prompt
stringlengths
94
42.6k
completion
stringlengths
6
120
api
stringlengths
14
68
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) orig_params = [] for param in mlp.parameters(): orig_params.append(np.copy(param.numpy())) opt.step() for param, orig_param in zip(mlp.parameters(), orig_params): assertTensorClose(param.numpy(), orig_param - param.grad.numpy() * 0.03) def test_adam(): data, data_shape, label, label_shape = get_input() mlp = MLP() beta0 = 0.8 beta1 = 0.9 eps = 1e-4 opt = Adam(mlp.parameters(), lr=0.01, betas=(beta0, beta1), eps=eps) m_slots =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) orig_params = [] for param in mlp.parameters(): orig_params.append(np.copy(param.numpy())) opt.step() for param, orig_param in zip(mlp.parameters(), orig_params): assertTensorClose(param.numpy(), orig_param - param.grad.numpy() * 0.03) def test_adam(): data, data_shape, label, label_shape = get_input() mlp = MLP() beta0 = 0.8 beta1 = 0.9 eps = 1e-4 opt = Adam(mlp.parameters(), lr=0.01, betas=(beta0, beta1), eps=eps) m_slots = TensorDict() v_slots =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) orig_params = [] for param in mlp.parameters(): orig_params.append(np.copy(param.numpy())) opt.step() for param, orig_param in zip(mlp.parameters(), orig_params): assertTensorClose(param.numpy(), orig_param - param.grad.numpy() * 0.03) def test_adam(): data, data_shape, label, label_shape = get_input() mlp = MLP() beta0 = 0.8 beta1 = 0.9 eps = 1e-4 opt = Adam(mlp.parameters(), lr=0.01, betas=(beta0, beta1), eps=eps) m_slots = TensorDict() v_slots = TensorDict() for param in mlp.parameters(): m_slots[param] = np.zeros(param.shape).astype(np.float32) v_slots[param] = np.zeros(param.shape).astype(np.float32) step_size = 0 def check_value(): for param in mlp.parameters(): grad = param.grad.numpy() orig_param = orig_params[param] m = m_slots[param] v = v_slots[param] m *= beta0 m += (1 - beta0) * grad v *= beta1 v += (1 - beta1) * grad * grad update = (m / (1 - beta0 ** step_size)) / ( np.sqrt(v / (1 - beta1 ** step_size)) + eps ) assertTensorClose(param.numpy(), orig_param - 0.01 * update) # eager for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() grads = opt.backward(loss) orig_params = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) opt.step() step_size += 1 check_value() # static @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.backward(loss) for _ in range(3): opt.zero_grad() orig_params = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) opt.step() step_size += 1 check_value() @graph_mode("eager", "static") def test_optimizer_serialization(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) orig_params = [] for param in mlp.parameters(): orig_params.append(np.copy(param.numpy())) opt.step() for param, orig_param in zip(mlp.parameters(), orig_params): assertTensorClose(param.numpy(), orig_param - param.grad.numpy() * 0.03) def test_adam(): data, data_shape, label, label_shape = get_input() mlp = MLP() beta0 = 0.8 beta1 = 0.9 eps = 1e-4 opt = Adam(mlp.parameters(), lr=0.01, betas=(beta0, beta1), eps=eps) m_slots = TensorDict() v_slots = TensorDict() for param in mlp.parameters(): m_slots[param] = np.zeros(param.shape).astype(np.float32) v_slots[param] = np.zeros(param.shape).astype(np.float32) step_size = 0 def check_value(): for param in mlp.parameters(): grad = param.grad.numpy() orig_param = orig_params[param] m = m_slots[param] v = v_slots[param] m *= beta0 m += (1 - beta0) * grad v *= beta1 v += (1 - beta1) * grad * grad update = (m / (1 - beta0 ** step_size)) / ( np.sqrt(v / (1 - beta1 ** step_size)) + eps ) assertTensorClose(param.numpy(), orig_param - 0.01 * update) # eager for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() grads = opt.backward(loss) orig_params =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) orig_params = [] for param in mlp.parameters(): orig_params.append(np.copy(param.numpy())) opt.step() for param, orig_param in zip(mlp.parameters(), orig_params): assertTensorClose(param.numpy(), orig_param - param.grad.numpy() * 0.03) def test_adam(): data, data_shape, label, label_shape = get_input() mlp = MLP() beta0 = 0.8 beta1 = 0.9 eps = 1e-4 opt = Adam(mlp.parameters(), lr=0.01, betas=(beta0, beta1), eps=eps) m_slots = TensorDict() v_slots = TensorDict() for param in mlp.parameters(): m_slots[param] = np.zeros(param.shape).astype(np.float32) v_slots[param] = np.zeros(param.shape).astype(np.float32) step_size = 0 def check_value(): for param in mlp.parameters(): grad = param.grad.numpy() orig_param = orig_params[param] m = m_slots[param] v = v_slots[param] m *= beta0 m += (1 - beta0) * grad v *= beta1 v += (1 - beta1) * grad * grad update = (m / (1 - beta0 ** step_size)) / ( np.sqrt(v / (1 - beta1 ** step_size)) + eps ) assertTensorClose(param.numpy(), orig_param - 0.01 * update) # eager for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() grads = opt.backward(loss) orig_params = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) opt.step() step_size += 1 check_value() # static @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.backward(loss) for _ in range(3): opt.zero_grad() orig_params =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) orig_params = [] for param in mlp.parameters(): orig_params.append(np.copy(param.numpy())) opt.step() for param, orig_param in zip(mlp.parameters(), orig_params): assertTensorClose(param.numpy(), orig_param - param.grad.numpy() * 0.03) def test_adam(): data, data_shape, label, label_shape = get_input() mlp = MLP() beta0 = 0.8 beta1 = 0.9 eps = 1e-4 opt = Adam(mlp.parameters(), lr=0.01, betas=(beta0, beta1), eps=eps) m_slots = TensorDict() v_slots = TensorDict() for param in mlp.parameters(): m_slots[param] = np.zeros(param.shape).astype(np.float32) v_slots[param] = np.zeros(param.shape).astype(np.float32) step_size = 0 def check_value(): for param in mlp.parameters(): grad = param.grad.numpy() orig_param = orig_params[param] m = m_slots[param] v = v_slots[param] m *= beta0 m += (1 - beta0) * grad v *= beta1 v += (1 - beta1) * grad * grad update = (m / (1 - beta0 ** step_size)) / ( np.sqrt(v / (1 - beta1 ** step_size)) + eps ) assertTensorClose(param.numpy(), orig_param - 0.01 * update) # eager for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() grads = opt.backward(loss) orig_params = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) opt.step() step_size += 1 check_value() # static @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.backward(loss) for _ in range(3): opt.zero_grad() orig_params = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) opt.step() step_size += 1 check_value() @graph_mode("eager", "static") def test_optimizer_serialization(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for param in mlp.parameters(): slot = slots[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 with BytesIO() as fout: save(opt.state_dict(), fout) fout.seek(0) state_dict =
load(fout)
megengine.load
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) orig_params = [] for param in mlp.parameters(): orig_params.append(np.copy(param.numpy())) opt.step() for param, orig_param in zip(mlp.parameters(), orig_params): assertTensorClose(param.numpy(), orig_param - param.grad.numpy() * 0.03) def test_adam(): data, data_shape, label, label_shape = get_input() mlp = MLP() beta0 = 0.8 beta1 = 0.9 eps = 1e-4 opt = Adam(mlp.parameters(), lr=0.01, betas=(beta0, beta1), eps=eps) m_slots = TensorDict() v_slots = TensorDict() for param in mlp.parameters(): m_slots[param] = np.zeros(param.shape).astype(np.float32) v_slots[param] = np.zeros(param.shape).astype(np.float32) step_size = 0 def check_value(): for param in mlp.parameters(): grad = param.grad.numpy() orig_param = orig_params[param] m = m_slots[param] v = v_slots[param] m *= beta0 m += (1 - beta0) * grad v *= beta1 v += (1 - beta1) * grad * grad update = (m / (1 - beta0 ** step_size)) / ( np.sqrt(v / (1 - beta1 ** step_size)) + eps ) assertTensorClose(param.numpy(), orig_param - 0.01 * update) # eager for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() grads = opt.backward(loss) orig_params = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) opt.step() step_size += 1 check_value() # static @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.backward(loss) for _ in range(3): opt.zero_grad() orig_params = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) opt.step() step_size += 1 check_value() @graph_mode("eager", "static") def test_optimizer_serialization(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for param in mlp.parameters(): slot = slots[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 with BytesIO() as fout: save(opt.state_dict(), fout) fout.seek(0) state_dict = load(fout) opt1 = SGD(mlp.parameters(), lr=0.02, momentum=0.8) opt1.load_state_dict(state_dict) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt1.zero_grad() opt1.backward(loss) orig_params =
TensorDict()
megengine.core.TensorDict
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad =
F.grad(loss, param, use_virtual_grad=False)
megengine.functional.grad
