import glob import logging import os import re import subprocess import sys import random from datetime import datetime import numpy as np import torch from torch import optim from torch.cuda.amp import GradScaler from a_cls.zeroshot_cls import evaluate_a_cls from i_cls.zeroshot_cls import evaluate_i_cls from d_cls.zeroshot_cls import evaluate_d_cls from t_cls.zeroshot_cls import evaluate_t_cls from v_cls.zeroshot_cls import evaluate_v_cls from vl_ret.retrieval import evaluate_vl_ret from model.process_clip import set_global_value, print_trainable_parameters try: import wandb except ImportError: wandb = None try: import tensorboardX as tensorboard except ImportError: tensorboard = None try: import horovod.torch as hvd except ImportError: hvd = None from data.build_datasets import get_data from open_clip import create_model_and_transforms, create_loss from training.distributed import is_master, init_distributed_device, broadcast_object from training.logger import setup_logging from training.params import parse_args from training.scheduler import cosine_lr, const_lr, const_lr_cooldown from training.file_utils import pt_load, start_sync_process, remote_sync from train import train_one_epoch from model.build_model import create_vat_model LATEST_CHECKPOINT_NAME = "epoch_latest.pt" MODEL_DICT = {"ViT-L-14": "laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K", "ViT-H-14": "laion/CLIP-ViT-H-14-laion2B-s32B-b79K"} CHECKPOINT_DICT = {"ViT-L-14": "models--laion--CLIP-ViT-L-14-DataComp.XL-s13B-b90K/snapshots/84c9828e63dc9a9351d1fe637c346d4c1c4db341/pytorch_model.bin", "ViT-H-14": "models--laion--CLIP-ViT-H-14-laion2B-s32B-b79K/snapshots/94a64189c3535c1cb44acfcccd7b0908c1c8eb23/pytorch_model.bin"} def random_seed(seed=42, rank=0): torch.manual_seed(seed + rank) np.random.seed(seed + rank) random.seed(seed + rank) def natural_key(string_): """See http://www.codinghorror.com/blog/archives/001018.html""" return [int(s) if s.isdigit() else s for s in re.split(r'(\d+)', string_.lower())] def get_latest_checkpoint(path: str, remote: bool): # as writen, this glob recurses, so can pick up checkpoints across multiple sub-folders if remote: result = subprocess.run(["aws", "s3", "ls", path + "/"], stdout=subprocess.PIPE, stderr=subprocess.PIPE) print(result) if result.returncode == 1: return None checkpoints = [os.path.join(path, x.split(' ')[-1]) for x in result.stdout.decode().split('\n')[:-1]] else: checkpoints = glob.glob(path + '**/*.pt', recursive=True) if checkpoints: checkpoints = sorted(checkpoints, key=natural_key) return checkpoints[-1] return None def SET_GLOBAL_VALUE(k, v): set_global_value(k, v) def main(args): args = parse_args(args) SET_GLOBAL_VALUE('PATCH_DROPOUT', args.force_patch_dropout) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) if torch.cuda.is_available(): # This enables tf32 on Ampere GPUs which is only 8% slower than # float16 and almost as accurate as float32 # This was a default in pytorch until 1.12 torch.backends.cuda.matmul.allow_tf32 = True torch.backends.cudnn.benchmark = True torch.backends.cudnn.deterministic = False # fully initialize distributed device environment device = init_distributed_device(args) # get the name of the experiments if args.name is None: # sanitize model name for filesystem / uri use, easier if we don't use / in name as a rule? model_name_safe = args.model.replace('/', '-') date_str = datetime.now().strftime("%Y_%m_%d-%H_%M_%S") if args.distributed: # sync date_str from master to all ranks date_str = broadcast_object(args, date_str) args.name = '-'.join([ date_str, f"pt_{args.clip_type}", f"text_{args.text_type}", f"bs_{args.batch_size}", f"ep_{args.epochs}", f"mask_{args.force_patch_dropout}", f"lorar_{args.lora_r}" if args.convert_to_lora else "", f"lr_{args.lr}", f"coeflr_{args.coef_lr}", f"warm_{args.warmup}", f"accum_{args.accum_freq}", f"tattn_{args.add_time_attn}" if args.clip_type == 'vl' else "", f"model_{model_name_safe}", f"frm_{args.num_frames}", f"vdb_{args.video_decode_backend}", ]) args.pretrained = CHECKPOINT_DICT[args.model] args.model = MODEL_DICT[args.model] resume_latest = args.resume == 'latest' log_base_path = os.path.join(args.logs, args.name) args.log_base_path = log_base_path args.log_path = None if is_master(args, local=args.log_local): os.makedirs(log_base_path, exist_ok=True) log_filename = f'out-{args.rank}' if args.log_local else 'out.log' args.log_path = os.path.join(log_base_path, log_filename) if os.path.exists(args.log_path) and not resume_latest: print( "Error. Experiment already exists. Use --name {} to specify a new experiment." ) return -1 # Setup text logger args.log_level = logging.DEBUG if args.debug else logging.INFO setup_logging(args.log_path, args.log_level) # Setup wandb, tensorboard, checkpoint logging args.wandb = 'wandb' in args.report_to or 'all' in args.report_to args.tensorboard = 'tensorboard' in args.report_to or 'all' in args.report_to args.checkpoint_path = os.path.join(log_base_path, "checkpoints") if is_master(args): args.tensorboard_path = os.path.join(log_base_path, "tensorboard") if args.tensorboard else '' for dirname in [args.tensorboard_path, args.checkpoint_path]: if dirname: os.makedirs(dirname, exist_ok=True) else: args.tensorboard_path = '' if resume_latest: resume_from = None checkpoint_path = args.checkpoint_path # If using remote_sync, need to check the remote instead of the local checkpoints folder. if args.remote_sync is not None: checkpoint_path = os.path.join(args.remote_sync, args.name, "checkpoints") if args.save_most_recent: print('Error. Cannot use save-most-recent with remote_sync and resume latest.') return -1 if args.remote_sync_protocol != 's3': print('Error. Sync protocol not supported when using resume latest.') return -1 if is_master(args): # Checking for existing checkpoint via master rank only. It is possible for # different rank processes to see different files if a shared file-system is under # stress, however it's very difficult to fully work around such situations. if args.save_most_recent: # if --save-most-recent flag is set, look for latest at a fixed filename resume_from = os.path.join(checkpoint_path, LATEST_CHECKPOINT_NAME) if not os.path.exists(resume_from): # If no latest checkpoint has been saved yet, don't try to resume resume_from = None else: # otherwise, list checkpoint dir contents and pick the newest checkpoint resume_from = get_latest_checkpoint(checkpoint_path, remote=args.remote_sync is not None) if resume_from: logging.info(f'Found latest resume checkpoint at {resume_from}.') else: logging.info(f'No latest resume checkpoint found in {checkpoint_path}.') if args.distributed: # sync found checkpoint path to all ranks resume_from = broadcast_object(args, resume_from) args.resume = resume_from if args.copy_codebase: copy_codebase(args) # start the sync proces if remote-sync is not None remote_sync_process = None if is_master(args) and args.remote_sync is not None: # first make sure it works result = remote_sync( os.path.join(args.logs, args.name), os.path.join(args.remote_sync, args.name), args.remote_sync_protocol ) if result: logging.info('remote sync successful.') else: logging.info('Error: remote sync failed. Exiting.') return -1 # if all looks good, start a process to do this every args.remote_sync_frequency seconds remote_sync_process = start_sync_process( args.remote_sync_frequency, os.path.join(args.logs, args.name), os.path.join(args.remote_sync, args.name), args.remote_sync_protocol ) remote_sync_process.start() if args.precision == 'fp16': logging.warning( 'It is recommended to use AMP mixed-precision instead of FP16. ' 'FP16 support needs further verification and tuning, especially for train.') if args.horovod: logging.info( f'Running in horovod mode with multiple processes / nodes. Device: {args.device}.' f'Process (global: {args.rank}, local {args.local_rank}), total {args.world_size}.') elif args.distributed: logging.info( f'Running in distributed mode with multiple processes. Device: {args.device}.' f'Process (global: {args.rank}, local {args.local_rank}), total {args.world_size}.') else: logging.info(f'Running with a single process. Device {args.device}.') dist_model = None args.distill = args.distill_model is not None and args.distill_pretrained is not None if args.distill: # FIXME: support distillation with grad accum. assert args.accum_freq == 1 # FIXME: support distillation with coca. assert 'coca' not in args.model.lower() if isinstance(args.force_image_size, (tuple, list)) and len(args.force_image_size) == 1: # arg is nargs, single (square) image size list -> int args.force_image_size = args.force_image_size[0] random_seed(args.seed, 0) ############################################################################# # model, preprocess_train, preprocess_val = create_model_and_transforms( # args.model, # args.pretrained, # precision=args.precision, # device=device, # jit=args.torchscript, # force_quick_gelu=args.force_quick_gelu, # force_custom_text=args.force_custom_text, # force_patch_dropout=args.force_patch_dropout, # force_image_size=args.force_image_size, # pretrained_image=args.pretrained_image, # image_mean=args.image_mean, # image_std=args.image_std, # aug_cfg=args.aug_cfg, # output_dict=True, # ) model = create_vat_model(args) args.image_size = model.vision_model.config.image_size ############################################################################# if args.distill: # FIXME: currenlty assumes the model your distilling from has the same tokenizer & transforms. dist_model, _, _ = create_model_and_transforms( args.distill_model, args.distill_pretrained, device=device, precision=args.precision, output_dict=True, ) if args.use_bnb_linear is not None: print('=> using a layer from bitsandbytes.\n' ' this is an experimental feature which requires two extra pip installs\n' ' pip install bitsandbytes triton' ' please make sure to use triton 2.0.0') import bitsandbytes as bnb from open_clip.utils import replace_linear print(f'=> replacing linear layers with {args.use_bnb_linear}') linear_replacement_cls = getattr(bnb.nn.triton_based_modules, args.use_bnb_linear) replace_linear(model, linear_replacement_cls) model = model.to(device) random_seed(args.seed, args.rank) # if args.trace: # model = trace_model(model, batch_size=args.batch_size, device=device) if args.lock_image: # if args.clip_type == 'al' or args.clip_type == 'dl': # for param in model.vision_model.embeddings.parameters(): # param.requires_grad = True # for param in model.vision_model.pre_layrnorm.parameters(): # param.requires_grad = True # else: for param in model.vision_model.embeddings.parameters(): param.requires_grad = False for param in model.vision_model.pre_layrnorm.parameters(): param.requires_grad = False for param in model.vision_model.embeddings.position_embedding.parameters(): param.requires_grad = False model.vision_model.embeddings.class_embedding.requires_grad = True # else: # for param in model.vision_model.embeddings.parameters(): # param.requires_grad = True # for param in model.vision_model.pre_layrnorm.parameters(): # param.requires_grad = True # for param in model.vision_model.post_layernorm.parameters(): # param.requires_grad = True # for param in model.visual_projection.parameters(): # param.requires_grad = True if args.add_time_attn: for name, param in model.vision_model.encoder.layers.named_parameters(): if 'temporal_layer_norm' in name or 'temporal_embedding' in name: param.requires_grad = True # if args.add_time_attn and args.unlock_time_attn: # model.unlock_time_attn() if args.lock_text: for param in model.text_model.parameters(): param.requires_grad = False for param in model.text_projection.parameters(): param.requires_grad = False # else: # for param in model.text_model.embeddings.parameters(): # param.requires_grad = True # for param in