils/dist.pyfind_free_network_port s  $rc Cs|jjd|jjdd\}}dt|jd|d|d|dt|jd|jjd }t d j d d t j d t |dddt d dd}||Wd|jS1sYwY|jS)a* Generate a DDP (Distributed Data Parallel) file for multi-GPU training. This function creates a temporary Python file that enables distributed training across multiple GPUs. The file contains the necessary configuration to initialize the trainer in a distributed environment. Args: trainer (object): The trainer object containing training configuration and arguments. Must have args attribute and be a class instance. Returns: (str): Path to the generated temporary DDP file. Notes: The generated file is saved in the USER_CONFIG_DIR/DDP directory and includes: - Trainer class import - Configuration overrides from the trainer arguments - Model path configuration - Training initialization code .rzd # Ultralytics Multi-GPU training temp file (should be automatically deleted after use) overrides = z& if __name__ == "__main__": from z import z from ultralytics.utils import DEFAULT_CFG_DICT cfg = DEFAULT_CFG_DICT.copy() cfg.update(save_dir='') # handle the extra key 'save_dir' trainer = z9(cfg=cfg, overrides=overrides) trainer.args.model = " model_urlz " results = trainer.train() DDPT)exist_ok_temp_.pyzw+zutf-8F)prefixsuffixmodeencodingdirdeleteN) __class__ __module____name__rsplitvarsargsgetattr hub_sessionmodelrmkdirtempfileNamedTemporaryFileidwritename)trainermoduler)contentfiler r r generate_ddp_files6"       r.cCsVddl}|js t|jt|}trdnd}t}tj d|d|d||g}||fS)a Generate command for distributed training. Args: world_size (int): Number of processes to spawn for distributed training. trainer (object): The trainer object containing configuration for distributed training. Returns: cmd (List[str]): The command to execute for distributed training. file (str): Path to the temporary file created for DDP training. rNztorch.distributed.runztorch.distributed.launchz-mz--nproc_per_nodez --master_port) __main__resumeshutilrmtreesave_dirr.rrsys executable) world_sizer*r/r-dist_cmdportcmdr r r generate_ddp_commandNs   r:cCs$t|d|vrt|dSdS)a0 Delete temporary file if created during distributed data parallel (DDP) training. This function checks if the provided file contains the trainer's ID in its name, indicating it was created as a temporary file for DDP training, and deletes it if so. Args: trainer (object): The trainer object used for distributed training. file (str): Path to the file that might need to be deleted. Examples: >>> trainer = YOLOTrainer() >>> file = "/tmp/ddp_temp_123456789.py" >>> ddp_cleanup(trainer, file) rN)r'osremove)r*r-r r r ddp_cleanupesr=)r;r1rr4r%r torch_utilsrintrr.r:r=r r r r s  2