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. from io import BytesIO import numpy as np from helpers import MLP, graph_mode import megengine.functional as F from megengine import load, save from megengine.core import TensorDict, tensor from megengine.jit import trace from megengine.optimizer import SGD, Adam from megengine.test import assertTensorClose def get_input(): batch_size = 2 input_dim = 28 data_shape = (batch_size, input_dim) label_shape = (batch_size,) data = tensor() label = tensor(dtype=np.int32) data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) return data, data_shape, label, label_shape def test_sgd_simple(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, weight_decay=0.1) for idx in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) if idx % 2: opt.zero_grad() else: mlp.zero_grad() opt.backward(loss) grads = TensorDict() orig_params = TensorDict() for param in mlp.parameters(): grad = F.grad(loss, param, use_virtual_grad=False) assertTensorClose(grad.numpy(), param.grad.numpy()) grads[param] = np.copy(grad.numpy()) orig_params[param] = np.copy(param.numpy()) opt.step() for param in mlp.parameters(): assertTensorClose( param.numpy(), orig_params[param] * 0.999 - grads[param] * 0.01 ) def test_sgd_momentum(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) # TODO: put opt.step() inside trace def test_sgd_momentum_static(): _, data_shape, _, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01, momentum=0.9) @trace def f(data, label): pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) slots = TensorDict() for param in mlp.parameters(): slots[param] = np.zeros(param.shape).astype(np.float32) for _ in range(3): f( np.random.random(data_shape).astype(np.float32), np.random.randint(0, 10, label_shape).astype(np.int32), ) orig_params = TensorDict() grads = TensorDict() for param in mlp.parameters(): orig_params[param] = np.copy(param.numpy()) grads[param] = np.copy(param.grad.numpy()) opt.step() for param in mlp.parameters(): slot = slots[param] orig_param = orig_params[param] slot *= 0.9 slot -= param.grad.numpy() * 0.01 assertTensorClose(param.numpy(), orig_param + slot) def test_update_lr(): data, data_shape, label, label_shape = get_input() mlp = MLP() opt = SGD(mlp.parameters(), lr=0.01) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) opt.step() for group in opt.param_groups: group["lr"] += 0.02 for _ in range(3): data.set_value(np.random.random(data_shape).astype(np.float32)) label.set_value(np.random.randint(0, 10, label_shape)) pred = mlp(data) loss = F.square_loss(pred, label.reshape(-1, 1)) opt.zero_grad() opt.backward(loss) for param in mlp.parameters(): grad =
F.grad(loss, param, use_virtual_grad=False)
megengine.functional.grad
#!/usr/bin/env python # -*- encoding: utf-8 -*- # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. import megengine.functional as F import megengine.module as M from .darknet import Darknet from .network_blocks import BaseConv, UpSample class YOLOFPN(M.Module): """ YOLOFPN module. Darknet 53 is the default backbone of this model. """ def __init__( self, depth=53, in_features=["dark3", "dark4", "dark5"], ): super().__init__() self.backbone = Darknet(depth) self.in_features = in_features # out 1 self.out1_cbl = self._make_cbl(512, 256, 1) self.out1 = self._make_embedding([256, 512], 512 + 256) # out 2 self.out2_cbl = self._make_cbl(256, 128, 1) self.out2 = self._make_embedding([128, 256], 256 + 128) # upsample self.upsample = UpSample(scale_factor=2, mode="bilinear") def _make_cbl(self, _in, _out, ks): return BaseConv(_in, _out, ks, stride=1, act="lrelu") def _make_embedding(self, filters_list, in_filters): m = M.Sequential( *[ self._make_cbl(in_filters, filters_list[0], 1), self._make_cbl(filters_list[0], filters_list[1], 3), self._make_cbl(filters_list[1], filters_list[0], 1), self._make_cbl(filters_list[0], filters_list[1], 3), self._make_cbl(filters_list[1], filters_list[0], 1), ] ) return m def forward(self, inputs): """ Args: inputs (Tensor): input image. Returns: Tuple[Tensor]: FPN output features.. """ # backbone out_features = self.backbone(inputs) x2, x1, x0 = [out_features[f] for f in self.in_features] # yolo branch 1 x1_in = self.out1_cbl(x0) x1_in = self.upsample(x1_in) x1_in =
F.concat([x1_in, x1], 1)
megengine.functional.concat
#!/usr/bin/env python # -*- encoding: utf-8 -*- # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. import megengine.functional as F import megengine.module as M from .darknet import Darknet from .network_blocks import BaseConv, UpSample class YOLOFPN(M.Module): """ YOLOFPN module. Darknet 53 is the default backbone of this model. """ def __init__( self, depth=53, in_features=["dark3", "dark4", "dark5"], ): super().__init__() self.backbone = Darknet(depth) self.in_features = in_features # out 1 self.out1_cbl = self._make_cbl(512, 256, 1) self.out1 = self._make_embedding([256, 512], 512 + 256) # out 2 self.out2_cbl = self._make_cbl(256, 128, 1) self.out2 = self._make_embedding([128, 256], 256 + 128) # upsample self.upsample = UpSample(scale_factor=2, mode="bilinear") def _make_cbl(self, _in, _out, ks): return BaseConv(_in, _out, ks, stride=1, act="lrelu") def _make_embedding(self, filters_list, in_filters): m = M.Sequential( *[ self._make_cbl(in_filters, filters_list[0], 1), self._make_cbl(filters_list[0], filters_list[1], 3), self._make_cbl(filters_list[1], filters_list[0], 1), self._make_cbl(filters_list[0], filters_list[1], 3), self._make_cbl(filters_list[1], filters_list[0], 1), ] ) return m def forward(self, inputs): """ Args: inputs (Tensor): input image. Returns: Tuple[Tensor]: FPN output features.. """ # backbone out_features = self.backbone(inputs) x2, x1, x0 = [out_features[f] for f in self.in_features] # yolo branch 1 x1_in = self.out1_cbl(x0) x1_in = self.upsample(x1_in) x1_in = F.concat([x1_in, x1], 1) out_dark4 = self.out1(x1_in) # yolo branch 2 x2_in = self.out2_cbl(out_dark4) x2_in = self.upsample(x2_in) x2_in =
F.concat([x2_in, x2], 1)
megengine.functional.concat
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging =
megengine.logger.get_logger()
megengine.logger.get_logger
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @
amp.autocast(enabled=args.mode == "mp")
megengine.amp.autocast
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train()
Q.enable_observer(model)
megengine.quantization.enable_observer
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model)
Q.enable_fake_quant(model)
megengine.quantization.enable_fake_quant
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) loss.item() for step in range(0, steps): t = time.time() if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) objs.update(loss.item()) clck.update(time.time() - t) if step % args.print_freq == 0 and dist.get_rank() == 0: print( "Step {}, {}, {}".format( step, objs, clck, )) objs.reset() if
dist.get_rank()
megengine.distributed.get_rank
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm =
autodiff.GradManager()
megengine.autodiff.GradManager
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss =
F.nn.cross_entropy(logits, label, label_smooth=0.1)
megengine.functional.nn.cross_entropy
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step =
jit.trace(train_step, symbolic=False, symbolic_shape=False)
megengine.jit.trace
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image =
megengine.tensor(image, dtype="float32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label =
megengine.tensor(label, dtype="int32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) loss.item() for step in range(0, steps): t = time.time() if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image =
megengine.tensor(image, dtype="float32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) loss.item() for step in range(0, steps): t = time.time() if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label =
megengine.tensor(label, dtype="int32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=
dist.make_allreduce_cb("SUM")
megengine.distributed.make_allreduce_cb
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image =
megengine.tensor(image, dtype="float32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label =
megengine.tensor(label, dtype="int32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) loss.item() for step in range(0, steps): t = time.time() if args.loader: image,label = next(dataloader) if not args.preload: image =
megengine.tensor(image, dtype="float32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) loss.item() for step in range(0, steps): t = time.time() if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label =
megengine.tensor(label, dtype="int32")
megengine.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=jit.SublinearMemoryConfig(genetic_nr_iter=50), symbolic_shape=False) else: train_step = jit.trace(train_step, symbolic=False, symbolic_shape=False) else: assert args.symbolic==False, "invalid arguments: trace=Trace, symbolic=True" # start training objs = AverageMeter("Loss") clck = AverageMeter("Time") if args.loader: dataloader = iter(get_dataloader(args)) image,label = next(dataloader) else: image = np.random.randn(args.batch_size, 3, 224, 224).astype("float32") label = np.random.randint(0, 1000, size=(args.batch_size,)).astype("int32") # warm up for step in range(10): if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) loss.item() for step in range(0, steps): t = time.time() if args.loader: image,label = next(dataloader) if not args.preload: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") else: image = megengine.tensor(image, dtype="float32") label = megengine.tensor(label, dtype="int32") loss = train_step(image, label) objs.update(loss.item()) clck.update(time.time() - t) if step % args.print_freq == 0 and
dist.get_rank()
megengine.distributed.get_rank
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import os import time import numpy as np # pylint: disable=import-error import model as snet_model import quantizable_model as quantizable_snet_model import megengine import megengine.device as device import megengine.autodiff as autodiff import megengine.data as data import megengine.data.transform as T import megengine.distributed as dist import megengine.functional as F import megengine.optimizer as optim import megengine.jit as jit import megengine.amp as amp import megengine.quantization as Q logging = megengine.logger.get_logger() from dataset import get_dataloader DEFAULT_QAT_CONFIG = { "ema":Q.ema_fakequant_qconfig, "ema_lowbi":Q.ema_lowbit_fakequant_qconfig, "sync_ema":Q.sync_ema_fakequant_qconfig, "min_max":Q.min_max_fakequant_qconfig, "tqt":Q.tqt_qconfig } def get_qconifg(config_name: str): return DEFAULT_QAT_CONFIG[config_name] def main(): parser = argparse.ArgumentParser(description="shufflenet benchmark") parser.add_argument( "-a", "--arch", default="shufflenet_v2_x2_0", help="model architecture (default: shufflenet_v2_x1_0)", ) parser.add_argument( "-n", "--ngpus", default=1, type=int, help="number of GPUs per node (default: None, use all available GPUs)", ) parser.add_argument( "-s", "--steps", default=200, type=int, help="number of train steps (default: 200)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 128)", ) parser.add_argument( "--trace", action='store_true', default=False, help="whether use trace or not (default: False)", ) parser.add_argument( "--symbolic", action='store_true', default=False, help="whether use symbolic trace or not (default: False)", ) parser.add_argument( "--lr", metavar="LR", default=0.001, help="learning rate for single GPU (default: 0.001)", ) parser.add_argument("--momentum", default=0.9, help="momentum (default: 