model.text_model.final_layer_norm.parameters(): # param.requires_grad = True # for param in model.text_projection.parameters(): # param.requires_grad = True model.logit_scale.requires_grad = args.learn_temp if is_master(args): print_trainable_parameters(model, msg='The model: ') if args.grad_checkpointing: model.set_grad_checkpointing() if is_master(args): logging.info("Model:") # logging.info(f"{str(model)}") logging.info("Args:") args_file = os.path.join(args.logs, args.name, "args.txt") with open(args_file, "w") as f: for name in sorted(vars(args)): val = getattr(args, name) logging.info(f" {name}: {val}") f.write(f"{name}: {val}\n") if args.distributed and not args.horovod: if args.use_bn_sync: model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model) ddp_args = {} if args.ddp_static_graph: # this doesn't exist in older PyTorch, arg only added if enabled ddp_args['static_graph'] = True model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[device], **ddp_args) if args.distill: dist_model = torch.nn.parallel.DistributedDataParallel(dist_model, device_ids=[device], **ddp_args) # create optimizer and scaler ############################################################ # if args.train_data or args.dataset_type == "synthetic": assert not args.trace, 'Cannot train with traced model' no_decay = lambda n, p: p.ndim < 2 or "bn" in n or "ln" in n or "bias" in n or 'logit_scale' in n decay = lambda n, p: not no_decay(n, p) lora = lambda n, p: "lora" in n non_lora = lambda n, p: not lora(n, p) named_parameters = list(model.named_parameters()) no_decay_non_lora_params = [[n, p] for n, p in named_parameters if no_decay(n, p) and non_lora(n, p) and p.requires_grad] decay_non_lora_params = [[n, p] for n, p in named_parameters if decay(n, p) and non_lora(n, p) and p.requires_grad] no_decay_lora_params = [[n, p] for n, p in named_parameters if no_decay(n, p) and lora(n, p) and p.requires_grad] decay_lora_params = [[n, p] for n, p in named_parameters if decay(n, p) and lora(n, p) and p.requires_grad] param_groups = [] if no_decay_non_lora_params: param_groups.append({"params": [p for n, p in no_decay_non_lora_params], "weight_decay": 0., 'lr': args.lr * args.coef_lr}) if decay_non_lora_params: param_groups.append({"params": [p for n, p in decay_non_lora_params], "weight_decay": args.wd, 'lr': args.lr * args.coef_lr}) if no_decay_lora_params: param_groups.append({"params": [p for n, p in no_decay_lora_params], "weight_decay": 0.}) if decay_lora_params: param_groups.append({"params": [p for n, p in decay_lora_params], "weight_decay": args.wd}) optimizer = optim.AdamW( # [ # {"params": no_decay_non_visual_params, "weight_decay": 0.}, # {"params": decay_non_visual_params, "weight_decay": args.wd}, # {"params": no_decay_visual_params, "weight_decay": 0., 'lr': args.lr * args.coef_lr}, # {"params": decay_visual_params, "weight_decay": args.wd, 'lr': args.lr * args.coef_lr}, # ], param_groups, lr=args.lr, betas=(args.beta1, args.beta2), eps=args.eps, ) name_groups = {} if no_decay_non_lora_params: name_groups['no_decay_non_lora_params'] = [{"name": n, "weight_decay": 0., 'lr': args.lr * args.coef_lr} for n, p in no_decay_non_lora_params] if decay_non_lora_params: name_groups['decay_non_lora_params'] = [{"name": n, "weight_decay": args.wd, 'lr': args.lr * args.coef_lr} for n, p in decay_non_lora_params] if no_decay_lora_params: name_groups['no_decay_lora_params'] = [{"name": n, "weight_decay": 0., 'lr': args.lr} for n, p in no_decay_lora_params] if decay_lora_params: name_groups['decay_lora_params'] = [{"name": n, "weight_decay": args.wd, 'lr': args.lr} for n, p in decay_lora_params] if is_master(args): params_file = os.path.join(args.logs, args.name, "params.txt") with open(params_file, "w") as f: for group_name, group in name_groups.items(): logging.info(f"Group name: {group_name}:") f.write(f"Group name: {group_name}:\n") for i in group: logging.info(f"Parameter name: {i['name']}. Learning