pytorch-image-models/tests/test_optim.py

""" Optimzier Tests

These tests were adapted from PyTorch' optimizer tests.

"""
import functools
import importlib
import os
from copy import deepcopy

import pytest
import torch
from torch.nn import Parameter
from torch.testing._internal.common_utils import TestCase

from timm.optim import create_optimizer_v2, list_optimizers, get_optimizer_class, get_optimizer_info, OptimInfo
from timm.optim import param_groups_layer_decay, param_groups_weight_decay
from timm.scheduler import PlateauLRScheduler

torch_backend = os.environ.get('TORCH_BACKEND')
if torch_backend is not None:
    importlib.import_module(torch_backend)
torch_device = os.environ.get('TORCH_DEVICE', 'cuda')

# HACK relying on internal PyTorch test functionality for comparisons that I don't want to write
torch_tc = TestCase()


def _test_basic_cases_template(weight, bias, input, constructor, scheduler_constructors):
    weight = Parameter(weight)
    bias = Parameter(bias)
    input = Parameter(input)
    optimizer = constructor(weight, bias)
    schedulers = []
    for scheduler_constructor in scheduler_constructors:
        schedulers.append(scheduler_constructor(optimizer))

    # to check if the optimizer can be printed as a string
    optimizer.__repr__()

    def fn():
        optimizer.zero_grad()
        y = weight.mv(input)
        if y.is_cuda and bias.is_cuda and y.get_device() != bias.get_device():
            y = y.cuda(bias.get_device())
        loss = (y + bias).pow(2).sum()
        loss.backward()
        return loss

    initial_value = fn().item()
    for _i in range(200):
        for scheduler in schedulers:
            if isinstance(scheduler, PlateauLRScheduler):
                val_loss = fn()
                scheduler.step(val_loss)
            else:
                scheduler.step()
        optimizer.step(fn)

    assert fn().item() < initial_value


def _test_state_dict(weight, bias, input, constructor):
    weight = Parameter(weight)
    bias = Parameter(bias)
    input = Parameter(input)

    def fn_base(optimizer, weight, bias):
        optimizer.zero_grad()
        i = input_device if weight.device.type != 'cpu' else input
        loss = (weight.mv(i) + bias).pow(2).sum()
        loss.backward()
        return loss

    optimizer = constructor(weight, bias)
    fn = functools.partial(fn_base, optimizer, weight, bias)

    # Prime the optimizer
    for _i in range(20):
        optimizer.step(fn)
    # Clone the weights and construct new optimizer for them
    with torch.no_grad():
        weight_c = Parameter(weight.clone().detach())
        bias_c = Parameter(bias.clone().detach())
    optimizer_c = constructor(weight_c, bias_c)
    fn_c = functools.partial(fn_base, optimizer_c, weight_c, bias_c)
    # Load state dict
    state_dict = deepcopy(optimizer.state_dict())
    state_dict_c = deepcopy(optimizer.state_dict())
    optimizer_c.load_state_dict(state_dict_c)

    # Run both optimizations in parallel
    for _i in range(20):
        optimizer.step(fn)
        optimizer_c.step(fn_c)
        torch_tc.assertEqual(weight, weight_c)
        torch_tc.assertEqual(bias, bias_c)
    # Make sure state dict is deterministic with equal but not identical parameters
    torch_tc.assertEqual(optimizer.state_dict(), optimizer_c.state_dict())
    # Make sure repeated parameters have identical representation in state dict
    optimizer_c.param_groups.extend(optimizer_c.param_groups)
    torch_tc.assertEqual(optimizer.state_dict()['param_groups'][-1], optimizer_c.state_dict()['param_groups'][-1])