0.9)") parser.add_argument( "--weight-decay", default=4e-5, help="weight decay (default: 4e-5)" ) parser.add_argument( "-p", "--print-freq", default=1, type=int, metavar="N", help="print frequency (default: 1)", ) parser.add_argument( "-m", "--mode", default="normal", type=str, choices=["normal", "mp", "qat"], help="Quantization Mode\n" "normal: no quantization, using float32\n" "mp: input type is fp16\n" "qat: quantization aware training" ) parser.add_argument( "--qat-config", default="min_max", type=str, choices=["min_max", "ema", "ema_lowbit", "sync_ema", "tqt"], help="quantization aware training config\n" "min_max: min_max_fakequant_qconfig\n" "ema: ema_fakequant_qconfig\n" "ema_lowbit: ema_lowbit_fakequant_qconfig\n" "sync_ema: sync_ema_fakequant_qconfig\n" "tqt: tqt_qconfig" ) parser.add_argument("--dist-addr", default="localhost") parser.add_argument("--dist-port", type=int, default=0) parser.add_argument("--world-size", type=int, default=None) parser.add_argument("--rank", default=0) parser.add_argument("--loader", default=False, action="store_true", help="whether use loader") parser.add_argument("--preload", default=False, action="store_true", help="whether use preload") args = parser.parse_args() if args.world_size is None: args.world_size = args.ngpus if args.world_size > 1: # launch processes train_func = dist.launcher(worker, master_ip=args.dist_addr, port=args.dist_port, world_size=args.world_size, n_gpus=args.ngpus, rank_start=args.rank * args.ngpus) train_func(args) else: worker(args) def worker(args): steps = args.steps # build model shufflenet = quantizable_snet_model if args.mode == "qat" else snet_model model = shufflenet.__dict__[args.arch]() if args.mode == "qat": if args.qat_config == "sync_ema": assert args.ngpus > 1, "sync_ema does not support ngpus={}".format(args.ngpus) qconfig = get_qconifg(args.qat_config) model = Q.quantize_qat(module=model, qconfig= qconfig) model.train() Q.enable_observer(model) Q.enable_fake_quant(model) # Sync parameters if args.world_size > 1: dist.bcast_list_(model.parameters(), dist.WORLD) # Autodiff gradient manager gm = autodiff.GradManager().attach( model.parameters(), callbacks=dist.make_allreduce_cb("SUM") if args.world_size > 1 else None, ) # Optimizer params_wd = [] params_nwd = [] params_scale = [] for n, p in model.named_parameters(): if n.find("weight") >= 0 and len(p.shape) > 1: params_wd.append(p) elif n.find("scale") >= 0: params_scale.append(p) else: params_nwd.append(p) opt = optim.SGD( [{"params": params_wd}, {"params": params_nwd, "weight_decay": 0}, ], lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay * args.world_size, # scale weight decay in "SUM" mode ) # train and valid func @amp.autocast(enabled=args.mode == "mp") def train_step(image, label): with gm: logits = model(image) loss = F.nn.cross_entropy(logits, label, label_smooth=0.1) gm.backward(loss) opt.step().clear_grad() return loss if args.trace: if args.symbolic: train_step = jit.trace(train_step, symbolic=True, sublinear_memory_config=
jit.SublinearMemoryConfig(genetic_nr_iter=50)
megengine.jit.SublinearMemoryConfig
import os.path as osp from abc import ABCMeta, abstractmethod import megengine as mge import megengine.distributed as dist from megengine.optimizer.optimizer import Optimizer from megengine.module import Module from edit.utils import mkdir_or_exist, build_from_cfg, get_root_logger from ..hook import Hook, HOOKS, get_priority module_ckpt_suffix = "_module.mge" optim_ckpt_suffix = "_optim.mge" class BaseRunner(metaclass=ABCMeta): """The base class of Runner, a training helper for Mge. All subclasses should implement the following APIs: - ``run()`` - ``train()`` - ``test()`` - ``save_checkpoint()`` - ``resume()`` Args: model (:obj:`megengine.module.Module`): The model to be run. optimizers_cfg (dict): optimizer configs work_dir (str, optional): The working directory to save checkpoints and logs. Defaults to None. """ def __init__(self, model, optimizers_cfg=None, work_dir=None): assert hasattr(model, 'train_step') assert hasattr(model, 'test_step') assert hasattr(model, 'create_gradmanager_and_optimizers') assert hasattr(model, 'cal_for_eval') self.model = model self.optimizers_cfg = optimizers_cfg self.logger = get_root_logger() self.work_dir = work_dir assert self.work_dir is not None # get model name from the model class self._model_name = self.model.__class__.__name__ self.mode = None self._hooks = [] self._epoch = 0 self._iter = 0 self._inner_iter = 0 self._max_epochs = 0 self._max_iters = 0 @property def model_name(self): """str: Name of the model, usually the module class name.""" return self._model_name @property def hooks(self): """list[:obj:`Hook`]: A list of registered hooks.""" return self._hooks @property def epoch(self): """int: Current epoch.""" return self._epoch @property def iter(self): """int: Current iteration.""" return self._iter @property def inner_iter(self): """int: Iteration in an epoch.""" return self._inner_iter @property def max_epochs(self): """int: Maximum training epochs.""" return self._max_epochs @property def max_iters(self): """int: Maximum training iterations.""" return self._max_iters @abstractmethod def train(self, data_loader): pass @abstractmethod def test(self, data_loader): pass @abstractmethod def run(self, data_loaders, workflow, max_iters): pass @abstractmethod def save_checkpoint(self, out_dir, create_symlink=True): pass @abstractmethod def resume(self, path2checkpoint): pass @abstractmethod def register_training_hooks(self, lr_config, checkpoint_config, log_config): """Register default hooks for training. Default hooks include: - LrUpdaterHook - CheckpointSaverHook - log_config """ pass def create_gradmanager_and_optimizers(self): self.model.create_gradmanager_and_optimizers(self.optimizers_cfg) def sync_model_params(self): if
dist.is_distributed()
megengine.distributed.is_distributed
import os.path as osp from abc import ABCMeta, abstractmethod import megengine as mge import megengine.distributed as dist from megengine.optimizer.optimizer import Optimizer from megengine.module import Module from edit.utils import mkdir_or_exist, build_from_cfg, get_root_logger from ..hook import Hook, HOOKS, get_priority module_ckpt_suffix = "_module.mge" optim_ckpt_suffix = "_optim.mge" class BaseRunner(metaclass=ABCMeta): """The base class of Runner, a training helper for Mge. All subclasses should implement the following APIs: - ``run()`` - ``train()`` - ``test()`` - ``save_checkpoint()`` - ``resume()`` Args: model (:obj:`megengine.module.Module`): The model to be run. optimizers_cfg (dict): optimizer configs work_dir (str, optional): The working directory to save checkpoints and logs. Defaults to None. """ def __init__(self, model, optimizers_cfg=None, work_dir=None): assert hasattr(model, 'train_step') assert hasattr(model, 'test_step') assert hasattr(model, 'create_gradmanager_and_optimizers') assert hasattr(model, 'cal_for_eval') self.model = model self.optimizers_cfg = optimizers_cfg self.logger = get_root_logger() self.work_dir = work_dir assert self.work_dir is not None # get model name from the model class self._model_name = self.model.__class__.__name__ self.mode = None self._hooks = [] self._epoch = 0 self._iter = 0 self._inner_iter = 0 self._max_epochs = 0 self._max_iters = 0 @property def model_name(self): """str: Name of the model, usually the module class name.""" return self._model_name @property def hooks(self): """list[:obj:`Hook`]: A list of registered hooks.""" return self._hooks @property def epoch(self): """int: Current epoch.""" return self._epoch @property def iter(self): """int: Current iteration.""" return self._iter @property def inner_iter(self): """int: Iteration in an epoch.""" return self._inner_iter @property def max_epochs(self): """int: Maximum training epochs.""" return self._max_epochs @property def max_iters(self): """int: Maximum training iterations.""" return self._max_iters @abstractmethod def train(self, data_loader): pass @abstractmethod def test(self, data_loader): pass @abstractmethod def run(self, data_loaders, workflow, max_iters): pass @abstractmethod def save_checkpoint(self, out_dir, create_symlink=True): pass @abstractmethod def resume(self, path2checkpoint): pass @abstractmethod def register_training_hooks(self, lr_config, checkpoint_config, log_config): """Register default hooks for training. Default hooks include: - LrUpdaterHook - CheckpointSaverHook - log_config """ pass def create_gradmanager_and_optimizers(self): self.model.create_gradmanager_and_optimizers(self.optimizers_cfg) def sync_model_params(self): if dist.is_distributed(): self.logger.info("syncing the model's parameters...") dist.bcast_list_(self.model.parameters(), dist.WORLD) else: pass # do nothing def current_lr(self): """Get current learning rates. Returns: list[float] | dict[str, list[float]]: Current learning rates of all param groups. If the runner has a dict of optimizers, this method will return a dict. """ raise NotImplementedError("") # if isinstance(self.optimizer, Optimizer): # lr = [group['lr'] for group in self.optimizer.param_groups] # elif isinstance(self.optimizer, dict): # lr = dict() # for name, optim in self.optimizer.items(): # lr[name] = [group['lr'] for group in optim.param_groups] # else: # raise RuntimeError('lr is not applicable because optimizer does not exist.') # return lr def current_momentum(self): """Get current momentums. Returns: list[float] | dict[str, list[float]]: Current momentums of all param groups. If the runner has a dict of optimizers, this method will return a dict. """ raise NotImplementedError("") # def _get_momentum(optimizer): # momentums = [] # for group in optimizer.param_groups: # if 'momentum' in group.keys(): # momentums.append(group['momentum']) # elif 'betas' in group.keys(): # momentums.append(group['betas'][0]) # else: # momentums.append(0) # return momentums # # if self.optimizer is None: # raise RuntimeError('momentum is not applicable because optimizer does not exist.') # elif isinstance(self.optimizer, Optimizer): # momentums = _get_momentum(self.optimizer) # elif isinstance(self.optimizer, dict): # momentums = dict() # for name, optim in self.optimizer.items(): # momentums[name] = _get_momentum(optim) # return momentums def register_hook(self, hook, priority='NORMAL'): """Register a hook into the hook list. The hook will be inserted into a priority queue, with the