code
stringlengths
101
5.91M
def parse_args(): parser = argparse.ArgumentParser(description='Video Classification') parser.add_argument('--mode', type=str, default='test', help='train/test') parser.add_argument('--model', type=str, default='r3d', help='c3d/r3d/r21d') parser.add_argument('--dataset', type=str, default='ucf101', help='ucf101/hmdb51') parser.add_argument('--split', type=str, default='1', help='dataset split') parser.add_argument('--cl', type=int, default=16, help='clip length') parser.add_argument('--gpu', type=int, default=0, help='GPU id') parser.add_argument('--lr', type=float, default=1e-05, help='learning rate') parser.add_argument('--momentum', type=float, default=0.9, help='momentum') parser.add_argument('--wd', type=float, default=0.0005, help='weight decay') parser.add_argument('--log', type=str, help='log directory') parser.add_argument('--ckpt', type=str, default='log/UCF101_TCGL_split1_112_acc_finetuned_r3d_cl16_/best_model_144.pt', help='checkpoint path') parser.add_argument('--desp', type=str, help='additional description') parser.add_argument('--epochs', type=int, default=20, help='number of total epochs to run') parser.add_argument('--start-epoch', type=int, default=1, help='manual epoch number (useful on restarts)') parser.add_argument('--bs', type=int, default=16, help='mini-batch size') parser.add_argument('--workers', type=int, default=4, help='number of data loading workers') parser.add_argument('--pf', type=int, default=100, help='print frequency every batch') parser.add_argument('--seed', type=int, default=632, help='seed for initializing training.') args = parser.parse_args() return args
_model def ig_resnext101_32x48d(pretrained=True, **kwargs): model_args = dict(block=Bottleneck, layers=[3, 4, 23, 3], cardinality=32, base_width=48, **kwargs) return _create_resnet('ig_resnext101_32x48d', pretrained, **model_args)
def generate_data_ratio(rows): x_array = [] y_array = [] while (len(x_array) < rows): x = [np.random.uniform(0, 1), np.random.uniform(0.01, 1)] try: y = (x[0] / x[1]) x_array.append(x) y_array.append(y) except (ValueError, ZeroDivisionError): pass return (np.array(x_array, dtype=np.float32), np.array(y_array, dtype=np.float32))
def create_reverse_dependency_tree(): modules = [str(f.relative_to(PATH_TO_TRANFORMERS)) for f in (Path(PATH_TO_TRANFORMERS) / 'src/transformers').glob('**/*.py')] module_edges = [(d, m) for m in modules for d in get_module_dependencies(m)] tests = [str(f.relative_to(PATH_TO_TRANFORMERS)) for f in (Path(PATH_TO_TRANFORMERS) / 'tests').glob('**/*.py')] test_edges = [(d, t) for t in tests for d in get_test_dependencies(t)] return (module_edges + test_edges)
def _no_grad_trunc_normal_(tensor, mean, std, a, b): def norm_cdf(x): return ((1.0 + math.erf((x / math.sqrt(2.0)))) / 2.0) if ((mean < (a - (2 * std))) or (mean > (b + (2 * std)))): warnings.warn('mean is more than 2 std from [a, b] in nn.init.trunc_normal_. The distribution of values may be incorrect.', stacklevel=2) with torch.no_grad(): l = norm_cdf(((a - mean) / std)) u = norm_cdf(((b - mean) / std)) tensor.uniform_(((2 * l) - 1), ((2 * u) - 1)) tensor.erfinv_() tensor.mul_((std * math.sqrt(2.0))) tensor.add_(mean) tensor.clamp_(min=a, max=b) return tensor
class ROIAlign(nn.Module): def __init__(self, output_size, spatial_scale, sampling_ratio): super(ROIAlign, self).__init__() self.output_size = output_size self.spatial_scale = spatial_scale self.sampling_ratio = sampling_ratio _function def forward(self, input, rois): return roi_align(input, rois, self.output_size, self.spatial_scale, self.sampling_ratio) def __repr__(self): tmpstr = (self.__class__.__name__ + '(') tmpstr += ('output_size=' + str(self.output_size)) tmpstr += (', spatial_scale=' + str(self.spatial_scale)) tmpstr += (', sampling_ratio=' + str(self.sampling_ratio)) tmpstr += ')' return tmpstr
class TripletNet(nn.Module): def __init__(self, embedding_net): super(TripletNet, self).__init__() self.embedding_net = embedding_net def forward(self, x1, x2, x3): output1 = self.embedding_net(x1) output2 = self.embedding_net(x2) output3 = self.embedding_net(x3) return (output1, output2, output3) def get_embedding(self, x): return self.embedding_net(x)
def get_parser(): parser = argparse.ArgumentParser(description='Cumulative Reasoning') parser.add_argument('--temperature', type=float, default=0.1, help='temperature') parser.add_argument('--propnum', type=int, choices=range(0, 21), default=2, help='numbers of props') parser.add_argument('--reasoningnum', type=int, choices=range(0, 21), default=16, help='numbers of reasoning, when > 1, majority voting is used') parser.add_argument('--choices', type=int, choices=range(0, 21), default=5, help='numbers of premises to be chosen') parser.add_argument('--trycnt', type=int, choices=range(1, 1001), default=16, help='numbers of try times') parser.add_argument('--exploration_prob', type=float, default=1.0, help='exploration probability') parser.add_argument('--verified_reasoning', type=ast.literal_eval, default=False, help='self verified reasoning, may not work well for small models') parser.add_argument('--model', type=str, default='gpt-4', help='model to use') parser.add_argument('--dataset', type=str, default='data/folio/folio-wiki.jsonl', help='dataset to use') parser.add_argument('--verbose', type=ast.literal_eval, default=True, help='verbose mode') return parser
def read_labeled_image_list(image_list_file, isSkip=True): f = open(image_list_file, 'r') content = f.readlines() glen1 = len(content) pair1 = [] pair2 = [] labels = [] st = 2 it = 25 if (not isSkip): st = 0 it = 1 for i in range(st, glen1, it): line = content[i] (fn1, fn2, label) = line[:(- 1)].split(' ') fn1 = (FLAGS.training_img_dir + fn1) fn2 = (FLAGS.training_img_dir + fn2) pair1.append(fn1) pair2.append(fn2) labels.append(int(label)) return (pair1, pair2, np.asarray(labels))
def modularize(f): class Transform(nn.Module): def __init__(self, f): super(Transform, self).__init__() self.f = f def forward(self, x): return self.f(x) return Transform(f)