    # Check that state dict can be loaded even when we cast parameters
    # to a different type and move to a different device.
    if torch_device == 'cpu':
        return
    elif torch_device == 'cuda' and not torch.cuda.is_available():
        return

    with torch.no_grad():
        input_device = Parameter(input.clone().detach().float().to(torch_device))
        weight_device = Parameter(weight.clone().detach().to(torch_device))
        bias_device = Parameter(bias.clone().detach().to(torch_device))
    optimizer_device = constructor(weight_device, bias_device)
    fn_device = functools.partial(fn_base, optimizer_device, weight_device, bias_device)

    state_dict = deepcopy(optimizer.state_dict())
    state_dict_c = deepcopy(optimizer.state_dict())
    optimizer_device.load_state_dict(state_dict_c)

    # Make sure state dict wasn't modified
    torch_tc.assertEqual(state_dict, state_dict_c)

    for _i in range(20):
        optimizer.step(fn)
        optimizer_device.step(fn_device)
        torch_tc.assertEqual(weight, weight_device)
        torch_tc.assertEqual(bias, bias_device)

    # validate deepcopy() copies all public attributes
    def getPublicAttr(obj):
        return set(k for k in obj.__dict__ if not k.startswith('_'))

    assert getPublicAttr(optimizer) == getPublicAttr(deepcopy(optimizer))


def _test_basic_cases(constructor, scheduler_constructors=None):
    if scheduler_constructors is None:
        scheduler_constructors = []
    _test_state_dict(
        torch.randn(10, 5),
        torch.randn(10),
        torch.randn(5),
        constructor
    )
    _test_basic_cases_template(
        torch.randn(10, 5),
        torch.randn(10),
        torch.randn(5),
        constructor,
        scheduler_constructors
    )
    # non-contiguous parameters
    _test_basic_cases_template(
        torch.randn(10, 5, 2)[..., 0],
        torch.randn(10, 2)[..., 0],
        torch.randn(5),
        constructor,
        scheduler_constructors
    )
    # CUDA
    if torch_device == 'cpu':
        return
    elif torch_device == 'cuda' and not torch.cuda.is_available():
        return

    _test_basic_cases_template(
        torch.randn(10, 5).to(torch_device),
        torch.randn(10).to(torch_device),
        torch.randn(5).to(torch_device),
        constructor,
        scheduler_constructors
    )


def _test_model(optimizer, params, device=torch.device('cpu'), after_step=0):
    weight = torch.tensor(
        [[-0.2109, -0.4976], [-0.1413, -0.3420], [-0.2524, 0.6976]],
        device=device, requires_grad=True)
    bias = torch.tensor([-0.1085, -0.2979, 0.6892], device=device, requires_grad=True)
    weight2 = torch.tensor([[-0.0508, -0.3941, -0.2843]], device=device, requires_grad=True)
    bias2 = torch.tensor([-0.0711], device=device, requires_grad=True)
    input = torch.tensor([0.1, 0.2, 0.3, 0.4, 0.5, 0.6], device=device).reshape(3, 2)

    model = torch.nn.Sequential(torch.nn.Linear(2, 3),
                                torch.nn.Sigmoid(),
                                torch.nn.Linear(3, 1),
                                torch.nn.Sigmoid())
    model.to(device)

    pretrained_dict = model.state_dict()
    pretrained_dict['0.weight'] = weight
    pretrained_dict['0.bias'] = bias
    pretrained_dict['2.weight'] = weight2
    pretrained_dict['2.bias'] = bias2
    model.load_state_dict(pretrained_dict)

    optimizer = create_optimizer_v2(model, opt=optimizer, **params)

    prev_loss = float('inf')
    for i in range(20):
        optimizer.zero_grad()
        output = model(input)
        loss = output.sum()
        loss.backward()
        loss = loss.item()
        if i > after_step:
            assert loss < prev_loss
        prev_loss = loss
        optimizer.step()


def rosenbrock(tensor):
    x, y = tensor
    return (1 - x) ** 2 + 100 * (y - x ** 2) ** 2


def drosenbrock(tensor):
    x, y = tensor
    return torch.tensor((-400 * x * (y - x ** 2) - 2 * (1 - x), 200 * (y - x ** 2)))