specified priority (See :class:`Priority` for details of priorities). For hooks with the same priority, they will be triggered in the same order as they are registered. Args: hook (:obj:`Hook`): The hook to be registered. priority (int or str or :obj:`Priority`): Hook priority. Lower value means higher priority. """ assert isinstance(hook, Hook) if hasattr(hook, 'priority'): raise ValueError('"priority" is a reserved attribute for hook') priority = get_priority(priority) hook.priority = priority # insert the hook to a sorted list inserted = False for i in range(len(self._hooks) - 1, -1, -1): if priority >= self._hooks[i].priority: self._hooks.insert(i + 1, hook) inserted = True break if not inserted: self._hooks.insert(0, hook) def call_hook(self, fn_name): """Call all hooks. Args: fn_name (str): The function name in each hook to be called, such as "before_train_epoch". """ for hook in self._hooks: getattr(hook, fn_name)(self) def load_checkpoint(self, path2checkpoint, load_optim=True): """ :param path2checkpoint: e.g. workdirs/xxxxx/checkpoint/epoch_10 :return: dict """ assert osp.exists(path2checkpoint), "{} do not exist".format(path2checkpoint) dirname = osp.split(path2checkpoint)[-1] epoch, nums = dirname.split("_") assert epoch in ("epoch", ) self.logger.info('load checkpoint from {}'.format(path2checkpoint)) # 遍历model中的所有配置optimizer的model,并进行load res = dict() res['nums'] = int(nums) for submodule_name in self.optimizers_cfg.keys(): submodule = getattr(self.model, submodule_name, None) assert submodule is not None, "model should have submodule {}".format(submodule_name) assert isinstance(submodule, Module), "submodule should be instance of mge.module.Module" if
dist.get_rank()
megengine.distributed.get_rank
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl") def vgg11(**kwargs): model_args = dict(depths=[1, 1, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11_bn/vgg11_bn.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl") def vgg11(**kwargs): model_args = dict(depths=[1, 1, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11_bn/vgg11_bn.pkl") def vgg11_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg11(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13/vgg13.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl") def vgg11(**kwargs): model_args = dict(depths=[1, 1, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11_bn/vgg11_bn.pkl") def vgg11_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg11(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13/vgg13.pkl") def vgg13(**kwargs): model_args = dict(depths=[2, 2, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13_bn/vgg13_bn.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl") def vgg11(**kwargs): model_args = dict(depths=[1, 1, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11_bn/vgg11_bn.pkl") def vgg11_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg11(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13/vgg13.pkl") def vgg13(**kwargs): model_args = dict(depths=[2, 2, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13_bn/vgg13_bn.pkl") def vgg13_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg13(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg16/vgg16.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl") def vgg11(**kwargs): model_args = dict(depths=[1, 1, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11_bn/vgg11_bn.pkl") def vgg11_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg11(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13/vgg13.pkl") def vgg13(**kwargs): model_args = dict(depths=[2, 2, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13_bn/vgg13_bn.pkl") def vgg13_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg13(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg16/vgg16.pkl") def vgg16(**kwargs): model_args = dict(depths=[2, 2, 3, 3, 3], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg16_bn/vgg16_bn.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl") def vgg11(**kwargs): model_args = dict(depths=[1, 1, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11_bn/vgg11_bn.pkl") def vgg11_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg11(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13/vgg13.pkl") def vgg13(**kwargs): model_args = dict(depths=[2, 2, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13_bn/vgg13_bn.pkl") def vgg13_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg13(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg16/vgg16.pkl") def vgg16(**kwargs): model_args = dict(depths=[2, 2, 3, 3, 3], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg16_bn/vgg16_bn.pkl") def vgg16_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg16(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg19/vgg19.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool = M.MaxPool2d(kernel_size=2, stride=2) def __len__(self): return self.depth def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(self.depth): block = getattr(self, f"b{i + 1}") x = block(x) x = self.max_pool(x) return x @registers.models.register() class VGG(M.Module): """VGG model. Args: depths: depth for each stage (number of blocks in the stage). widths: width for each stage (width of each block in the stage). norm_name: normalization function. Default: ``None`` act_name: activation function. Default: ``"relu"`` head: head args. Default: ``None`` """ def __init__( self, depths: Sequence[int], widths: Sequence[int], norm_name: str = None, act_name: str = "relu", head: Mapping[str, Any] = None, ): super().__init__() self.depths = depths model_args = [depths, widths] prev_w = 3 for i, (d, w) in enumerate(zip(*model_args)): stage = VGGStage(prev_w, w, d, norm_name, act_name) setattr(self, f"s{i + 1}", stage) prev_w = w self.head = build_head(prev_w, head, None, act_name) self.apply(init_weights) def forward(self, x: mge.Tensor) -> mge.Tensor: for i in range(len(self.depths)): stage = getattr(self, f"s{i + 1}") x = stage(x) if getattr(self, "head", None) is not None: x = self.head(x) return x def _build_vgg(**kwargs): model_args = dict(head=dict(name="VGGHead", dropout_prob=0.5)) recursive_update(model_args, kwargs) return VGG(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11/vgg11.pkl") def vgg11(**kwargs): model_args = dict(depths=[1, 1, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg11_bn/vgg11_bn.pkl") def vgg11_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg11(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13/vgg13.pkl") def vgg13(**kwargs): model_args = dict(depths=[2, 2, 2, 2, 2], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg13_bn/vgg13_bn.pkl") def vgg13_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg13(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg16/vgg16.pkl") def vgg16(**kwargs): model_args = dict(depths=[2, 2, 3, 3, 3], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg16_bn/vgg16_bn.pkl") def vgg16_bn(**kwargs): model_args = dict(norm_name="BN") recursive_update(model_args, kwargs) return vgg16(**model_args) @registers.models.register() @hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg19/vgg19.pkl") def vgg19(**kwargs): model_args = dict(depths=[2, 2, 4, 4, 4], widths=[64, 128, 256, 512, 512]) recursive_update(model_args, kwargs) return _build_vgg(**model_args) @registers.models.register() @
hub.pretrained("https://data.megengine.org.cn/research/basecls/models/vgg/vgg19_bn/vgg19_bn.pkl")
megengine.hub.pretrained
#!/usr/bin/env python3 # Copyright (c) 2014-2021 Megvii Inc. All rights reserved. """VGG Series VGG: `"Very Deep Convolutional Networks for Large-Scale Image Recognition" <https://arxiv.org/abs/1409.1556>`_ """ from typing import Any, Mapping, Sequence import megengine as mge import megengine.hub as hub import megengine.module as M from basecls.layers import activation, build_head, conv2d, init_weights, norm2d from basecls.utils import recursive_update, registers __all__ = ["VGGStage", "VGG"] class VGGStage(M.Module): """VGG stage (sequence of blocks w/ the same output shape).""" def __init__(self, w_in: int, w_out: int, depth: int, norm_name: str, act_name: str): super().__init__() self.depth = depth for i in range(depth): block = M.Sequential( conv2d(w_in, w_out, 3), norm2d(norm_name, w_out), activation(act_name) ) setattr(self, f"b{i + 1}", block) w_in = w_out self.max_pool =
M.MaxPool2d(kernel_size=2, stride=2)
megengine.module.MaxPool2d
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return
mge_get_logger(*args)
megengine.get_logger
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return
cgtools.get_dep_vars(x, type)
megengine.utils.comp_graph_tools.get_dep_vars
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return
cgtools.get_opr_type(x)
megengine.utils.comp_graph_tools.get_opr_type
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs = mgb.load_comp_graph_from_file(path) else: ret = G.load_graph(path) cg = ret.graph outputs = ret.output_vars_list return cg, outputs def graph_traversal(outputs): ( map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree, ) =
cgtools.graph_traversal(outputs)
megengine.utils.comp_graph_tools.graph_traversal
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs = mgb.load_comp_graph_from_file(path) else: ret = G.load_graph(path) cg = ret.graph outputs = ret.output_vars_list return cg, outputs def graph_traversal(outputs): ( map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree, ) = cgtools.graph_traversal(outputs) return map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree def get_oprs_seq(outputs, prune_reshape=True): all_oprs =
cgtools.get_oprs_seq(outputs, prune_reshape=prune_reshape)
megengine.utils.comp_graph_tools.get_oprs_seq
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs = mgb.load_comp_graph_from_file(path) else: ret = G.load_graph(path) cg = ret.graph outputs = ret.output_vars_list return cg, outputs def graph_traversal(outputs): ( map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree, ) = cgtools.graph_traversal(outputs) return map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree def get_oprs_seq(outputs, prune_reshape=True): all_oprs = cgtools.get_oprs_seq(outputs, prune_reshape=prune_reshape) return all_oprs def eval_partial(inp, oup): if not isinstance(oup, (list, tuple)): oup = (oup,) inputs =
cgtools.get_dep_vars(oup, "Host2DeviceCopy")