def bbox_ious(boxes1, boxes2, x1y1x2y2=True): if x1y1x2y2: mx = torch.min(boxes1[0], boxes2[0]) Mx = torch.max(boxes1[2], boxes2[2]) my = torch.min(boxes1[1], boxes2[1]) My = torch.max(boxes1[3], boxes2[3]) w1 = (boxes1[2] - boxes1[0]) h1 = (boxes1[3] - boxes1[1]) w2 = (boxes2[2] - boxes2[0]) h2 = (boxes2[3] - boxes2[1]) else: mx = torch.min((boxes1[0] - (boxes1[2] / 2.0)), (boxes2[0] - (boxes2[2] / 2.0))) Mx = torch.max((boxes1[0] + (boxes1[2] / 2.0)), (boxes2[0] + (boxes2[2] / 2.0))) my = torch.min((boxes1[1] - (boxes1[3] / 2.0)), (boxes2[1] - (boxes2[3] / 2.0))) My = torch.max((boxes1[1] + (boxes1[3] / 2.0)), (boxes2[1] + (boxes2[3] / 2.0))) w1 = boxes1[2] h1 = boxes1[3] w2 = boxes2[2] h2 = boxes2[3] uw = (Mx - mx) uh = (My - my) cw = ((w1 + w2) - uw) ch = ((h1 + h2) - uh) mask = (((cw <= 0) + (ch <= 0)) > 0) area1 = (w1 * h1) area2 = (w2 * h2) carea = (cw * ch) carea[mask] = 0 uarea = ((area1 + area2) - carea) return (carea / uarea)
def ffmpeg_read(bpayload: bytes, sampling_rate: int) -> np.array: ar = f'{sampling_rate}' ac = '1' format_for_conversion = 'f32le' ffmpeg_command = ['ffmpeg', '-i', 'pipe:0', '-ac', ac, '-ar', ar, '-f', format_for_conversion, '-hide_banner', '-loglevel', 'quiet', 'pipe:1'] try: with subprocess.Popen(ffmpeg_command, stdin=subprocess.PIPE, stdout=subprocess.PIPE) as ffmpeg_process: output_stream = ffmpeg_process.communicate(bpayload) except FileNotFoundError as error: raise ValueError('ffmpeg was not found but is required to load audio files from filename') from error out_bytes = output_stream[0] audio = np.frombuffer(out_bytes, np.float32) if (audio.shape[0] == 0): raise ValueError('Malformed soundfile') return audio
def test_fpn(): s = 64 in_channels = [8, 16, 32, 64] feat_sizes = [(s // (2 ** i)) for i in range(4)] out_channels = 8 with pytest.raises(AssertionError): FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, num_outs=2) with pytest.raises(AssertionError): FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, end_level=4, num_outs=2) with pytest.raises(AssertionError): FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, end_level=3, num_outs=1) with pytest.raises(AssertionError): FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, add_extra_convs='on_xxx', num_outs=5) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, add_extra_convs=True, num_outs=5) feats = [torch.rand(1, in_channels[i], feat_sizes[i], feat_sizes[i]) for i in range(len(in_channels))] outs = fpn_model(feats) assert (fpn_model.add_extra_convs == 'on_input') assert (len(outs) == fpn_model.num_outs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, add_extra_convs=False, num_outs=5) outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) assert (not fpn_model.add_extra_convs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, add_extra_convs=True, no_norm_on_lateral=False, norm_cfg=dict(type='BN', requires_grad=True), num_outs=5) outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) assert (fpn_model.add_extra_convs == 'on_input') for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) bn_exist = False for m in fpn_model.modules(): if isinstance(m, _BatchNorm): bn_exist = True assert bn_exist fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, add_extra_convs=True, upsample_cfg=dict(mode='bilinear', align_corners=True), num_outs=5) fpn_model(feats) outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) assert (fpn_model.add_extra_convs == 'on_input') for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, start_level=1, add_extra_convs=True, upsample_cfg=dict(scale_factor=2), num_outs=5) outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, add_extra_convs='on_input', start_level=1, num_outs=5) assert (fpn_model.add_extra_convs == 'on_input') outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, add_extra_convs='on_lateral', start_level=1, num_outs=5) assert (fpn_model.add_extra_convs == 'on_lateral') outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, add_extra_convs='on_output', start_level=1, num_outs=5) assert (fpn_model.add_extra_convs == 'on_output') outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, add_extra_convs=True, extra_convs_on_inputs=False, start_level=1, num_outs=5) assert (fpn_model.add_extra_convs == 'on_output') outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i))) fpn_model = FPN(in_channels=in_channels, out_channels=out_channels, add_extra_convs=True, extra_convs_on_inputs=True, start_level=1, num_outs=5) assert (fpn_model.add_extra_convs == 'on_input') outs = fpn_model(feats) assert (len(outs) == fpn_model.num_outs) for i in range(fpn_model.num_outs): (outs[i].shape[1] == out_channels) (outs[i].shape[2] == outs[i].shape[3] == (s // (2 ** i)))
class InceptionV4(nn.Module): def __init__(self, num_classes=1001): super(InceptionV4, self).__init__() self.input_space = None self.input_size = (299, 299, 3) self.mean = None self.std = None self.features = nn.Sequential(BasicConv2d(3, 32, kernel_size=3, stride=2), BasicConv2d(32, 32, kernel_size=3, stride=1), BasicConv2d(32, 64, kernel_size=3, stride=1, padding=1), Mixed_3a(), Mixed_4a(), Mixed_5a(), Inception_A(), Inception_A(), Inception_A(), Inception_A(), Reduction_A(), Inception_B(), Inception_B(), Inception_B(), Inception_B(), Inception_B(), Inception_B(), Inception_B(), Reduction_B(), Inception_C(), Inception_C(), Inception_C()) self.avg_pool = nn.AvgPool2d(8, count_include_pad=False) self.last_linear = nn.Linear(1536, num_classes) def logits(self, features): x = self.avg_pool(features) x = x.view(x.size(0), (- 1)) x = self.last_linear(x) return x def forward(self, input): x = self.features(input) x = self.logits(x) return x
def session_indexed(s): action_to_idx = {'start': 0, 'end': 1, 'add': 2, 'remove': 3, 'purchase': 4, 'detail': 5, 'view': 6} return (([action_to_idx['start']] + [action_to_idx[e] for e in s]) + [action_to_idx['end']])
def is_number(s): try: float(s) return True except ValueError: pass return False
def test_register_new_sensors_and_measures(): if (not PointNavDatasetV1.check_config_paths_exist(config=habitat.get_config().DATASET)): pytest.skip('Please download Habitat test data to data folder.') register_new_sensors_and_measures.main()