def _test_rosenbrock(constructor, scheduler_constructors=None):
    if scheduler_constructors is None:
        scheduler_constructors = []
    params_t = torch.tensor([1.5, 1.5])

    params = Parameter(params_t)
    optimizer = constructor([params])
    schedulers = []
    for scheduler_constructor in scheduler_constructors:
        schedulers.append(scheduler_constructor(optimizer))

    solution = torch.tensor([1, 1])
    initial_dist = params.clone().detach().dist(solution)


    def get_grad(_param, _sparse_grad, _w):
        grad = drosenbrock(params.clone().detach())
        # Depending on w, provide only the x or y gradient
        if _sparse_grad:
            if _w:
                i = torch.tensor([[0, 0]], dtype=torch.int64)
                x = grad[0]
                v = torch.tensor([x / 4.0, x - x / 4.0])
            else:
                i = torch.tensor([[1, 1]], dtype=torch.int64)
                y = grad[1]
                v = torch.tensor([y - y / 4.0, y / 4.0])
            grad_out = torch.sparse_coo_tensor(i, v, (2,), dtype=v.dtype)
        else:
            if _w:
                grad_out = torch.tensor([grad[0], 0], dtype=_param.dtype)
            else:
                grad_out = torch.tensor([0, grad[1]], dtype=_param.dtype)
        return grad_out


    def eval(_param, _sparse_grad, _w):
        # Depending on w, provide only the x or y gradient
        optimizer.zero_grad()
        loss = rosenbrock(_param)
        loss.backward()

        grad_out = get_grad(_param, _sparse_grad, _w)
        with torch.no_grad():
            _param.grad = grad_out.to_dense()

        return loss

    for i in range(2000):
        # Do cyclic coordinate descent
        w = i % 2
        optimizer.step(functools.partial(eval, params, True, w))
        for scheduler in schedulers:
            if isinstance(scheduler, PlateauLRScheduler):
                scheduler.step(rosenbrock(params))
            else:
                scheduler.step()

    torch_tc.assertLessEqual(params.clone().detach().dist(solution), initial_dist)


def _build_params_dict(weight, bias, **kwargs):
    return [{'params': [weight]}, dict(params=[bias], **kwargs)]


def _build_params_dict_single(weight, bias, **kwargs):
    return [dict(params=bias, **kwargs)]


@pytest.mark.parametrize('optimizer', list_optimizers(exclude_filters=('fused*', 'bnb*')))
def test_optim_factory(optimizer):
    assert issubclass(get_optimizer_class(optimizer, bind_defaults=False), torch.optim.Optimizer)

    opt_info = get_optimizer_info(optimizer)
    assert isinstance(opt_info, OptimInfo)

    lr = (1e-2,) * 4
    if optimizer in ('mars', 'nadam', 'claprop', 'crmsproptf', 'cadafactorbv', 'csgdw', 'clamb'):
        lr = (1e-3,) * 4
    elif optimizer in ('cmars',):
        lr = (1e-4,) * 4

    try:
        if not opt_info.second_order:  # basic tests don't support second order right now
            # test basic cases that don't need specific tuning via factory test
            _test_basic_cases(
                lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=lr[0])
            )
            _test_basic_cases(
                lambda weight, bias: create_optimizer_v2(
                    _build_params_dict(weight, bias, lr=lr[1]),
                    optimizer,
                    lr=lr[1] / 10)
            )
            _test_basic_cases(
                lambda weight, bias: create_optimizer_v2(
                    _build_params_dict_single(weight, bias, lr=lr[2]),
                    optimizer,
                    lr=lr[2] / 10)
            )
            _test_basic_cases(
                lambda weight, bias: create_optimizer_v2(
                    _build_params_dict_single(weight, bias, lr=lr[3]),
                    optimizer)
            )
    except TypeError as e:
        if 'radamw' in optimizer:
            pytest.skip("Expected for 'radamw' (decoupled decay) to fail in older PyTorch versions.")
        else:
            raise e