megengine.utils.comp_graph_tools.get_dep_vars
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return
cgtools.get_type(x._var)
megengine.utils.comp_graph_tools.get_type
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return
cgtools.get_owner_opr_type(x._var)
megengine.utils.comp_graph_tools.get_owner_opr_type
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs =
mgb.load_comp_graph_from_file(path)
megengine._internal.load_comp_graph_from_file
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs = mgb.load_comp_graph_from_file(path) else: ret =
G.load_graph(path)
megengine.core.tensor.megbrain_graph.load_graph
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs = mgb.load_comp_graph_from_file(path) else: ret = G.load_graph(path) cg = ret.graph outputs = ret.output_vars_list return cg, outputs def graph_traversal(outputs): ( map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree, ) = cgtools.graph_traversal(outputs) return map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree def get_oprs_seq(outputs, prune_reshape=True): all_oprs = cgtools.get_oprs_seq(outputs, prune_reshape=prune_reshape) return all_oprs def eval_partial(inp, oup): if not isinstance(oup, (list, tuple)): oup = (oup,) inputs = cgtools.get_dep_vars(oup, "Host2DeviceCopy") if mge_version <= "0.6.0": cg = oup[0].owner_graph outputs = list(map(mgb.copy_output, oup)) f = cg.compile(inputs, outputs) result = f(inp) else: if not isinstance(inp, (list, tuple)): inp = (inp,) replace_dict = {} inp_node_list = [] for i in inputs: inp_node = G.InputNode( device="xpux", dtype=inputs[0].dtype, graph=inputs[0].graph ) replace_dict[i] = inp_node.outputs[0] inp_node_list.append(inp_node) new_out =
cgtools.replace_vars(oup, replace_dict)
megengine.utils.comp_graph_tools.replace_vars
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node =
G.ValueOutputNode(sym_var)
megengine.core.tensor.megbrain_graph.ValueOutputNode
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs = mgb.load_comp_graph_from_file(path) else: ret = G.load_graph(path) cg = ret.graph outputs = ret.output_vars_list return cg, outputs def graph_traversal(outputs): ( map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree, ) = cgtools.graph_traversal(outputs) return map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree def get_oprs_seq(outputs, prune_reshape=True): all_oprs = cgtools.get_oprs_seq(outputs, prune_reshape=prune_reshape) return all_oprs def eval_partial(inp, oup): if not isinstance(oup, (list, tuple)): oup = (oup,) inputs = cgtools.get_dep_vars(oup, "Host2DeviceCopy") if mge_version <= "0.6.0": cg = oup[0].owner_graph outputs = list(map(mgb.copy_output, oup)) f = cg.compile(inputs, outputs) result = f(inp) else: if not isinstance(inp, (list, tuple)): inp = (inp,) replace_dict = {} inp_node_list = [] for i in inputs: inp_node = G.InputNode( device="xpux", dtype=inputs[0].dtype, graph=inputs[0].graph ) replace_dict[i] = inp_node.outputs[0] inp_node_list.append(inp_node) new_out = cgtools.replace_vars(oup, replace_dict) out_node_list = [
G.OutputNode(i)
megengine.core.tensor.megbrain_graph.OutputNode
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import megengine as mge import numpy as np from megengine import get_logger as mge_get_logger mge_version = mge.__version__ if mge_version <= "0.6.0": # pylint: disable=import-error, no-name-in-module import megengine._internal as mgb from megengine._internal import cgtools else: import megengine.core.tensor.megbrain_graph as G import megengine.core._imperative_rt as rt import megengine.utils.comp_graph_tools as cgtools if mge_version <= "1.1.0": from megengine.core.tensor.raw_tensor import ( # pylint: disable=no-name-in-module,import-error as_raw_tensor as Tensor, ) else: from megengine.tensor import Tensor def get_logger(*args): return mge_get_logger(*args) def get_mge_version(): return mge_version def get_symvar_value(sym_var): if mge_version <= "0.6.0": if sym_var.inferred_value is not None: val = sym_var.inferred_value return val else: cg = sym_var.owner_graph func = cg.compile_outonly(sym_var) val = func() return val else: if sym_var.value is not None: return sym_var.value else: out_node = G.ValueOutputNode(sym_var) cg = out_node.outputs[0].graph func = cg.compile(out_node.outputs) func.execute() return out_node.get_value() def isnum(x): return isinstance(x, (int, float)) def isconst(x): return x.np_data is not None def isvar(x): return ( isinstance(x, mgb.SymbolVar) if mge_version <= "0.6.0" else isinstance(x, rt.VarNode) # pylint: disable=c-extension-no-member ) def get_shape(x): return x._get_imm_shape() if mge_version <= "0.6.0" else x.shape def get_dep_vars(x, type=None): return cgtools.get_dep_vars(x, type) def get_dtype_name(x): return ( x.dtype.metadata["mgb_dtype"]["name"] if isinstance(x.dtype, np.dtype) else None ) def get_opr_type(x): return cgtools.get_opr_type(x) def get_owner_opr_type(x): if mge_version <= "0.6.0": return cgtools.get_type(x._var) else: return cgtools.get_owner_opr_type(x._var) def load_comp_graph_from_file(path): if mge_version <= "0.6.0": cg, _, outputs = mgb.load_comp_graph_from_file(path) else: ret = G.load_graph(path) cg = ret.graph outputs = ret.output_vars_list return cg, outputs def graph_traversal(outputs): ( map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree, ) = cgtools.graph_traversal(outputs) return map_oprs, map_vars, var2oprs, opr2receivers, indegree2opr, opr2indegree def get_oprs_seq(outputs, prune_reshape=True): all_oprs = cgtools.get_oprs_seq(outputs, prune_reshape=prune_reshape) return all_oprs def eval_partial(inp, oup): if not isinstance(oup, (list, tuple)): oup = (oup,) inputs = cgtools.get_dep_vars(oup, "Host2DeviceCopy") if mge_version <= "0.6.0": cg = oup[0].owner_graph outputs = list(map(mgb.copy_output, oup)) f = cg.compile(inputs, outputs) result = f(inp) else: if not isinstance(inp, (list, tuple)): inp = (inp,) replace_dict = {} inp_node_list = [] for i in inputs: inp_node = G.InputNode( device="xpux", dtype=inputs[0].dtype, graph=inputs[0].graph ) replace_dict[i] = inp_node.outputs[0] inp_node_list.append(inp_node) new_out = cgtools.replace_vars(oup, replace_dict) out_node_list = [G.OutputNode(i) for i in new_out] new_out_list = [i.outputs[0] for i in out_node_list] cg = new_out_list[0].graph func = cg.compile(new_out_list) for node, value in zip(inp_node_list, inp): node.set_value(
Tensor(value)
megengine.tensor.Tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn =
BatchNorm1d(nr_chan, momentum=momentum)
megengine.module.BatchNorm1d
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn = BatchNorm1d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1), dtype=np.float32) data =
tensor()
megengine.core.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn = BatchNorm1d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) xv_transposed = np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose( running_mean.reshape(-1), bn.running_mean.numpy().reshape(-1), max_err=5e-6 ) assertTensorClose( running_var.reshape(-1), bn.running_var.numpy().reshape(-1), max_err=5e-6 ) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm2d(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) momentum = 0.9 bn =
BatchNorm2d(nr_chan, momentum=momentum)
megengine.module.BatchNorm2d
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn = BatchNorm1d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) xv_transposed = np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose( running_mean.reshape(-1), bn.running_mean.numpy().reshape(-1), max_err=5e-6 ) assertTensorClose( running_var.reshape(-1), bn.running_var.numpy().reshape(-1), max_err=5e-6 ) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm2d(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) momentum = 0.9 bn = BatchNorm2d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1, 1), dtype=np.float32) data =
tensor()
megengine.core.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn = BatchNorm1d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) xv_transposed = np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose( running_mean.reshape(-1), bn.running_mean.numpy().reshape(-1), max_err=5e-6 ) assertTensorClose( running_var.reshape(-1), bn.running_var.numpy().reshape(-1), max_err=5e-6 ) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm2d(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) momentum = 0.9 bn = BatchNorm2d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) xv_transposed = np.transpose(xv, [0, 2, 3, 1]).reshape( (data_shape[0] * data_shape[2] * data_shape[3], nr_chan) ) mean = np.mean(xv_transposed, axis=0).reshape(1, nr_chan, 1, 1) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose(running_mean, bn.running_mean.numpy(), max_err=5e-6) assertTensorClose(running_var, bn.running_var.numpy(), max_err=5e-6) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm_no_stats(): nr_chan = 8 data_shape = (3, nr_chan, 4) bn =
BatchNorm1d(8, track_running_stats=False)
megengine.module.BatchNorm1d
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn = BatchNorm1d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) xv_transposed = np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose( running_mean.reshape(-1), bn.running_mean.numpy().reshape(-1), max_err=5e-6 ) assertTensorClose( running_var.reshape(-1), bn.running_var.numpy().reshape(-1), max_err=5e-6 ) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm2d(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) momentum = 0.9 bn = BatchNorm2d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) xv_transposed = np.transpose(xv, [0, 2, 3, 1]).reshape( (data_shape[0] * data_shape[2] * data_shape[3], nr_chan) ) mean = np.mean(xv_transposed, axis=0).reshape(1, nr_chan, 1, 1) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose(running_mean, bn.running_mean.numpy(), max_err=5e-6) assertTensorClose(running_var, bn.running_var.numpy(), max_err=5e-6) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm_no_stats(): nr_chan = 8 data_shape = (3, nr_chan, 4) bn = BatchNorm1d(8, track_running_stats=False) data =
tensor()
megengine.core.tensor