class VisualNavigationModel(nn.Module): def init_weights(self, module): if (type(module) in [nn.GRU, nn.LSTM, nn.RNN]): for (name, param) in module.named_parameters(): if ('weight_ih' in name): nn.init.xavier_uniform_(param.data) elif ('weight_hh' in name): nn.init.orthogonal_(param.data) elif ('bias' in name): param.data.fill_(0) elif (type(module) in [nn.Conv2d, nn.ConvTranspose2d, nn.Linear]): nn.init.zeros_(module.bias.data) (fan_in, fan_out) = nn.init._calculate_fan_in_and_fan_out(module.weight.data) d = (1.0 / math.sqrt(fan_in)) nn.init.uniform_(module.weight.data, (- d), d) def __init__(self, num_inputs, num_outputs): super().__init__() self.main_output_size = 512 self.shared_base = TimeDistributed(nn.Sequential(nn.Conv2d(num_inputs, 16, 8, stride=4), nn.ReLU(True))) self.conv_base = TimeDistributed(nn.Sequential(nn.Conv2d(32, 32, 4, stride=2), nn.ReLU(True), nn.Conv2d(32, 32, 1), nn.ReLU())) self.conv_merge = TimeDistributed(nn.Sequential(Flatten(), nn.Linear(((9 ** 2) * 32), self.main_output_size), nn.ReLU())) self.critic = TimeDistributed(nn.Linear(self.main_output_size, 1)) self.policy_logits = TimeDistributed(nn.Linear(self.main_output_size, num_outputs)) self.lstm_layers = 1 self.lstm_hidden_size = self.main_output_size self.rnn = MaskedRNN(nn.LSTM(((self.main_output_size + num_outputs) + 1), hidden_size=self.lstm_hidden_size, num_layers=self.lstm_layers, batch_first=True)) self._create_pixel_control_network(num_outputs) self._create_rp_network() self._create_deconv_networks() self.apply(self.init_weights) self.pc_cell_size = 2 self.deconv_cell_size = 2 def _create_deconv_networks(self): self.deconv_depth = TimeDistributed(nn.Sequential(Unflatten(32, 9, 9), nn.ConvTranspose2d(32, 16, kernel_size=4, stride=2), nn.ReLU(), nn.ConvTranspose2d(16, 1, kernel_size=4, stride=2))) self.deconv_rgb = TimeDistributed(nn.Sequential(Unflatten(32, 9, 9), nn.ConvTranspose2d(32, 16, kernel_size=4, stride=2), nn.ReLU())) self.deconv_rgb_goal = TimeDistributed(nn.Sequential(Unflatten(32, 9, 9), nn.ConvTranspose2d(32, 16, kernel_size=4, stride=2), nn.ReLU())) self.deconv_rgb_shared = TimeDistributed(nn.Sequential(nn.ConvTranspose2d(16, 3, kernel_size=4, stride=2))) self.deconv_depth.apply(self.init_weights) self.deconv_rgb_goal.apply(self.init_weights) self.deconv_rgb.apply(self.init_weights) self.deconv_rgb_shared.apply(self.init_weights) def initial_states(self, batch_size): return tuple([torch.zeros([batch_size, self.lstm_layers, self.lstm_hidden_size], dtype=torch.float32) for _ in range(2)]) def forward(self, inputs, masks, states): (features, states) = self._forward_base(inputs, masks, states) policy_logits = self.policy_logits(features) critic = self.critic(features) return [policy_logits, critic, states] def _forward_base(self, inputs, masks, states): (observations, last_reward_action) = inputs image = observations[0] goal = observations[1] (image, goal) = (self.shared_base(image), self.shared_base(goal)) features = torch.cat((image, goal), 2) features = self.conv_base(features) features = self.conv_merge(features) features = torch.cat((features, last_reward_action), dim=2) return self.rnn(features, masks, states) def _create_pixel_control_network(self, num_outputs): self.pc_base = TimeDistributed(nn.Sequential(nn.Linear(self.lstm_hidden_size, ((32 * 9) * 9)), nn.ReLU())) self.pc_action = TimeDistributed(nn.Sequential(nn.ConvTranspose2d(32, 32, kernel_size=4, stride=2), nn.ReLU(), nn.ConvTranspose2d(32, 1, kernel_size=4, stride=2), nn.ReLU())) self.pc_value = TimeDistributed(nn.Sequential(nn.ConvTranspose2d(32, 32, kernel_size=4, stride=2), nn.ReLU(), nn.ConvTranspose2d(32, num_outputs, kernel_size=4, stride=2), nn.ReLU())) def _create_rp_network(self): self.rp = nn.Sequential(Flatten(), nn.Linear((((9 ** 2) * 32) * 3), 3)) def reward_prediction(self, inputs): (observations, _) = inputs image = observations[0] goal = observations[1] (image, goal) = (self.shared_base(image), self.shared_base(goal)) features = torch.cat((image, goal), 2) features = self.conv_base(features) features = self.rp(features) return features def pixel_control(self, inputs, masks, states): (features, states) = self._forward_base(inputs, masks, states) features = self.pc_base(features) features = features.view(*(features.size()[:2] + (32, 9, 9))) action_features = self.pc_action(features) features = ((self.pc_value(features) + action_features) - action_features.mean(2, keepdim=True)) return (features, states) def value_prediction(self, inputs, masks, states): (features, states) = self._forward_base(inputs, masks, states) critic = self.critic(features) return (critic, states) def forward_deconv(self, inputs, masks, states): (observations, _) = inputs image = observations[0] goal = observations[1] (image, goal) = (self.shared_base(image), self.shared_base(goal)) features = torch.cat((image, goal), 2) features = self.conv_base(features) depth = self.deconv_depth(features) rgb = self.deconv_rgb_shared(self.deconv_rgb(features)) rgb_goal = self.deconv_rgb_shared(self.deconv_rgb_goal(features)) return ((rgb, rgb_goal, depth), states)
class TFBertForMaskedLM(metaclass=DummyObject): _backends = ['tf'] def __init__(self, *args, **kwargs): requires_backends(self, ['tf'])
def get_bs_per_stream(batch_size, stream_number): result = [] batch_per_instance = (batch_size // stream_number) if (batch_per_instance >= 1): used_num_streams = stream_number instance_need_extra_input = (batch_size % stream_number) else: batch_per_instance = 1 used_num_streams = batch_size instance_need_extra_input = 0 start_idx = 0 end_idx = 0 for j in range(used_num_streams): if (j < instance_need_extra_input): end_idx = (end_idx + (batch_per_instance + 1)) else: end_idx = (end_idx + batch_per_instance) result.append((end_idx - start_idx)) start_idx = end_idx return result
def init_weights(m): if (isinstance(m, nn.Conv2d) or isinstance(m, nn.Linear)): m.weight.data.normal_(0, 0.001) if (m.bias is not None): m.bias.data.zero_() elif isinstance(m, nn.ConvTranspose2d): m.weight.data.normal_(0, 0.001) if (m.bias is not None): m.bias.data.zero_() elif isinstance(m, nn.BatchNorm2d): m.weight.data.fill_(1) m.bias.data.zero_()
.parametrize('size', list_tensor_sizes()) .parametrize('dtype', list_non_bool_dtypes()) .parametrize('op', list_binary_ops()) def test_binary_ew_ops(benchmark, size, dtype, op): np_a = np.array(np.random.uniform(1, 127, size), dtype=to_numpy_dtype(dtype)) np_b = np.array(np.random.uniform(1, 127, size), dtype=to_numpy_dtype(dtype)) a = o3c.Tensor(np_a, dtype=dtype, device=o3c.Device('CPU:0')) b = o3c.Tensor(np_b, dtype=dtype, device=o3c.Device('CPU:0')) benchmark(op, a, b)