#@pytest.mark.parametrize('optimizer', ['sgd', 'momentum'])
# FIXME momentum variant frequently fails in GitHub runner, but never local after many attempts
@pytest.mark.parametrize('optimizer', ['sgd'])
def test_sgd(optimizer):
    # _test_basic_cases(
    #     lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=1e-3),
    #     [lambda opt: StepLR(opt, gamma=0.9, step_size=10)]
    # )
    # _test_basic_cases(
    #     lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=1e-3),
    #     [lambda opt: WarmUpLR(opt, warmup_factor=0.4, warmup_iters=4, warmup_method="linear")]
    # )
    # _test_basic_cases(
    #     lambda weight, bias: optimizer([weight, bias], lr=1e-3),
    #     [lambda opt: WarmUpLR(opt, warmup_factor=0.4, warmup_iters=4, warmup_method="constant")]
    # )
    # _test_basic_cases(
    #     lambda weight, bias: optimizer([weight, bias], lr=1e-3),
    #     [lambda opt: StepLR(opt, gamma=0.9, step_size=10),
    #      lambda opt: WarmUpLR(opt, warmup_factor=0.4, warmup_iters=4)]
    # )
    # _test_basic_cases(
    #     lambda weight, bias: optimizer([weight, bias], lr=1e-3),
    #     [lambda opt: StepLR(opt, gamma=0.9, step_size=10),
    #      lambda opt: ReduceLROnPlateau(opt)]
    # )
    # _test_basic_cases(
    #     lambda weight, bias: optimizer([weight, bias], lr=1e-3),
    #     [lambda opt: StepLR(opt, gamma=0.99, step_size=10),
    #      lambda opt: ExponentialLR(opt, gamma=0.99),
    #      lambda opt: ReduceLROnPlateau(opt)]
    # )
    _test_basic_cases(
        lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=3e-3, momentum=1)
    )
    _test_basic_cases(
        lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=3e-3, momentum=1, weight_decay=.1)
    )
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=1e-3))


@pytest.mark.parametrize('optimizer',  ['adamw', 'adam', 'nadam', 'adamax', 'nadamw'])
def test_adam(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=5e-2)
    )
    _test_model(optimizer, dict(lr=5e-2))


@pytest.mark.parametrize('optimizer',  ['adopt', 'adoptw'])
def test_adopt(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=3e-3)
    )
    _test_model(optimizer, dict(lr=5e-2), after_step=1)  # note no convergence in first step for ADOPT


@pytest.mark.parametrize('optimizer',  ['adan', 'adanw'])
def test_adan(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=5e-2), after_step=1)  # note no convergence in first step for ADOPT


@pytest.mark.parametrize('optimizer',  ['adabelief'])
def test_adabelief(optimizer):
    _test_basic_cases(
        lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=1e-3, weight_decay=1)
    )
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=5e-2)
    )
    _test_model(optimizer, dict(lr=5e-2))


@pytest.mark.parametrize('optimizer',  ['radam', 'radabelief'])
def test_rectified(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=1e-3))


@pytest.mark.parametrize('optimizer',   ['adadelta', 'adagrad'])
def test_adaother(optimizer):
    _test_basic_cases(
        lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=1e-3, weight_decay=1)
    )
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-1)
    )
    _test_model(optimizer, dict(lr=5e-2))


@pytest.mark.parametrize('optimizer',   ['adafactor', 'adafactorbv'])
def test_adafactor(optimizer):
    _test_basic_cases(
        lambda weight, bias: create_optimizer_v2([weight, bias], optimizer, lr=1e-3, weight_decay=1)
    )
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=5e-2)
    )
    _test_model(optimizer, dict(lr=5e-2))


@pytest.mark.parametrize('optimizer',  ['lamb', 'lambc'])
def test_lamb(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=1e-3))


@pytest.mark.parametrize('optimizer', ['laprop'])
def test_laprop(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-2)
    )
    _test_model(optimizer, dict(lr=1e-2))


