WZMIAOMIAO/deep-learning-for-image-processing · error · FileNotFoundError

not found weights file: {}

Error message

not found weights file: {}

What it means

This FileNotFoundError is raised when the --weights path supplied for resuming/fine-tuning does not exist on disk. The code checks os.path.exists(args.weights) and, if the file is missing, fails with the path in the message instead of letting torch.load throw a vaguer error.

Source

Thrown at pytorch_classification/train_multi_GPU/train_single_gpu.py:80

                                               collate_fn=train_data_set.collate_fn)

    val_loader = torch.utils.data.DataLoader(val_data_set,
                                             batch_size=batch_size,
                                             shuffle=False,
                                             pin_memory=True,
                                             num_workers=nw,
                                             collate_fn=val_data_set.collate_fn)

    # 如果存在预训练权重则载入
    model = resnet34(num_classes=args.num_classes).to(device)
    if args.weights != "":
        if os.path.exists(args.weights):
            weights_dict = torch.load(args.weights, map_location=device)
            load_weights_dict = {k: v for k, v in weights_dict.items()
                                 if model.state_dict()[k].numel() == v.numel()}
            print(model.load_state_dict(load_weights_dict, strict=False))
        else:
            raise FileNotFoundError("not found weights file: {}".format(args.weights))

    # 是否冻结权重
    if args.freeze_layers:
        for name, para in model.named_parameters():
            # 除最后的全连接层外，其他权重全部冻结
            if "fc" not in name:
                para.requires_grad_(False)

    pg = [p for p in model.parameters() if p.requires_grad]
    optimizer = optim.SGD(pg, lr=args.lr, momentum=0.9, weight_decay=0.005)
    # Scheduler https://arxiv.org/pdf/1812.01187.pdf
    lf = lambda x: ((1 + math.cos(x * math.pi / args.epochs)) / 2) * (1 - args.lrf) + args.lrf  # cosine
    scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lf)

    for epoch in range(args.epochs):
        # train
        mean_loss = train_one_epoch(model=model,
                                    optimizer=optimizer,

View on GitHub (pinned to 1ec3fe6f37)

Solutions

1. Print/verify the exact path: ls -l <args.weights> from the same working directory the script runs in.
2. Pass an absolute path to --weights to avoid cwd ambiguity.
3. Download or train the checkpoint first (run the training script without --weights, or fetch the released model file).
4. If you intend to train from scratch, omit --weights so the branch is skipped entirely.
5. Add a caller-side os.path.exists() precheck before launching the script.

Example fix

# before
python train_single_gpu.py --weights ./checkpoints/model.pth
# FileNotFoundError if missing
# after
import os
cmd = ["python", "train_single_gpu.py", "--weights", "/abs/path/model.pth"]
assert os.path.exists("/abs/path/model.pth"), "checkpoint missing"
subprocess.run(cmd)

import os, sys
weights = args.weights
if weights and not os.path.isfile(weights):
    sys.exit(f"weights not found: {os.path.abspath(weights)}")

try:
    weights_dict = torch.load(args.weights, map_location=device)
except FileNotFoundError as e:
    logging.error("Checkpoint missing: %s — start training from scratch", e)
    weights_dict = None

When it happens

Trigger: Invoking train_single_gpu.py with --weights pointing to a nonexistent file: typo in path, weights never trained/downloaded, relative path resolved from a different working directory, or the checkpoint was deleted/moved before the run.

Common situations: Copying a training command from docs without first running the training that produces weights; running from a different cwd so relative paths break; wrong filename (weights.pth vs best.pth); checkpoint stored in cloud storage not yet downloaded.

Understand the failure class

Background: "File not found" and ENOENT errors: why libraries can't find a file that should exist — this error's family across 50 libraries.

Related errors

• not found weights file: {}
• not found weights file: {}
• not found weights file: {}
• not found weights file: {}
• not found weights file: {}

AI-assisted analysis of WZMIAOMIAO/deep-learning-for-image-processing@1ec3fe6f37 (2026-08-30). Data as JSON: /api/errors/9012500de688d027. Report an issue: GitHub.