_vision class CLIPProcessorTest(unittest.TestCase): def setUp(self): self.tmpdirname = tempfile.mkdtemp() vocab = ['l', 'o', 'w', 'e', 'r', 's', 't', 'i', 'd', 'n', 'lo', 'l</w>', 'w</w>', 'r</w>', 't</w>', 'low</w>', 'er</w>', 'lowest</w>', 'newer</w>', 'wider', '<unk>', '<|startoftext|>', '<|endoftext|>'] vocab_tokens = dict(zip(vocab, range(len(vocab)))) merges = ['#version: 0.2', 'l o', 'lo w</w>', 'e r</w>', ''] self.special_tokens_map = {'unk_token': '<unk>'} self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES['vocab_file']) self.merges_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES['merges_file']) with open(self.vocab_file, 'w', encoding='utf-8') as fp: fp.write((json.dumps(vocab_tokens) + '\n')) with open(self.merges_file, 'w', encoding='utf-8') as fp: fp.write('\n'.join(merges)) image_processor_map = {'do_resize': True, 'size': 20, 'do_center_crop': True, 'crop_size': 18, 'do_normalize': True, 'image_mean': [0., 0.4578275, 0.], 'image_std': [0., 0., 0.]} self.image_processor_file = os.path.join(self.tmpdirname, IMAGE_PROCESSOR_NAME) with open(self.image_processor_file, 'w', encoding='utf-8') as fp: json.dump(image_processor_map, fp) def get_tokenizer(self, **kwargs): return CLIPTokenizer.from_pretrained(self.tmpdirname, **kwargs) def get_rust_tokenizer(self, **kwargs): return CLIPTokenizerFast.from_pretrained(self.tmpdirname, **kwargs) def get_image_processor(self, **kwargs): return CLIPImageProcessor.from_pretrained(self.tmpdirname, **kwargs) def tearDown(self): shutil.rmtree(self.tmpdirname) def prepare_image_inputs(self): image_inputs = [np.random.randint(255, size=(3, 30, 400), dtype=np.uint8)] image_inputs = [Image.fromarray(np.moveaxis(x, 0, (- 1))) for x in image_inputs] return image_inputs def test_save_load_pretrained_default(self): tokenizer_slow = self.get_tokenizer() tokenizer_fast = self.get_rust_tokenizer() image_processor = self.get_image_processor() processor_slow = CLIPProcessor(tokenizer=tokenizer_slow, image_processor=image_processor) processor_slow.save_pretrained(self.tmpdirname) processor_slow = CLIPProcessor.from_pretrained(self.tmpdirname, use_fast=False) processor_fast = CLIPProcessor(tokenizer=tokenizer_fast, image_processor=image_processor) processor_fast.save_pretrained(self.tmpdirname) processor_fast = CLIPProcessor.from_pretrained(self.tmpdirname) self.assertEqual(processor_slow.tokenizer.get_vocab(), tokenizer_slow.get_vocab()) self.assertEqual(processor_fast.tokenizer.get_vocab(), tokenizer_fast.get_vocab()) self.assertEqual(tokenizer_slow.get_vocab(), tokenizer_fast.get_vocab()) self.assertIsInstance(processor_slow.tokenizer, CLIPTokenizer) self.assertIsInstance(processor_fast.tokenizer, CLIPTokenizerFast) self.assertEqual(processor_slow.image_processor.to_json_string(), image_processor.to_json_string()) self.assertEqual(processor_fast.image_processor.to_json_string(), image_processor.to_json_string()) self.assertIsInstance(processor_slow.image_processor, CLIPImageProcessor) self.assertIsInstance(processor_fast.image_processor, CLIPImageProcessor) def test_save_load_pretrained_additional_features(self): processor = CLIPProcessor(tokenizer=self.get_tokenizer(), image_processor=self.get_image_processor()) processor.save_pretrained(self.tmpdirname) tokenizer_add_kwargs = self.get_tokenizer(bos_token='(BOS)', eos_token='(EOS)') image_processor_add_kwargs = self.get_image_processor(do_normalize=False, padding_value=1.0) processor = CLIPProcessor.from_pretrained(self.tmpdirname, bos_token='(BOS)', eos_token='(EOS)', do_normalize=False, padding_value=1.0) self.assertEqual(processor.tokenizer.get_vocab(), tokenizer_add_kwargs.get_vocab()) self.assertIsInstance(processor.tokenizer, CLIPTokenizerFast) self.assertEqual(processor.image_processor.to_json_string(), image_processor_add_kwargs.to_json_string()) self.assertIsInstance(processor.image_processor, CLIPImageProcessor) def test_image_processor(self): image_processor = self.get_image_processor() tokenizer = self.get_tokenizer() processor = CLIPProcessor(tokenizer=tokenizer, image_processor=image_processor) image_input = self.prepare_image_inputs() input_image_proc = image_processor(image_input, return_tensors='np') input_processor = processor(images=image_input, return_tensors='np') for key in input_image_proc.keys(): self.assertAlmostEqual(input_image_proc[key].sum(), input_processor[key].sum(), delta=0.01) def test_tokenizer(self): image_processor = self.get_image_processor() tokenizer = self.get_tokenizer() processor = CLIPProcessor(tokenizer=tokenizer, image_processor=image_processor) input_str = 'lower newer' encoded_processor = processor(text=input_str) encoded_tok = tokenizer(input_str) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key], encoded_processor[key]) def test_processor(self): image_processor = self.get_image_processor() tokenizer = self.get_tokenizer() processor = CLIPProcessor(tokenizer=tokenizer, image_processor=image_processor) input_str = 'lower newer' image_input = self.prepare_image_inputs() inputs = processor(text=input_str, images=image_input) self.assertListEqual(list(inputs.keys()), ['input_ids', 'attention_mask', 'pixel_values']) with pytest.raises(ValueError): processor() def test_tokenizer_decode(self): image_processor = self.get_image_processor() tokenizer = self.get_tokenizer() processor = CLIPProcessor(tokenizer=tokenizer, image_processor=image_processor) predicted_ids = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] decoded_processor = processor.batch_decode(predicted_ids) decoded_tok = tokenizer.batch_decode(predicted_ids) self.assertListEqual(decoded_tok, decoded_processor) def test_model_input_names(self): image_processor = self.get_image_processor() tokenizer = self.get_tokenizer() processor = CLIPProcessor(tokenizer=tokenizer, image_processor=image_processor) input_str = 'lower newer' image_input = self.prepare_image_inputs() inputs = processor(text=input_str, images=image_input) self.assertListEqual(list(inputs.keys()), processor.model_input_names)