rate: {i['lr']}. Weight decay: {i['weight_decay']}") f.write(f"Parameter name: {i['name']}. Learning rate: {i['lr']}. Weight decay: {i['weight_decay']}\n") if args.horovod: optimizer = hvd.DistributedOptimizer(optimizer, named_parameters=model.named_parameters()) hvd.broadcast_parameters(model.state_dict(), root_rank=0) hvd.broadcast_optimizer_state(optimizer, root_rank=0) scaler = GradScaler() if args.precision == "amp" else None ############################################################ # optionally resume from a checkpoint start_epoch = 0 if args.resume is not None: checkpoint = pt_load(args.resume, map_location='cpu') if 'epoch' in checkpoint: # resuming a train checkpoint w/ epoch and optimizer state start_epoch = checkpoint["epoch"] sd = checkpoint["state_dict"] if not args.distributed and next(iter(sd.items()))[0].startswith('module'): sd = {k[len('module.'):]: v for k, v in sd.items()} model.load_state_dict(sd) if optimizer is not None: optimizer.load_state_dict(checkpoint["optimizer"]) if scaler is not None and 'scaler' in checkpoint: scaler.load_state_dict(checkpoint['scaler']) logging.info(f"=> resuming checkpoint '{args.resume}' (epoch {start_epoch})") else: # loading a bare (model only) checkpoint for fine-tune or evaluation model.load_state_dict(checkpoint) logging.info(f"=> loaded checkpoint '{args.resume}' (epoch {start_epoch})") # initialize datasets data = get_data(args, epoch=start_epoch) if is_master(args): logging.info(f"{data})") assert len(data), 'At least one train or eval dataset must be specified.' # create scheduler if train scheduler = None if f'{args.clip_type}_pt' in data and optimizer is not None: total_steps = (data[f'{args.clip_type}_pt'].dataloader.num_batches // args.accum_freq) * args.epochs if args.lr_scheduler == "cosine": scheduler = cosine_lr(optimizer, args.lr, args.warmup, total_steps) elif args.lr_scheduler == "const": scheduler = const_lr(optimizer, args.lr, args.warmup, total_steps) elif args.lr_scheduler == "const-cooldown": assert args.epochs_cooldown is not None, \ "Please specify the number of cooldown epochs for this lr schedule." cooldown_steps = (data[f'{args.clip_type}_pt'].dataloader.num_batches // args.accum_freq) * args.epochs_cooldown scheduler = const_lr_cooldown( optimizer, args.lr, args.warmup, total_steps, cooldown_steps, args.lr_cooldown_power, args.lr_cooldown_end) else: logging.error( f'Unknown scheduler, {args.lr_scheduler}. Available options are: cosine, const, const-cooldown.') exit(1) # determine if this worker should save logs and checkpoints. only do so if it is rank == 0 args.save_logs = args.logs and args.logs.lower() != 'none' and is_master(args) writer = None if args.save_logs and args.tensorboard: assert tensorboard is not None, "Please install tensorboard." writer = tensorboard.SummaryWriter(args.tensorboard_path) # if args.wandb and is_master(args): # assert wandb is not None, 'Please install wandb.' # logging.debug('Starting wandb.') # args.train_sz = data["train"].dataloader.num_samples # if args.val_data is not None: # args.val_sz = data["val"].dataloader.num_samples # # you will have to configure this for your project! # wandb.init( # project=args.wandb_project_name, # name=args.name, # id=args.name, # notes=args.wandb_notes, # tags=[], # resume='auto' if args.resume == "latest" else None, # config=vars(args), # ) # if args.debug: # wandb.watch(model, log='all') # wandb.save(params_file) # logging.debug('Finished loading wandb.') if args.torchcompile: logging.info('Compiling model...') model = torch.compile(model) if f'{args.clip_type}_pt' not in data: # If using int8, convert to inference mode. if args.use_bnb_linear is not None: from open_clip.utils import convert_int8_model_to_inference_mode convert_int8_model_to_inference_mode(model) # Evaluate. if "i_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) evaluate_i_cls(model, data, start_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) if "vl_ret" in data: for sub_data in data['vl_ret']: evaluate_vl_ret(model, sub_data, start_epoch, args, writer) if "a_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) for sub_data in data['a_cls']: evaluate_a_cls(model, sub_data, start_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) if "v_cls" in data: evaluate_v_cls(model, data, start_epoch, args, writer) if "d_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) for sub_data in data['d_cls']: evaluate_d_cls(model, sub_data, start_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) if "t_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) for sub_data in data['t_cls']: evaluate_t_cls(model, sub_data, start_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) return loss = create_loss(args) for epoch in range(start_epoch, args.epochs): if is_master(args): logging.info(f'Start epoch {epoch}') train_one_epoch(model, data, loss, epoch, optimizer, scaler, scheduler, dist_model, args, tb_writer=writer) completed_epoch = epoch + 1 if "i_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) evaluate_i_cls(model, data, completed_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) if "vl_ret" in data: for sub_data in data['vl_ret']: evaluate_vl_ret(model, sub_data, completed_epoch, args, writer) if "a_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) for sub_data in data['a_cls']: evaluate_a_cls(model, sub_data, completed_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) if "v_cls" in data: evaluate_v_cls(model, data, completed_epoch, args, writer) if "d_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) for sub_data in data['d_cls']: evaluate_d_cls(model, sub_data, completed_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) if "t_cls" in data: SET_GLOBAL_VALUE('NUM_FRAMES', 1) for sub_data in data['t_cls']: evaluate_t_cls(model, sub_data, completed_epoch, args, writer) SET_GLOBAL_VALUE('NUM_FRAMES', args.num_frames) # Saving checkpoints. if args.save_logs: checkpoint_dict = { "epoch": completed_epoch, "name": args.name, "state_dict": model.state_dict(), "optimizer": optimizer.state_dict(), } if scaler is not None: checkpoint_dict["scaler"] = scaler.state_dict() if completed_epoch == args.epochs or ( args.save_frequency > 0 and (completed_epoch % args.save_frequency) == 0 ): torch.save( checkpoint_dict, os.path.join(args.checkpoint_path, f"epoch_{completed_epoch}.pt"), ) if args.delete_previous_checkpoint: previous_checkpoint = os.path.join(args.checkpoint_path, f"epoch_{completed_epoch - 1}.pt") if os.path.exists(previous_checkpoint): os.remove(previous_checkpoint) if args.save_most_recent: # try not to corrupt the latest checkpoint if save fails tmp_save_path = os.path.join(args.checkpoint_path, "tmp.pt") latest_save_path = os.path.join(args.checkpoint_path, LATEST_CHECKPOINT_NAME) torch.save(checkpoint_dict, tmp_save_path) os.replace(tmp_save_path, latest_save_path) if args.wandb and is_master(args): wandb.finish() # run a final sync. if remote_sync_process is not None: logging.info('Final remote sync.') remote_sync_process.terminate() result = remote_sync( os.path.join(args.logs, args.name), os.path.join(args.remote_sync, args.name), args.remote_sync_protocol ) if result: logging.info('Final remote sync successful.') else: logging.info('Final remote sync failed.') def copy_codebase(args): from shutil import copytree, ignore_patterns new_code_path = os.path.join(args.logs, args.name, "code") if os.path.exists(new_code_path): print( f"Error. Experiment already exists at {new_code_path}. Use --name to specify a new experiment." ) return -1 print(f"Copying codebase to {new_code_path}") current_code_path = os.path.realpath(__file__) for _ in range(3): current_code_path = os.path.dirname(current_code_path) copytree(current_code_path, new_code_path, ignore=ignore_patterns('log', 'logs', 'wandb')) print("Done copying code.") return 1 if __name__ == "__main__": main(sys.argv[1:])