|hJ`ddlZddlZddlZddlZddlmZddlmZdefdZ dZ defd Z d Z y) N)USER_CONFIG_DIR) TORCH_1_9returncddl}|j|j|j5}|jd|j dcdddS#1swYyxYw)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. rN)z 127.0.0.1rr)socketAF_INET SOCK_STREAMbind getsockname)rss U/var/www/html/test/engine/venv/lib/python3.12/site-packages/ultralytics/utils/dist.pyfind_free_network_portr sQ v~~v'9'9 :"a  }}q!"""s $AA"c*|jjd|jjjdd\}}dt |j d|d|d|dt |jd|j jd }td z jd tjd t|dddtd z d5}|j|ddd|jS#1swYjSxYw)a6 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 (ultralytics.engine.trainer.BaseTrainer): The trainer 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+contentfiles rgenerate_ddp_filer0s%*''2231W5F5F5O5O4PQXXY\^_`LFD ',,  !  $  f"7#6#6 W\\EWEWXYZ Gu##T#2  $ $W+c"  e #    7 99 99s C>>D world_sizecddl}|jstj|jt |}t rdnd}t}tjd|d|d||g}||fS)a Generate command for distributed training. Args: world_size (int): Number of processes to spawn for distributed training. trainer (ultralytics.engine.trainer.BaseTrainer): The trainer 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_dirr0rrsys executable)r1r,r3r/dist_cmdportcmds rgenerate_ddp_commandr=Osj >> g&&' W %D*3&9SH ! #D >>4+=*P_dhcilp qC 9cPt|d|vrtj|yy)aI 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 (ultralytics.engine.trainer.BaseTrainer): The trainer 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/s r ddp_cleanuprBfs' W+cd" $#r>) r@r5r8r'r torch_utilsrintrr0r=rBr>rrGs< """"/dS.r>