class Abstractor(object): def __init__(self, abs_dir, max_len=30, cuda=True): abs_meta = json.load(open(os.path.join(abs_dir, 'meta.json'))) assert (abs_meta['net'] == 'base_abstractor') abs_args = abs_meta['net_args'] abs_ckpt = load_best_ckpt(abs_dir) word2id = pkl.load(open(os.path.join(abs_dir, 'vocab.pkl'), 'rb')) abstractor = CopySumm(**abs_args) abstractor.load_state_dict(abs_ckpt) self._device = torch.device(('cuda' if cuda else 'cpu')) self._net = abstractor.to(self._device) self._word2id = word2id self._id2word = {i: w for (w, i) in word2id.items()} self._max_len = max_len def _prepro(self, raw_article_sents): ext_word2id = dict(self._word2id) ext_id2word = dict(self._id2word) for raw_words in raw_article_sents: for w in raw_words: if (not (w in ext_word2id)): ext_word2id[w] = len(ext_word2id) ext_id2word[len(ext_id2word)] = w articles = conver2id(UNK, self._word2id, raw_article_sents) art_lens = [len(art) for art in articles] article = pad_batch_tensorize(articles, PAD, cuda=False).to(self._device) extend_arts = conver2id(UNK, ext_word2id, raw_article_sents) extend_art = pad_batch_tensorize(extend_arts, PAD, cuda=False).to(self._device) extend_vsize = len(ext_word2id) dec_args = (article, art_lens, extend_art, extend_vsize, START, END, UNK, self._max_len) return (dec_args, ext_id2word) def __call__(self, raw_article_sents): self._net.eval() (dec_args, id2word) = self._prepro(raw_article_sents) (decs, attns) = self._net.batch_decode(*dec_args) def argmax(arr, keys): return arr[max(range(len(arr)), key=(lambda i: keys[i].item()))] dec_sents = [] for (i, raw_words) in enumerate(raw_article_sents): dec = [] for (id_, attn) in zip(decs, attns): if (id_[i] == END): break elif (id_[i] == UNK): dec.append(argmax(raw_words, attn[i])) else: dec.append(id2word[id_[i].item()]) dec_sents.append(dec) return dec_sents
class BNMomentumScheduler(object): def __init__(self, model, bn_lambda, last_epoch=(- 1), setter=set_bn_momentum_default): if (not isinstance(model, nn.Module)): raise RuntimeError("Class '{}' is not a PyTorch nn Module".format(type(model).__name__)) self.model = model self.setter = setter self.lmbd = bn_lambda self.step((last_epoch + 1)) self.last_epoch = last_epoch def step(self, epoch=None): if (epoch is None): epoch = (self.last_epoch + 1) self.last_epoch = epoch self.model.apply(self.setter(self.lmbd(epoch))) def get_momentum(self, epoch=None): if (epoch is None): epoch = (self.last_epoch + 1) return self.lmbd(epoch)
def lossfun(x, alpha, scale, approximate=False, epsilon=1e-06): assert torch.is_tensor(x) assert torch.is_tensor(scale) assert torch.is_tensor(alpha) assert (alpha.dtype == x.dtype) assert (scale.dtype == x.dtype) assert (scale > 0).all() if approximate: assert (epsilon > np.finfo(np.float32).eps) b = (torch.abs((alpha - 2)) + epsilon) d = torch.where((alpha >= 0), (alpha + epsilon), (alpha - epsilon)) loss = ((b / d) * (torch.pow(((((x / scale) ** 2) / b) + 1.0), (0.5 * d)) - 1.0)) else: squared_scaled_x = ((x / scale) ** 2) loss_two = (0.5 * squared_scaled_x) loss_zero = util.log1p_safe((0.5 * squared_scaled_x)) loss_neginf = (- torch.expm1(((- 0.5) * squared_scaled_x))) loss_posinf = util.expm1_safe((0.5 * squared_scaled_x)) machine_epsilon = torch.tensor(np.finfo(np.float32).eps).to(x) beta_safe = torch.max(machine_epsilon, torch.abs((alpha - 2.0))) alpha_safe = (torch.where((alpha >= 0), torch.ones_like(alpha), (- torch.ones_like(alpha))) * torch.max(machine_epsilon, torch.abs(alpha))) loss_otherwise = ((beta_safe / alpha_safe) * (torch.pow(((squared_scaled_x / beta_safe) + 1.0), (0.5 * alpha)) - 1.0)) loss = torch.where((alpha == (- float('inf'))), loss_neginf, torch.where((alpha == 0), loss_zero, torch.where((alpha == 2), loss_two, torch.where((alpha == float('inf')), loss_posinf, loss_otherwise)))) return loss
def one_of_k_encoding_unk(x, allowable_set): if (x not in allowable_set): return None return list(map((lambda s: int((x == s))), allowable_set))
(others=sampled_from([{'box': TFBoxTensor(tf.Variable([[[1, 1], [3, 5]], [[2, 0], [6, 2]]], dtype=tf.float32)), 'weights': None, 'mask': None, 'keepdim': True, 'dim': 0, 'expected': TFBoxTensor(tf.Variable([[(3.0 / 2.0), (1.0 / 2.0)], [(9.0 / 2.0), (7.0 / 2.0)]]))}, {'box': TFBoxTensor(tf.Variable([[[1, 1], [3, 5]], [[2, 0], [6, 2]]], dtype=tf.float32)), 'weights': tf.reshape(tf.Variable([0.5, 0.5]), (2, 1)), 'mask': None, 'keepdim': True, 'dim': 0, 'expected': TFBoxTensor(tf.Variable([[(3.0 / 2.0), (1.0 / 2.0)], [(9.0 / 2.0), (7.0 / 2.0)]]))}, {'box': TFBoxTensor(tf.Variable([[[1, 1], [3, 5]], [[2, 0], [6, 2]]], dtype=tf.float32)), 'weights': tf.reshape(tf.Variable([0.1, 0.9]), (2, 1)), 'mask': None, 'keepdim': True, 'dim': 0, 'expected': TFBoxTensor(tf.Variable([[((0.1 * 1.0) + (0.9 * 2.0)), ((0.1 * 1.0) + (0.9 * 0))], [((0.1 * 3) + (0.9 * 6.0)), ((0.1 * 5) + (0.9 * 2))]]))}, {'box': TFBoxTensor(tf.Variable([[[[1, 1], [3, 5]], [[2, 0], [6, 2]]]], dtype=tf.float32)), 'weights': tf.reshape(tf.Variable([0.1, 0.9]), (1, 2, 1)), 'mask': None, 'keepdim': True, 'dim': 1, 'expected': TFBoxTensor(tf.Variable([[[((0.1 * 1.0) + (0.9 * 2.0)), ((0.1 * 1.0) + (0.9 * 0))], [((0.1 * 3) + (0.9 * 6.0)), ((0.1 * 5) + (0.9 * 2))]]], dtype=tf.float32))}, {'box': TFBoxTensor(tf.Variable([[[1, 1, 1], [3, 5, 6]], [[2, 0, 1], [6, 2, 3]]], dtype=tf.float32)), 'weights': tf.Variable([0.5, 0.5], dtype=tf.float32), 'mask': None, 'keepdim': True, 'dim': 0, 'expected': InternalError}])) def test_bob(others) -> None: box = others['box'] bob = TFBagOfBoxesBoxPooler(dim=others['dim'], keepdim=others['keepdim']) expected = others['expected'] if isinstance(expected, TFBoxTensor): result = bob(box, mask=others['mask'], weights=others['weights']) assert np.allclose(result.z, others['expected'].z) assert np.allclose(result.Z, others['expected'].Z) else: with pytest.raises(expected): result = bob(box, mask=others['mask'], weights=others['weights'])
def merge_new_config(config, new_config): for (key, val) in new_config.items(): if (not isinstance(val, dict)): if (key == '_base_'): with open(new_config['_base_'], 'r') as f: try: val = yaml.load(f, Loader=yaml.FullLoader) except: val = yaml.load(f) config[key] = EasyDict() merge_new_config(config[key], val) else: config[key] = val continue if (key not in config): config[key] = EasyDict() merge_new_config(config[key], val) return config
class TransNorm2d(_TransNorm): def _check_input(self, x): if (x.dim() != 4): raise ValueError('Expected the input to be 4-D, but got {}-D'.format(x.dim()))
def dc_state_dict(dc_vars, *name_list): return {(name + '_state_dict'): dc_vars[name].state_dict() for name in name_list if hasattr(dc_vars[name], 'state_dict')}