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn = BatchNorm1d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) xv_transposed = np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose( running_mean.reshape(-1), bn.running_mean.numpy().reshape(-1), max_err=5e-6 ) assertTensorClose( running_var.reshape(-1), bn.running_var.numpy().reshape(-1), max_err=5e-6 ) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm2d(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) momentum = 0.9 bn = BatchNorm2d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) xv_transposed = np.transpose(xv, [0, 2, 3, 1]).reshape( (data_shape[0] * data_shape[2] * data_shape[3], nr_chan) ) mean = np.mean(xv_transposed, axis=0).reshape(1, nr_chan, 1, 1) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose(running_mean, bn.running_mean.numpy(), max_err=5e-6) assertTensorClose(running_var, bn.running_var.numpy(), max_err=5e-6) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm_no_stats(): nr_chan = 8 data_shape = (3, nr_chan, 4) bn = BatchNorm1d(8, track_running_stats=False) data = tensor() for i in range(4): if i == 2: bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) var = np.var( np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ), axis=0, ).reshape((1, nr_chan, 1)) sd = np.sqrt(var + bn.eps) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) def test_batchnorm2d_no_stats(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) bn =
BatchNorm2d(8, track_running_stats=False)
megengine.module.BatchNorm2d
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import numpy as np import pytest import megengine as mge from megengine.core import tensor from megengine.module import BatchNorm1d, BatchNorm2d from megengine.test import assertTensorClose def test_batchnorm(): nr_chan = 8 data_shape = (3, nr_chan, 4) momentum = 0.9 bn = BatchNorm1d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) xv_transposed = np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose( running_mean.reshape(-1), bn.running_mean.numpy().reshape(-1), max_err=5e-6 ) assertTensorClose( running_var.reshape(-1), bn.running_var.numpy().reshape(-1), max_err=5e-6 ) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm2d(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) momentum = 0.9 bn = BatchNorm2d(nr_chan, momentum=momentum) running_mean = np.zeros((1, nr_chan, 1, 1), dtype=np.float32) running_var = np.ones((1, nr_chan, 1, 1), dtype=np.float32) data = tensor() for i in range(3): xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) xv_transposed = np.transpose(xv, [0, 2, 3, 1]).reshape( (data_shape[0] * data_shape[2] * data_shape[3], nr_chan) ) mean = np.mean(xv_transposed, axis=0).reshape(1, nr_chan, 1, 1) var_biased = np.var(xv_transposed, axis=0).reshape((1, nr_chan, 1, 1)) sd = np.sqrt(var_biased + bn.eps) var_unbiased = np.var(xv_transposed, axis=0, ddof=1).reshape((1, nr_chan, 1, 1)) running_mean = running_mean * momentum + mean * (1 - momentum) running_var = running_var * momentum + var_unbiased * (1 - momentum) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) assertTensorClose(running_mean, bn.running_mean.numpy(), max_err=5e-6) assertTensorClose(running_var, bn.running_var.numpy(), max_err=5e-6) # test set 'training' flag to False mean_backup = bn.running_mean.numpy() var_backup = bn.running_var.numpy() bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) data.set_value(xv) yv1 = bn(data) yv2 = bn(data) assertTensorClose(yv1.numpy(), yv2.numpy(), max_err=0) assertTensorClose(mean_backup, bn.running_mean.numpy(), max_err=0) assertTensorClose(var_backup, bn.running_var.numpy(), max_err=0) yv_expect = (xv - running_mean) / np.sqrt(running_var + bn.eps) assertTensorClose(yv_expect, yv1.numpy(), max_err=5e-6) def test_batchnorm_no_stats(): nr_chan = 8 data_shape = (3, nr_chan, 4) bn = BatchNorm1d(8, track_running_stats=False) data = tensor() for i in range(4): if i == 2: bn.training = False xv = np.random.normal(loc=2.3, size=data_shape).astype(np.float32) mean = np.mean(np.mean(xv, axis=0, keepdims=True), axis=2, keepdims=True) var = np.var( np.transpose(xv, [0, 2, 1]).reshape( (data_shape[0] * data_shape[2], nr_chan) ), axis=0, ).reshape((1, nr_chan, 1)) sd = np.sqrt(var + bn.eps) data.set_value(xv) yv = bn(data) yv_expect = (xv - mean) / sd assertTensorClose(yv_expect, yv.numpy(), max_err=5e-6) def test_batchnorm2d_no_stats(): nr_chan = 8 data_shape = (3, nr_chan, 16, 16) bn = BatchNorm2d(8, track_running_stats=False) data =
tensor()
megengine.core.tensor
#!/usr/bin/env python3 # -*- coding:utf-8 -*- import megengine as mge import megengine.module as M from models.yolo_fpn import YOLOFPN from models.yolo_head import YOLOXHead from models.yolo_pafpn import YOLOPAFPN from models.yolox import YOLOX def build_yolox(name="yolox-s"): num_classes = 80 # value meaning: depth, width param_dict = { "yolox-nano": (0.33, 0.25), "yolox-tiny": (0.33, 0.375), "yolox-s": (0.33, 0.50), "yolox-m": (0.67, 0.75), "yolox-l": (1.0, 1.0), "yolox-x": (1.33, 1.25), } if name == "yolov3": depth = 1.0 width = 1.0 backbone = YOLOFPN() head = YOLOXHead(num_classes, width, in_channels=[128, 256, 512], act="lrelu") model = YOLOX(backbone, head) else: assert name in param_dict kwargs = {} depth, width = param_dict[name] if name == "yolox-nano": kwargs["depthwise"] = True in_channels = [256, 512, 1024] backbone = YOLOPAFPN(depth, width, in_channels=in_channels, **kwargs) head = YOLOXHead(num_classes, width, in_channels=in_channels, **kwargs) model = YOLOX(backbone, head) for m in model.modules(): if isinstance(m, M.BatchNorm2d): m.eps = 1e-3 return model def build_and_load(weight_file, name="yolox-s"): model = build_yolox(name) model_weights =
mge.load(weight_file)
megengine.load
import numpy as np import argparse from datetime import datetime import time import model as resnet_model import megengine as mge import megengine.autodiff as ad import megengine.functional as F import megengine.optimizer as optim parser = argparse.ArgumentParser(description="MegEngine ResNet Training") parser.add_argument( "-a", "--arch", default="resnet50", help="model architecture (default: resnet50)", ) parser.add_argument( "--steps", default=10, type=int, help="number of total steps to run (default: 10)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 64)", ) parser.add_argument( "--enable-dtr", dest="enable_dtr", action="store_true", help="Enable DTR") parser.add_argument( "--memory-budget", dest="mem_budget", default=5, type=int, help="memory budget for DTR, measured in GB (default: 5)", ) args = parser.parse_args() if args.enable_dtr: from megengine.utils.dtr import DTR ds =
DTR(memory_budget=args.mem_budget*1024**3)
megengine.utils.dtr.DTR
import numpy as np import argparse from datetime import datetime import time import model as resnet_model import megengine as mge import megengine.autodiff as ad import megengine.functional as F import megengine.optimizer as optim parser = argparse.ArgumentParser(description="MegEngine ResNet Training") parser.add_argument( "-a", "--arch", default="resnet50", help="model architecture (default: resnet50)", ) parser.add_argument( "--steps", default=10, type=int, help="number of total steps to run (default: 10)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 64)", ) parser.add_argument( "--enable-dtr", dest="enable_dtr", action="store_true", help="Enable DTR") parser.add_argument( "--memory-budget", dest="mem_budget", default=5, type=int, help="memory budget for DTR, measured in GB (default: 5)", ) args = parser.parse_args() if args.enable_dtr: from megengine.utils.dtr import DTR ds = DTR(memory_budget=args.mem_budget*1024**3) batch_size = args.batch_size image = mge.tensor(np.random.random((batch_size, 3, 224, 224))) label = mge.tensor(np.random.randint(100, size=(batch_size,))) #model = resnet_model.__dict__["resnet50"]() model = resnet_model.__dict__[args.arch]() gm=
ad.GradManager()
megengine.autodiff.GradManager
import numpy as np import argparse from datetime import datetime import time import model as resnet_model import megengine as mge import megengine.autodiff as ad import megengine.functional as F import megengine.optimizer as optim parser = argparse.ArgumentParser(description="MegEngine ResNet Training") parser.add_argument( "-a", "--arch", default="resnet50", help="model architecture (default: resnet50)", ) parser.add_argument( "--steps", default=10, type=int, help="number of total steps to run (default: 10)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 64)", ) parser.add_argument( "--enable-dtr", dest="enable_dtr", action="store_true", help="Enable DTR") parser.add_argument( "--memory-budget", dest="mem_budget", default=5, type=int, help="memory budget for DTR, measured in GB (default: 5)", ) args = parser.parse_args() if args.enable_dtr: from megengine.utils.dtr import DTR ds = DTR(memory_budget=args.mem_budget*1024**3) batch_size = args.batch_size image = mge.tensor(np.random.random((batch_size, 3, 224, 224))) label = mge.tensor(np.random.randint(100, size=(batch_size,))) #model = resnet_model.__dict__["resnet50"]() model = resnet_model.__dict__[args.arch]() gm=ad.GradManager().attach(model.parameters()) opt=optim.SGD(model.parameters(), lr=0.0125, momentum=0.9, weight_decay=1e-4) # miliseconds print(datetime.now().timetz()) time_list = [] cur_time = int(round(time.time()*1000)) for i in range(args.steps): with gm: logits=model(image) loss=
F.nn.cross_entropy(logits, label)
megengine.functional.nn.cross_entropy
import numpy as np import argparse from datetime import datetime import time import model as resnet_model import megengine as mge import megengine.autodiff as ad import megengine.functional as F import megengine.optimizer as optim parser = argparse.ArgumentParser(description="MegEngine ResNet Training") parser.add_argument( "-a", "--arch", default="resnet50", help="model architecture (default: resnet50)", ) parser.add_argument( "--steps", default=10, type=int, help="number of total steps to run (default: 10)", ) parser.add_argument( "-b", "--batch-size", metavar="SIZE", default=64, type=int, help="batch size for single GPU (default: 64)", ) parser.add_argument( "--enable-dtr", dest="enable_dtr", action="store_true", help="Enable DTR") parser.add_argument( "--memory-budget", dest="mem_budget", default=5, type=int, help="memory budget for DTR, measured in GB (default: 5)", ) args = parser.parse_args() if args.enable_dtr: from megengine.utils.dtr import DTR ds = DTR(memory_budget=args.mem_budget*1024**3) batch_size = args.batch_size image = mge.tensor(np.random.random((batch_size, 3, 224, 224))) label = mge.tensor(np.random.randint(100, size=(batch_size,))) #model = resnet_model.__dict__["resnet50"]() model = resnet_model.__dict__[args.arch]() gm=ad.GradManager().attach(model.parameters()) opt=optim.SGD(model.parameters(), lr=0.0125, momentum=0.9, weight_decay=1e-4) # miliseconds print(datetime.now().timetz()) time_list = [] cur_time = int(round(time.time()*1000)) for i in range(args.steps): with gm: logits=model(image) loss=F.nn.cross_entropy(logits, label) gm.backward(loss) total, free =