@pytest.mark.parametrize('optimizer',  ['lars', 'larc', 'nlars', 'nlarc'])
def test_lars(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=1e-3))


@pytest.mark.parametrize('optimizer',  ['madgrad', 'madgradw'])
def test_madgrad(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-2)
    )
    _test_model(optimizer, dict(lr=1e-2))


@pytest.mark.parametrize('optimizer',  ['mars'])
def test_mars(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=5e-2), after_step=1)  # note no convergence in first step for ADOPT


@pytest.mark.parametrize('optimizer',  ['novograd'])
def test_novograd(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=1e-3))


@pytest.mark.parametrize('optimizer', ['rmsprop', 'rmsproptf'])
def test_rmsprop(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-2)
    )
    _test_model(optimizer, dict(lr=1e-2))


@pytest.mark.parametrize('optimizer', ['adamp'])
def test_adamp(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=5e-2)
    )
    _test_model(optimizer, dict(lr=5e-2))


@pytest.mark.parametrize('optimizer', ['sgdp'])
def test_sgdp(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )
    _test_model(optimizer, dict(lr=1e-3))


@pytest.mark.parametrize('optimizer', ['lookahead_sgd', 'lookahead_momentum'])
def test_lookahead_sgd(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-3)
    )


@pytest.mark.parametrize('optimizer', ['lookahead_adamw', 'lookahead_adam'])
def test_lookahead_adam(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=5e-2)
    )


@pytest.mark.parametrize('optimizer', ['lookahead_radam'])
def test_lookahead_radam(optimizer):
    _test_rosenbrock(
        lambda params: create_optimizer_v2(params, optimizer, lr=1e-4)
    )


def test_param_groups_layer_decay_with_end_decay():
    model = torch.nn.Sequential(
        torch.nn.Linear(10, 5),
        torch.nn.ReLU(),
        torch.nn.Linear(5, 2)
    )
    
    param_groups = param_groups_layer_decay(
        model,
        weight_decay=0.05,
        layer_decay=0.75,
        end_layer_decay=0.5,
        verbose=True
    )
    
    assert len(param_groups) > 0
    # Verify layer scaling is applied with end decay
    for group in param_groups:
        assert 'lr_scale' in group
        assert group['lr_scale'] <= 1.0
        assert group['lr_scale'] >= 0.5


def test_param_groups_layer_decay_with_matcher():
    class ModelWithMatcher(torch.nn.Module):
        def __init__(self):
            super().__init__()
            self.layer1 = torch.nn.Linear(10, 5)
            self.layer2 = torch.nn.Linear(5, 2)
            
        def group_matcher(self, coarse=False):
            return lambda name: int(name.split('.')[0][-1])
            
    model = ModelWithMatcher()
    param_groups = param_groups_layer_decay(
        model,
        weight_decay=0.05,
        layer_decay=0.75,
        verbose=True
    )
    
    assert len(param_groups) > 0
    # Verify layer scaling is applied
    for group in param_groups:
        assert 'lr_scale' in group
        assert 'weight_decay' in group
        assert len(group['params']) > 0


def test_param_groups_weight_decay():
    model = torch.nn.Sequential(
        torch.nn.Linear(10, 5),
        torch.nn.ReLU(),
        torch.nn.Linear(5, 2)
    )
    weight_decay = 0.01
    no_weight_decay_list = ['1.weight']
    
    param_groups = param_groups_weight_decay(
        model, 
        weight_decay=weight_decay,
        no_weight_decay_list=no_weight_decay_list
    )
    
    assert len(param_groups) == 2
    assert param_groups[0]['weight_decay'] == 0.0
    assert param_groups[1]['weight_decay'] == weight_decay
    
    # Verify parameters are correctly grouped
    no_decay_params = set(param_groups[0]['params'])
    decay_params = set(param_groups[1]['params'])
    
    for name, param in model.named_parameters():
        if param.ndim <= 1 or name.endswith(".bias") or name in no_weight_decay_list:
            assert param in no_decay_params
        else:
            assert param in decay_params