def eval(args, val_loader, model, criterion): model.eval() losses = AverageMeter() top1 = AverageMeter() top5 = AverageMeter() device = args.device if args.cuda: torch.cuda.empty_cache() for (data, y) in val_loader: data = data.to(device, non_blocking=True) y = y.to(device, non_blocking=True) outputs = model(data) loss = criterion(outputs, y) (prec1, prec5) = accuracy(outputs.data, y.data, topk=(1, 5)) losses.update(loss.item(), data.size(0)) top1.update(prec1.item(), data.size(0)) top5.update(prec5.item(), data.size(0)) logging('Avg loss: {:.4f}, top1: {:.2f}%, top5: {:.2f}%'.format(losses.avg, top1.avg, top5.avg), args.log) return (losses.avg, top1.avg, top5.avg)
def convert_coco_poly_to_mask(segmentations, height, width): masks = [] for polygons in segmentations: rles = coco_mask.frPyObjects(polygons, height, width) mask = coco_mask.decode(rles) if (len(mask.shape) < 3): mask = mask[(..., None)] mask = torch.as_tensor(mask, dtype=torch.uint8) mask = mask.any(dim=2) masks.append(mask) if masks: masks = torch.stack(masks, dim=0) else: masks = torch.zeros((0, height, width), dtype=torch.uint8) return masks
def weights_to_cpu(state_dict): state_dict_cpu = OrderedDict() for (key, val) in state_dict.items(): state_dict_cpu[key] = val.cpu() return state_dict_cpu
class CmsPfSinglePi(tfds.core.GeneratorBasedBuilder): VERSION = tfds.core.Version('1.6.0') RELEASE_NOTES = {'1.0.0': 'Initial release.', '1.1.0': 'Add muon type, fix electron GSF association', '1.2.0': '12_1_0_pre3 generation, add corrected energy, cluster flags, 20k events', '1.4.0': 'Add genjet information', '1.5.0': 'Without padding', '1.5.1': 'Remove outlier caps', '1.6.0': 'Regenerate with ARRAY_RECORD'} MANUAL_DOWNLOAD_INSTRUCTIONS = '\n rsync -r --progress lxplus.cern.ch:/eos/user/j/jpata/mlpf/tensorflow_datasets/cms/cms_pf_single_pi ~/tensorflow_datasets/\n ' def __init__(self, *args, **kwargs): kwargs['file_format'] = tfds.core.FileFormat.ARRAY_RECORD super(CmsPfSinglePi, self).__init__(*args, **kwargs) def _info(self) -> tfds.core.DatasetInfo: return tfds.core.DatasetInfo(builder=self, description=_DESCRIPTION, features=tfds.features.FeaturesDict({'X': tfds.features.Tensor(shape=(None, len(X_FEATURES)), dtype=tf.float32), 'ygen': tfds.features.Tensor(shape=(None, len(Y_FEATURES)), dtype=tf.float32), 'ycand': tfds.features.Tensor(shape=(None, len(Y_FEATURES)), dtype=tf.float32)}), supervised_keys=('X', 'ycand'), homepage='', citation=_CITATION, metadata=tfds.core.MetadataDict(x_features=X_FEATURES, y_features=Y_FEATURES)) def _split_generators(self, dl_manager: tfds.download.DownloadManager): path = dl_manager.manual_dir sample_dir = 'SinglePiMinusFlatPt0p7To1000_cfi' return cms_utils.split_sample(((path / sample_dir) / 'raw')) def _generate_examples(self, files): return cms_utils.generate_examples(files)
def connect_nodes(n1, n2): if (n2['node'].name not in n1['outputs']): n1['outputs'].append(n2['node'].name) n2['inputs'].append(n1['node'].name) else: print('{} -> {} already connected'.format(n1['node'].name, n2['node'].name))
def load_actor(checkpointpath): import json with open(checkpointpath.replace('model.tar', 'meta.json'), 'r') as file: args = json.load(file)['args'] if ('breath' not in args): args['breath'] = 2 env = create_gymenv(AttributeDict(args)) return (env, create_model(AttributeDict(args), env), AttributeDict(args))
class Res16UNetSN14(Res16UNet14): NORM_TYPE = NormType.SPARSE_SWITCH_NORM BLOCK = BasicBlockSN
class MyMetric_keras(MyMetric): def __init__(self, *args): super(MyMetric_keras, self).__init__(*args)
class BasePose(nn.Module): __metaclass__ = ABCMeta def forward_train(self, img, img_metas, **kwargs): def forward_test(self, img, img_metas, **kwargs): def forward(self, img, img_metas, return_loss=True, **kwargs): def _parse_losses(losses): log_vars = OrderedDict() for (loss_name, loss_value) in losses.items(): if isinstance(loss_value, torch.Tensor): log_vars[loss_name] = loss_value.mean() elif isinstance(loss_value, float): log_vars[loss_name] = loss_value elif isinstance(loss_value, list): log_vars[loss_name] = sum((_loss.mean() for _loss in loss_value)) else: raise TypeError(f'{loss_name} is not a tensor or list of tensors or float') loss = sum((_value for (_key, _value) in log_vars.items() if ('loss' in _key))) log_vars['loss'] = loss for (loss_name, loss_value) in log_vars.items(): if (not isinstance(loss_value, float)): if (dist.is_available() and dist.is_initialized()): loss_value = loss_value.data.clone() dist.all_reduce(loss_value.div_(dist.get_world_size())) log_vars[loss_name] = loss_value.item() else: log_vars[loss_name] = loss_value return (loss, log_vars) def train_step(self, data_batch, optimizer, **kwargs): losses = self.forward(**data_batch) (loss, log_vars) = self._parse_losses(losses) outputs = dict(loss=loss, log_vars=log_vars, num_samples=len(next(iter(data_batch.values())))) return outputs def val_step(self, data_batch, optimizer, **kwargs): results = self.forward(return_loss=False, **data_batch) outputs = dict(results=results) return outputs def show_result(self, **kwargs): raise NotImplementedError
def symlink_force(target, link_name): try: os.symlink(target, link_name) except OSError as e: if (e.errno == errno.EEXIST): os.remove(link_name) os.symlink(target, link_name) else: raise e
class Trainer(): _pipeline: Union[(BaseNRHintPipeline, DDP)] def __init__(self, config: SystemConfig, shm_info: NRDataSHMInfo): self.config = config self.rank = local_rank() self.world_size = 1 if torch.cuda.is_available(): self.device = torch.device('cuda') torch.set_default_tensor_type('torch.cuda.FloatTensor') torch.backends.cudnn.benchmark = True gpu_count = torch.cuda.device_count() if (gpu_count > 1): torch.distributed.init_process_group(backend='nccl', init_method='env://') self.world_size = torch.distributed.get_world_size() torch_rank = torch.distributed.get_rank() assert (torch_rank == self.rank), f'torch_rank {torch_rank} != rank {self.rank}, initialization might have failed...' self.device = torch.device('cuda', self.rank) torch.cuda.set_device(self.device) else: self.device = torch.device('cpu') self.log_dir = ((pathlib.Path(self.config.base_dir) / self.config.exp_name) / self.config.scene_name) self.log_dir.mkdir(parents=True, exist_ok=True) if self.is_main_process: with open((self.log_dir / 'config.yaml'), 'w') as f: f.write(tyro.to_yaml(self.config)) wandb.init(project='NR2023', name=f'{config.exp_name}_{config.scene_name}', group=config.exp_name, notes='NR Hint