mge.get_mem_status_bytes()
megengine.get_mem_status_bytes
import numpy as np import megengine.functional as F import megengine.module as M from config import config from .anchors_generator import AnchorGenerator from .find_top_rpn_proposals import find_top_rpn_proposals from .fpn_anchor_target import fpn_anchor_target, fpn_rpn_reshape from det_opr.loss_opr import softmax_loss, smooth_l1_loss_rpn import pdb class RPN(M.Module): def __init__(self, rpn_channel=256): super().__init__() self.anchors_generator = AnchorGenerator( config.anchor_base_size, config.anchor_aspect_ratios, config.anchor_base_scale) self.rpn_conv =
M.Conv2d(256, rpn_channel, kernel_size=3, stride=1, padding=1)
megengine.module.Conv2d
import numpy as np import megengine.functional as F import megengine.module as M from config import config from .anchors_generator import AnchorGenerator from .find_top_rpn_proposals import find_top_rpn_proposals from .fpn_anchor_target import fpn_anchor_target, fpn_rpn_reshape from det_opr.loss_opr import softmax_loss, smooth_l1_loss_rpn import pdb class RPN(M.Module): def __init__(self, rpn_channel=256): super().__init__() self.anchors_generator = AnchorGenerator( config.anchor_base_size, config.anchor_aspect_ratios, config.anchor_base_scale) self.rpn_conv = M.Conv2d(256, rpn_channel, kernel_size=3, stride=1, padding=1) self.rpn_cls_score =
M.Conv2d(rpn_channel, config.num_cell_anchors * 2, kernel_size=1, stride=1)
megengine.module.Conv2d
import numpy as np import megengine.functional as F import megengine.module as M from config import config from .anchors_generator import AnchorGenerator from .find_top_rpn_proposals import find_top_rpn_proposals from .fpn_anchor_target import fpn_anchor_target, fpn_rpn_reshape from det_opr.loss_opr import softmax_loss, smooth_l1_loss_rpn import pdb class RPN(M.Module): def __init__(self, rpn_channel=256): super().__init__() self.anchors_generator = AnchorGenerator( config.anchor_base_size, config.anchor_aspect_ratios, config.anchor_base_scale) self.rpn_conv = M.Conv2d(256, rpn_channel, kernel_size=3, stride=1, padding=1) self.rpn_cls_score = M.Conv2d(rpn_channel, config.num_cell_anchors * 2, kernel_size=1, stride=1) self.rpn_bbox_offsets =
M.Conv2d(rpn_channel, config.num_cell_anchors * 4, kernel_size=1, stride=1)
megengine.module.Conv2d
import numpy as np import megengine.functional as F import megengine.module as M from config import config from .anchors_generator import AnchorGenerator from .find_top_rpn_proposals import find_top_rpn_proposals from .fpn_anchor_target import fpn_anchor_target, fpn_rpn_reshape from det_opr.loss_opr import softmax_loss, smooth_l1_loss_rpn import pdb class RPN(M.Module): def __init__(self, rpn_channel=256): super().__init__() self.anchors_generator = AnchorGenerator( config.anchor_base_size, config.anchor_aspect_ratios, config.anchor_base_scale) self.rpn_conv = M.Conv2d(256, rpn_channel, kernel_size=3, stride=1, padding=1) self.rpn_cls_score = M.Conv2d(rpn_channel, config.num_cell_anchors * 2, kernel_size=1, stride=1) self.rpn_bbox_offsets = M.Conv2d(rpn_channel, config.num_cell_anchors * 4, kernel_size=1, stride=1) for l in [self.rpn_conv, self.rpn_cls_score, self.rpn_bbox_offsets]:
M.init.normal_(l.weight, std=0.01)
megengine.module.init.normal_
import numpy as np import megengine.functional as F import megengine.module as M from config import config from .anchors_generator import AnchorGenerator from .find_top_rpn_proposals import find_top_rpn_proposals from .fpn_anchor_target import fpn_anchor_target, fpn_rpn_reshape from det_opr.loss_opr import softmax_loss, smooth_l1_loss_rpn import pdb class RPN(M.Module): def __init__(self, rpn_channel=256): super().__init__() self.anchors_generator = AnchorGenerator( config.anchor_base_size, config.anchor_aspect_ratios, config.anchor_base_scale) self.rpn_conv = M.Conv2d(256, rpn_channel, kernel_size=3, stride=1, padding=1) self.rpn_cls_score = M.Conv2d(rpn_channel, config.num_cell_anchors * 2, kernel_size=1, stride=1) self.rpn_bbox_offsets = M.Conv2d(rpn_channel, config.num_cell_anchors * 4, kernel_size=1, stride=1) for l in [self.rpn_conv, self.rpn_cls_score, self.rpn_bbox_offsets]: M.init.normal_(l.weight, std=0.01)
M.init.fill_(l.bias, 0)
megengine.module.init.fill_
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc2(x) return x def generate_eager_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data_inp = tensor(np.zeros(data_shape), dtype=np.float32) label_inp = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) def step(data, label): opt.zero_grad() data_inp.set_value(data) label_inp.set_value(label) pred = net(data_inp) loss = cross_entropy_with_softmax(pred, label_inp) opt.backward(loss) opt.step() return loss.numpy()[0] return step def generate_static_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data = tensor(np.zeros(data_shape), dtype=np.float32) label = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) # Save state to reset parameters later. state = copy.deepcopy(net.state_dict()) # Evaluate network in eager mode once. pred = net(data) loss =
cross_entropy_with_softmax(pred, label)
megengine.functional.cross_entropy_with_softmax
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc2(x) return x def generate_eager_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data_inp = tensor(np.zeros(data_shape), dtype=np.float32) label_inp = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) def step(data, label): opt.zero_grad() data_inp.set_value(data) label_inp.set_value(label) pred = net(data_inp) loss = cross_entropy_with_softmax(pred, label_inp) opt.backward(loss) opt.step() return loss.numpy()[0] return step def generate_static_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data = tensor(np.zeros(data_shape), dtype=np.float32) label = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) # Save state to reset parameters later. state = copy.deepcopy(net.state_dict()) # Evaluate network in eager mode once. pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() grads = opt.backward(loss) f =
mge.graph.compile(loss, grads)
megengine.graph.compile
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 =
Linear(self.num_class, self.mid_layers, bias=True)
megengine.module.Linear
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ =
init.calculate_fan_in_and_fan_out(self.fc0.weight)
megengine.module.init.calculate_fan_in_and_fan_out
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in))
init.zeros_(self.fc0.bias)
megengine.module.init.zeros_
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 =
Linear(self.mid_layers, self.mid_layers, bias=True)
megengine.module.Linear
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ =
init.calculate_fan_in_and_fan_out(self.fc1.weight)
megengine.module.init.calculate_fan_in_and_fan_out
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in))
init.zeros_(self.fc1.bias)
megengine.module.init.zeros_
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 =
Linear(self.mid_layers, self.num_class, bias=True)
megengine.module.Linear
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ =
init.calculate_fan_in_and_fan_out(self.fc2.weight)
megengine.module.init.calculate_fan_in_and_fan_out
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in))
init.zeros_(self.fc2.bias)
megengine.module.init.zeros_
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x =
relu(x)
megengine.functional.relu
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x =
relu(x)
megengine.functional.relu
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc2(x) return x def generate_eager_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data_inp = tensor(np.zeros(data_shape), dtype=np.float32) label_inp = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) def step(data, label): opt.zero_grad() data_inp.set_value(data) label_inp.set_value(label) pred = net(data_inp) loss =
cross_entropy_with_softmax(pred, label_inp)
megengine.functional.cross_entropy_with_softmax
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc2(x) return x def generate_eager_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data_inp = tensor(np.zeros(data_shape), dtype=np.float32) label_inp = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) def step(data, label): opt.zero_grad() data_inp.set_value(data) label_inp.set_value(label) pred = net(data_inp) loss = cross_entropy_with_softmax(pred, label_inp) opt.backward(loss) opt.step() return loss.numpy()[0] return step def generate_static_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data = tensor(np.zeros(data_shape), dtype=np.float32) label = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) # Save state to reset parameters later. state = copy.deepcopy(net.state_dict()) # Evaluate network in eager mode once. pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() grads = opt.backward(loss) f = mge.graph.compile(loss, grads) def step(data, label): opt.zero_grad() out = f(data=data, label=label) opt.step() loss = out[0][0] return loss # Reset parameters. net.load_state_dict(state) return step def generate_trace_step( net: Module, opt_factory: Callable[[Module], Optimizer], enable: bool ): opt = opt_factory(net) @trace def train(data, label): pred = net(data) loss =
cross_entropy_with_softmax(pred, label)
megengine.functional.cross_entropy_with_softmax
