Detection lr_scheduler.step() called every step instead of every epoch

🐛 Describe the bug

The lr_scheduler.step() function should be called once per epoch, not every step as seen in engine.py:

https://github.com/pytorch/vision/blob/main/references/detection/engine.py#L54

Reference: https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate

The --lr-steps argument provided to the train.py script don't make much sense: [16, 22], because these quickly change in the 2nd epoch and remain static for the rest of the training job.

Instead, I might suggest:


def train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq, scaler=None):
    model.train()
    metric_logger = utils.MetricLogger(delimiter="  ")
    metric_logger.add_meter("lr", utils.SmoothedValue(window_size=1, fmt="{value:.6f}"))
    header = f"Epoch: [{epoch}]"

    lr_scheduler = None
    if epoch == 0:
        warmup_factor = 1.0 / 1000
        warmup_iters = min(1000, len(data_loader) - 1)

        lr_scheduler = torch.optim.lr_scheduler.LinearLR(
            optimizer, start_factor=warmup_factor, total_iters=warmup_iters
        )

    for images, targets in metric_logger.log_every(data_loader, print_freq, header):
        images = list(image.to(device) for image in images)
        targets = [{k: v.to(device) if isinstance(v, torch.Tensor) else v for k, v in t.items()} for t in targets]
        with torch.cuda.amp.autocast(enabled=scaler is not None):
            loss_dict = model(images, targets)
            losses = sum(loss for loss in loss_dict.values())

        # reduce losses over all GPUs for logging purposes
        loss_dict_reduced = utils.reduce_dict(loss_dict)
        losses_reduced = sum(loss for loss in loss_dict_reduced.values())

        loss_value = losses_reduced.item()

        if not math.isfinite(loss_value):
            print(f"Loss is {loss_value}, stopping training")
            print(loss_dict_reduced)
            sys.exit(1)

        optimizer.zero_grad()
        if scaler is not None:
            scaler.scale(losses).backward()
            scaler.step(optimizer)
            scaler.update()
        else:
            losses.backward()
            optimizer.step()

        # Only step() the special LinearLR scheduler for the first epoch
        if lr_scheduler is not None and isinstance(lr_scheduler, torch.optim.lr_scheduler.LinearLR) and epoch == 0:
            lr_scheduler.step()

        metric_logger.update(loss=losses_reduced, **loss_dict_reduced)
        metric_logger.update(lr=optimizer.param_groups[0]["lr"])

    # one step for each epoch
    if lr_scheduler is not None:
        lr_scheduler.step()

    return metric_logger

Versions

Collecting environment information... PyTorch version: 2.4.0+cu121 Is debug build: False CUDA used to build PyTorch: 12.1 ROCM used to build PyTorch: N/A

OS: Ubuntu 20.04.6 LTS (x86_64) GCC version: (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0 Clang version: Could not collect CMake version: version 3.16.3 Libc version: glibc-2.31

Python version: 3.10.14 (main, Jun 12 2024, 11:12:41) [GCC 9.4.0] (64-bit runtime) Python platform: Linux-5.4.0-193-generic-x86_64-with-glibc2.31 Is CUDA available: True CUDA runtime version: Could not collect CUDA_MODULE_LOADING set to: LAZY GPU models and configuration: GPU 0: NVIDIA GeForce RTX 3070 Laptop GPU Nvidia driver version: 550.90.07 cuDNN version: Could not collect HIP runtime version: N/A MIOpen runtime version: N/A Is XNNPACK available: True

CPU: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian Address sizes: 48 bits physical, 48 bits virtual CPU(s): 16 On-line CPU(s) list: 0-15 Thread(s) per core: 2 Core(s) per socket: 8 Socket(s): 1 NUMA node(s): 1 Vendor ID: AuthenticAMD CPU family: 25 Model: 80 Model name: AMD Ryzen 9 5900HX with Radeon Graphics Stepping: 0 Frequency boost: enabled CPU MHz: 3916.790 CPU max MHz: 3300.0000 CPU min MHz: 1200.0000 BogoMIPS: 6587.59 Virtualization: AMD-V L1d cache: 256 KiB L1i cache: 256 KiB L2 cache: 4 MiB L3 cache: 16 MiB NUMA node0 CPU(s): 0-15 Vulnerability Gather data sampling: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Retbleed: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr wbnoinvd arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif umip pku ospke vaes vpclmulqdq rdpid overflow_recov succor smca

Versions of relevant libraries: [pip3] cjm-pytorch-utils==0.0.6 [pip3] cjm-torchvision-tfms==0.0.11 [pip3] numpy==1.26.3 [pip3] onnx==1.16.1 [pip3] onnxruntime==1.18.1 [pip3] onnxscript==0.1.0.dev20240628 [pip3] torch==2.4.0 [pip3] torchaudio==2.3.1+cu121 [pip3] torchvision==0.19.0 [pip3] triton==3.0.0 [conda] Could not collect

pytorch / vision

Detection lr_scheduler.step() called every step instead of every epoch #8629

🐛 Describe the bug

Versions