Relighting', config=asdict(config), resume='allow', id=str(self.log_dir).replace('/', '_')) wandb.run.log_code() seed_everything((config.seed + self.rank)) self._pipeline = BaseNRHintPipeline(config, shm_info).to(self.device) if (self.world_size > 1): self._pipeline = DDP(self._pipeline, device_ids=[self.rank], find_unused_parameters=False, broadcast_buffers=True) if self.is_main_process: summary(self.pipeline) self.optimizer = torch.optim.Adam(self.pipeline.get_param_groups()) alpha = config.model.lr_alpha warm_up_end = config.model.warm_up_end end_iter = config.model.end_iter def lr_lambda(iter_step): if (iter_step < warm_up_end): learning_factor = (iter_step / warm_up_end) else: progress = ((iter_step - warm_up_end) / (end_iter - warm_up_end)) learning_factor = ((((np.cos((np.pi * progress)) + 1.0) * 0.5) * (1 - alpha)) + alpha) return learning_factor self.lr_scheduler = torch.optim.lr_scheduler.LambdaLR(self.optimizer, lr_lambda) self.data_manager = NRSHMDataManager(shm_info=shm_info, batch_size=(config.model.batch_size // self.world_size), strategy=PixelSamplingStrategy.ALL_IMAGES, image_idx_rng_seed=config.seed, pixel_idx_rng_seed=config.seed, local_rank=self.rank) self.global_step = 0 self.load_ckpt() def pipeline(self): if isinstance(self._pipeline, DDP): return cast(BaseNRHintPipeline, self._pipeline.module) return self._pipeline def is_main_process(self): return (self.rank == 0) def use_ddp(self): return (self.world_size > 1) def wait_all(self): if self.use_ddp: torch.distributed.barrier() def model_states(self) -> Dict: return {'world_size': self.world_size, 'global_step': self.global_step, 'pipeline': self.pipeline.state_dict(), 'optimizer': self.optimizer.state_dict(), 'scheduler': self.lr_scheduler.state_dict()} def rng_states(self): return {'python.random': random.getstate(), 'np.random': np.random.get_state(), 'torch.random': torch.random.get_rng_state(), 'torch.cuda.random': torch.cuda.random.get_rng_state(self.device), 'ray_generator.image': self.data_manager.sampler.image_rng.__getstate__(), 'ray_generator.pixel': self.data_manager.sampler.pixel_rng.__getstate__()} def save_ckpt(self): if self.is_main_process: ckpt_path = ((self.log_dir / 'ckpt') / f'step_{self.global_step:07d}.ckpt') torch.save(self.model_states, ckpt_path) rng_state_path = ((self.log_dir / 'rng_state') / f'step_{self.global_step:07d}_device_{self.rank}.pickle') pickle.dump(self.rng_states, open(rng_state_path, 'wb')) def load_ckpt(self): ckpt_path = (self.log_dir / 'ckpt') ckpt_path.mkdir(parents=True, exist_ok=True) rng_state_path = (self.log_dir / 'rng_state') rng_state_path.mkdir(parents=True, exist_ok=True) if (self.config.ckpt_path is not None): ckpts = [self.config.ckpt_path] print(f'Loading given ckpt from {self.config.ckpt_path}') else: ckpts = [os.path.join(ckpt_path, f) for f in sorted(os.listdir(ckpt_path)) if ('ckpt' in f)] print(f'Found {len(ckpts)} ckpts in {ckpt_path}') if (len(ckpts) > 0): try: ckpt_path = ckpts[(- 1)] print(f'Resume from ckpt: {ckpt_path}') last_world_size = self._load_ckpt_file(ckpt_path) except EOFError: ckpt_path = ckpts[(- 2)] print(f'Retrying resume from ckpt: {ckpt_path}') last_world_size = self._load_ckpt_file(ckpt_path) if (last_world_size == self.world_size): try: rng_state_path = (rng_state_path / f'step_{self.global_step:07d}_device_{self.rank}.pickle') self._load_rng_states(rng_state_path) except Exception as e: print(e) print('rng state resume failed, the results might not be fully reproducible') def _load_ckpt_file(self, ckpt_file): ckpt = torch.load(ckpt_file, map_location=self.device) self.global_step = ckpt['global_step'] self.pipeline.load_state_dict(ckpt['pipeline']) self.optimizer.load_state_dict(ckpt['optimizer']) self.lr_scheduler.load_state_dict(ckpt['scheduler']) return ckpt['world_size'] def _load_rng_states(self, rng_state_path): rng_states = pickle.load(open(rng_state_path, 'rb')) random.setstate(rng_states['python.random']) np.random.set_state(rng_states['np.random']) torch.random.set_rng_state(rng_states['torch.random']) torch.cuda.set_rng_state(rng_states['torch.cuda.random']) self.data_manager.sampler.image_rng.__setstate__(rng_states['ray_generator.image']) self.data_manager.sampler.pixel_rng.__setstate__(rng_states['ray_generator.pixel']) def run(self): if (not self.config.evaluation_only): start_step = self.global_step for _ in tqdm(range(start_step, self.config.model.end_iter), desc=f'Training: ', initial=start_step, total=self.config.model.end_iter, dynamic_ncols=True, disable=(self.rank != 0)): loss_dict = self.train_iter() if (((self.global_step % self.config.intervals.log_metrics) == 0) and self.is_main_process): wandb.log(loss_dict, step=self.global_step) self.global_step += 1 if ((self.global_step % self.config.intervals.save_ckpt) == 0): self.save_ckpt() if ((self.global_step % self.config.intervals.render_test_views) == 0): self.render_test_views() if ((self.global_step % self.config.intervals.dump_mesh) == 0): self.dump_mesh() if ((self.global_step % self.config.intervals.render_video) == 0): self.render_video() self.dump_mesh(resolution=1024) self.render_test_views(is_final=True) def train_iter(self): pixel_bundle = self.data_manager.next_train_batch() pixel_bundle = pixel_bundle.to(self.device) rendering_res = self._pipeline.forward(pixel_bundle, global_step=self.global_step) loss_dict = self.pipeline.get_train_loss_dict(rendering_res, pixel_bundle) loss = loss_dict['loss'] self.optimizer.zero_grad() loss.backward() self.optimizer.step() self.lr_scheduler.step() return loss_dict def render_test_views(self, is_final=False): total_test_view = self.data_manager.test_view_num() skip_num = (self.config.data.testset_skip if (not is_final) else 1) start_idx = (self.rank * skip_num) metrics_dicts = [] for idx in tqdm(range(start_idx, total_test_view, (skip_num * self.world_size)), desc=f'Rendering test views on process {self.rank}: '): metrics_dict = self.render_single_view(idx) metrics_dicts.append(metrics_dict) self.wait_all() if self.use_ddp: output_list = [None for _ in range(self.world_size)] torch.distributed.gather_object(obj=metrics_dicts, object_gather_list=(output_list if self.is_main_process else None), dst=0) else: output_list = [metrics_dicts] if self.is_main_process: gather_output = {} image_cnt = 0 for item in itertools.chain(*output_list): image_cnt += 1 for (k, v) in item.items(): gather_output.setdefault(k, 0.0) gather_output[k] += v