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc2(x) return x def generate_eager_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data_inp = tensor(np.zeros(data_shape), dtype=np.float32) label_inp = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) def step(data, label): opt.zero_grad() data_inp.set_value(data) label_inp.set_value(label) pred = net(data_inp) loss = cross_entropy_with_softmax(pred, label_inp) opt.backward(loss) opt.step() return loss.numpy()[0] return step def generate_static_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data = tensor(np.zeros(data_shape), dtype=np.float32) label = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) # Save state to reset parameters later. state = copy.deepcopy(net.state_dict()) # Evaluate network in eager mode once. pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() grads = opt.backward(loss) f = mge.graph.compile(loss, grads) def step(data, label): opt.zero_grad() out = f(data=data, label=label) opt.step() loss = out[0][0] return loss # Reset parameters. net.load_state_dict(state) return step def generate_trace_step( net: Module, opt_factory: Callable[[Module], Optimizer], enable: bool ): opt = opt_factory(net) @trace def train(data, label): pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() opt.backward(loss) return loss train.enabled = enable def step(data, label): out = train(data, label) opt.step() loss = out[0][0] return loss return step def assert_network_equvilence(nets): net_state = [net.state_dict() for net in nets] for state in net_state[1:]: assert len(net_state[0]) == len(state) for k, v in net_state[0].items(): for state in net_state[1:]: assert k in state assertTensorClose(v, state[k]) @pytest.mark.slow def test_eager_equvilence(): eager_net = SimpleNet() trace_enable_net = copy.deepcopy(eager_net) trace_disable_net = copy.deepcopy(eager_net) opt_factory = lambda net: SGD( net.parameters(requires_grad=True), lr=0.01, momentum=0.01 ) estep = generate_eager_step(eager_net, opt_factory) te_step = generate_trace_step(trace_enable_net, opt_factory, True) td_step = generate_trace_step(trace_disable_net, opt_factory, False) assert_network_equvilence([eager_net, trace_enable_net, trace_disable_net]) # Use hard code number as limit, may increase if needed. for data, label in itertools.islice(minibatch_generator(), 200): eloss = estep(data, label) te_loss = te_step(data, label) td_loss = td_step(data, label)
assertTensorClose(eloss, te_loss)
megengine.test.assertTensorClose
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc2(x) return x def generate_eager_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data_inp = tensor(np.zeros(data_shape), dtype=np.float32) label_inp = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) def step(data, label): opt.zero_grad() data_inp.set_value(data) label_inp.set_value(label) pred = net(data_inp) loss = cross_entropy_with_softmax(pred, label_inp) opt.backward(loss) opt.step() return loss.numpy()[0] return step def generate_static_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data = tensor(np.zeros(data_shape), dtype=np.float32) label = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) # Save state to reset parameters later. state = copy.deepcopy(net.state_dict()) # Evaluate network in eager mode once. pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() grads = opt.backward(loss) f = mge.graph.compile(loss, grads) def step(data, label): opt.zero_grad() out = f(data=data, label=label) opt.step() loss = out[0][0] return loss # Reset parameters. net.load_state_dict(state) return step def generate_trace_step( net: Module, opt_factory: Callable[[Module], Optimizer], enable: bool ): opt = opt_factory(net) @trace def train(data, label): pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() opt.backward(loss) return loss train.enabled = enable def step(data, label): out = train(data, label) opt.step() loss = out[0][0] return loss return step def assert_network_equvilence(nets): net_state = [net.state_dict() for net in nets] for state in net_state[1:]: assert len(net_state[0]) == len(state) for k, v in net_state[0].items(): for state in net_state[1:]: assert k in state assertTensorClose(v, state[k]) @pytest.mark.slow def test_eager_equvilence(): eager_net = SimpleNet() trace_enable_net = copy.deepcopy(eager_net) trace_disable_net = copy.deepcopy(eager_net) opt_factory = lambda net: SGD( net.parameters(requires_grad=True), lr=0.01, momentum=0.01 ) estep = generate_eager_step(eager_net, opt_factory) te_step = generate_trace_step(trace_enable_net, opt_factory, True) td_step = generate_trace_step(trace_disable_net, opt_factory, False) assert_network_equvilence([eager_net, trace_enable_net, trace_disable_net]) # Use hard code number as limit, may increase if needed. for data, label in itertools.islice(minibatch_generator(), 200): eloss = estep(data, label) te_loss = te_step(data, label) td_loss = td_step(data, label) assertTensorClose(eloss, te_loss)
assertTensorClose(eloss, td_loss)
megengine.test.assertTensorClose
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import copy import itertools import os from typing import Callable import numpy as np import pytest import megengine as mge import megengine.module.init as init from megengine.core import tensor from megengine.functional import cross_entropy_with_softmax, relu from megengine.jit import trace from megengine.module import Linear, Module from megengine.optimizer import SGD, Optimizer from megengine.test import assertTensorClose batch_size = 64 data_shape = (batch_size, 2) label_shape = (batch_size,) def minibatch_generator(): while True: inp_data = np.zeros((batch_size, 2)) label = np.zeros(batch_size, dtype=np.int32) for i in range(batch_size): # [x0, x1], sampled from U[-1, 1] inp_data[i, :] = np.random.rand(2) * 2 - 1 label[i] = 0 if np.prod(inp_data[i]) < 0 else 1 yield inp_data.astype(np.float32), label.astype(np.int32) class SimpleNet(Module): def __init__(self): self.mid_layers = 14 self.num_class = 2 super().__init__() self.fc0 = Linear(self.num_class, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc0.weight) init.normal_(self.fc0.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc0.bias) self.fc1 = Linear(self.mid_layers, self.mid_layers, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc1.weight) init.normal_(self.fc1.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc1.bias) self.fc2 = Linear(self.mid_layers, self.num_class, bias=True) fan_in, _ = init.calculate_fan_in_and_fan_out(self.fc2.weight) init.normal_(self.fc2.weight, std=np.sqrt(float(1.0) / fan_in)) init.zeros_(self.fc2.bias) def forward(self, x): x = self.fc0(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc1(x) x = relu(x) # Should use tanh but it's not stable now. x = self.fc2(x) return x def generate_eager_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data_inp = tensor(np.zeros(data_shape), dtype=np.float32) label_inp = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) def step(data, label): opt.zero_grad() data_inp.set_value(data) label_inp.set_value(label) pred = net(data_inp) loss = cross_entropy_with_softmax(pred, label_inp) opt.backward(loss) opt.step() return loss.numpy()[0] return step def generate_static_step(net: Module, opt_factory: Callable[[Module], Optimizer]): data = tensor(np.zeros(data_shape), dtype=np.float32) label = tensor(np.zeros(label_shape), dtype=np.int32) opt = opt_factory(net) # Save state to reset parameters later. state = copy.deepcopy(net.state_dict()) # Evaluate network in eager mode once. pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() grads = opt.backward(loss) f = mge.graph.compile(loss, grads) def step(data, label): opt.zero_grad() out = f(data=data, label=label) opt.step() loss = out[0][0] return loss # Reset parameters. net.load_state_dict(state) return step def generate_trace_step( net: Module, opt_factory: Callable[[Module], Optimizer], enable: bool ): opt = opt_factory(net) @trace def train(data, label): pred = net(data) loss = cross_entropy_with_softmax(pred, label) opt.zero_grad() opt.backward(loss) return loss train.enabled = enable def step(data, label): out = train(data, label) opt.step() loss = out[0][0] return loss return step def assert_network_equvilence(nets): net_state = [net.state_dict() for net in nets] for state in net_state[1:]: assert len(net_state[0]) == len(state) for k, v in net_state[0].items(): for state in net_state[1:]: assert k in state
assertTensorClose(v, state[k])
megengine.test.assertTensorClose
#!/usr/bin/env python3 # -*- coding:utf-8 -*- # Copyright (c) Megvii, Inc. and its affiliates. import argparse import megengine as mge import numpy as np from megengine import jit from ..build import build_and_load def make_parser(): parser = argparse.ArgumentParser("YOLOX Demo Dump") parser.add_argument("-n", "--name", type=str, default="yolox-s", help="model name") parser.add_argument("-c", "--ckpt", default=None, type=str, help="ckpt for eval") parser.add_argument( "--dump_path", default="model.mge", help="path to save the dumped model" ) return parser def dump_static_graph(model, graph_name="model.mge"): model.eval() model.head.decode_in_inference = False data = mge.Tensor(np.random.random((1, 3, 640, 640))) @
jit.trace(capture_as_const=True)
megengine.jit.trace
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import json import os import subprocess import sys import time import numpy as np from resnet50 import Resnet50 import megengine as mge import megengine.distributed as dist import megengine.functional as F from megengine._internal.plugin import CompGraphProfiler from megengine.core import Graph, tensor from megengine.core.graph import get_default_graph from megengine.functional.debug_param import ( get_conv_execution_strategy, set_conv_execution_strategy, ) from megengine.jit import trace from megengine.module import BatchNorm2d, Conv2d, Linear, MaxPool2d, Module from megengine.optimizer import SGD sys.path.append(os.path.join(os.path.dirname(__file__), "..", "..", "..", "examples")) def init_profiler(comp_graph=
get_default_graph()
megengine.core.graph.get_default_graph
# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import argparse import json import os import subprocess import sys import time import numpy as np from resnet50 import Resnet50 import megengine as mge import megengine.distributed as dist import megengine.functional as F from megengine._internal.plugin import CompGraphProfiler from megengine.core import Graph, tensor from megengine.core.graph import get_default_graph from megengine.functional.debug_param import ( get_conv_execution_strategy, set_conv_execution_strategy, ) from megengine.jit import trace from megengine.module import BatchNorm2d, Conv2d, Linear, MaxPool2d, Module from megengine.optimizer import SGD sys.path.append(os.path.join(os.path.dirname(__file__), "..", "..", "..", "examples")) def init_profiler(comp_graph=get_default_graph()): profiler =
CompGraphProfiler(comp_graph)
megengine._internal.plugin.CompGraphProfiler