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

not found weights file: {}

Error message

not found weights file: {}

What it means

train.py only attempts to load pretrained weights if os.path.exists(args.weights); when the file path is set but missing, it raises FileNotFoundError instead of silently training from scratch.

Source

Thrown at pytorch_classification/Test10_regnet/train.py:76

                                             batch_size=batch_size,
                                             shuffle=False,
                                             pin_memory=True,
                                             num_workers=nw,
                                             collate_fn=val_dataset.collate_fn)

    # 如果存在预训练权重则载入
    model = create_regnet(model_name=args.model_name,
                          num_classes=args.num_classes).to(device)
    # print(model)

    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 "head" not in name:
                para.requires_grad_(False)
            else:
                print("train {}".format(name))

    pg = [p for p in model.parameters() if p.requires_grad]
    optimizer = optim.SGD(pg, lr=args.lr, momentum=0.9, weight_decay=5E-5)
    # 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

View on GitHub (pinned to 1ec3fe6f37)

Solutions

  1. Download the weights first (run pretrain_weights.py) so the file exists at args.weights.
  2. Verify/correct the --weights path (absolute path or correct relative path from the CWD).
  3. Pass --weights '' or omit it if you intend to train from scratch (per the script's empty-string convention).

Example fix

// before
python train.py --weights ./regnetx_400mf.pth   # file not downloaded
// after
python pretrain_weights.py && python train.py --weights ./regnetx_400mf.pth
Defensive patterns

Strategy: validation

Validate before calling

import os
weights = "./regnetx_400mf.pth"
if weights and not os.path.exists(weights):
    raise SystemExit(f"weights missing: {weights}; run pretrain_weights.py first")

Type guard

def weights_available(path) -> bool:
    return not path or os.path.isfile(path)

Try / catch

try:
    main()
except FileNotFoundError as e:
    print(f"{e}; downloading weights...")
    os.system("python pretrain_weights.py")
    main()

Prevention

When it happens

Trigger: Running train.py with --weights ./regnetx_400mf.pth where the .pth was never downloaded, renamed, or is in a different working directory.

Common situations: Forgot to run pretrain_weights.py first; typo in the path; weights downloaded under a different filename than the --weights argument.

Related errors


AI-assisted analysis of WZMIAOMIAO/deep-learning-for-image-processing@1ec3fe6f37 (2026-08-30). Data as JSON: /api/errors/3c0425ba40bdb3b0. Report an issue: GitHub.