final_output = {} for (k, v) in gather_output.items(): final_output[(f'val/{k}' if is_final else f'val/{k}')] = (v / image_cnt) wandb.log(final_output, step=self.global_step) self.wait_all() _grad() def render_single_view(self, view_index, is_training_view: bool=False): img_pixel_bundle = self.data_manager.get_test_view(view_index) (img_dict, metrics_dict, tensor_dict) = self.pipeline.get_eval_dicts(img_pixel_bundle, self.device) self.save_dumps(view_index, img_dict, tensor_dict) if ((view_index == 0) and self.is_main_process): self.log_images(img_dict) return metrics_dict def save_dumps(self, view_idx, image_dict, tensor_dict): dump_dir = ((self.log_dir / 'test_views') / f'step_{self.global_step:07d}') dump_dir.mkdir(parents=True, exist_ok=True) for (k, v) in image_dict.items(): if ('normal' in k): v = ((v * 0.5) + 0.5) if (v.shape[(- 1)] == 1): v = v[(..., 0)] imageio.v3.imwrite((dump_dir / f'{k}_{view_idx:03d}.png'), (v * 255).clip(0, 255).astype(np.uint8)) for (k, v) in tensor_dict.items(): np.save((dump_dir / f'{k}_{view_idx:03d}.npy'), v) def log_images(self, image_dict): for (k, v) in image_dict.items(): if ('normal' in k): v = ((v * 0.5) + 0.5) wandb.log({k: wandb.Image((v * 255).clip(0, 255).astype(np.uint8))}, step=self.global_step) def dump_mesh(self, resolution: int=256): if self.is_main_process: mesh_dir = (self.log_dir / 'mesh') mesh_dir.mkdir(parents=True, exist_ok=True) mesh_file_path = (mesh_dir / f'step_{self.global_step:07d}_res_{resolution}.obj') bound_min = torch.tensor([(- 1.01), (- 1.01), (- 1.01)], dtype=torch.float32) bound_max = torch.tensor([1.01, 1.01, 1.01], dtype=torch.float32) (vertices, triangles) = self.pipeline.renderer.extract_geometry(bound_min, bound_max, resolution=resolution, threshold=0.0) mesh = trimesh.Trimesh(vertices, triangles) mesh.export(mesh_file_path) self.wait_all() def render_video(self): video_dir = (self.log_dir / 'video') video_dir.mkdir(parents=True, exist_ok=True) video_frame_dir = (video_dir / f'step_{self.global_step:07d}') video_frame_dir.mkdir(parents=True, exist_ok=True) video_pixel_bundle = self.data_manager.get_video_pixel_bundle(self.config.data.video_frame_num, is_z_up=self.config.data.is_z_up) if self.is_main_process: video_frame_buffer = torch.empty(*video_pixel_bundle.shape, 3, dtype=torch.uint8, device='cpu') assert ((len(video_pixel_bundle) % self.world_size) == 0), 'video frame number should be divisible by world size' frames_per_process = (len(video_pixel_bundle) // self.world_size) for i in tqdm(range(frames_per_process), desc=f'Rendering video frames on process {self.rank}: '): idx = (i + (frames_per_process * self.rank)) (img_dict, _, _) = self.pipeline.get_eval_dicts(video_pixel_bundle[idx], self.device) uint8_rgb = (img_dict['rgb'] * 255).clip(0, 255).astype(np.uint8) imageio.v3.imwrite((video_frame_dir / f'{idx:03d}.png'), uint8_rgb) if self.is_main_process: video_frame_buffer[idx] = torch.tensor(uint8_rgb, dtype=torch.uint8, device='cpu') single_idx_buffer = torch.empty((1,), dtype=torch.long, device=self.device) single_frame_buffer = torch.empty_like(video_frame_buffer[idx], dtype=torch.uint8, device=self.device) for sub_process_rank in range(1, self.world_size): torch.distributed.recv(tensor=single_idx_buffer, src=sub_process_rank, tag=0) torch.distributed.recv(tensor=single_frame_buffer, src=sub_process_rank, tag=1) video_frame_buffer[single_idx_buffer.item()] = single_frame_buffer.cpu() else: torch.distributed.send(torch.tensor([idx], device=self.device), dst=0, tag=0) torch.distributed.send(torch.tensor(uint8_rgb, dtype=torch.uint8, device=self.device), dst=0, tag=1) self.wait_all() if self.is_main_process: video_rgb = video_frame_buffer.numpy().astype(np.uint8) imageio.v3.imwrite(os.path.join(video_dir, f'step_{self.global_step:06d}_rot_view.mp4'), video_rgb[:self.config.data.video_frame_num], fps=30, quality=9) imageio.v3.imwrite(os.path.join(video_dir, f'step_{self.global_step:06d}_rot_light.mp4'), video_rgb[self.config.data.video_frame_num:], fps=30, quality=9) self.wait_all() def release_shm(self): self.data_manager.release_shm() del self.data_manager
class ConvVAE(nn.Module): def __init__(self, latent_size): super(ConvVAE, self).__init__() self.latent_size = latent_size self.encoder = nn.Sequential(nn.Conv2d(3, 64, kernel_size=4, stride=2, padding=1), nn.ReLU(), nn.Conv2d(64, 128, kernel_size=4, stride=2, padding=1), nn.ReLU(), Flatten(), nn.Linear(131072, 1024), nn.ReLU()) self.fc1 = nn.Linear(1024, self.latent_size) self.fc2 = nn.Linear(1024, self.latent_size) self.decoder = nn.Sequential(nn.Linear(self.latent_size, 1024), nn.ReLU(), nn.Linear(1024, 131072), nn.ReLU(), Unflatten(128, 32, 32), nn.ReLU(), nn.ConvTranspose2d(128, 64, kernel_size=4, stride=2, padding=1), nn.ReLU(), nn.ConvTranspose2d(64, 3, kernel_size=4, stride=2, padding=1), nn.Sigmoid()) def encode(self, x): h = self.encoder(x) (mu, logvar) = (self.fc1(h), self.fc2(h)) return (mu, logvar) def decode(self, z): z = self.decoder(z) return z def reparameterize(self, mu, logvar): if self.training: std = torch.exp((0.5 * logvar)) eps = torch.randn_like(std) return eps.mul(std).add_(mu) else: return mu def reparameterize_eval(self, mu, logvar): std = torch.exp((0.5 * logvar)) eps = torch.randn_like(std) return eps.mul(std).add_(mu) def forward(self, x): if (x.shape[1] == 1): x = x.repeat(1, 3, 1, 1) (mu, logvar) = self.encode(x) z = self.reparameterize(mu, logvar) return (self.decode(z), mu, logvar) def loss_function(self, x: Tensor, y: Tensor, mu: Tensor, logvar: Tensor) -> dict: if (x.shape[1] == 1): x = x.repeat(1, 3, 1, 1) recon_loss = F.binary_cross_entropy(y, x, reduction='mean') kl_loss = torch.mean(((- 0.5) * (((1 + logvar) - (mu ** 2)) - logvar.exp()))) loss = (recon_loss + (1 * kl_loss)) return {'loss': loss, 'recon_loss': recon_loss, 'kl_loss': kl_loss}
def get_sorted_wordlist(path): freqs = defaultdict((lambda : 0)) with codecs.open(path, 'r', encoding='utf-8') as fin: for line in fin: words = line.strip().split() for word in words: freqs[word] += 1 sorted_words = sorted(freqs, key=freqs.get, reverse=True) wordlist = [word for word in sorted_words] return wordlist
def make_layers(cfg, batch_norm=False): layers = [] in_channels = 3 for (i, v) in enumerate(cfg): if (v == 'M'): layers += [nn.MaxPool2d(kernel_size=2, stride=2)] else: padding = (v[1] if isinstance(v, tuple) else 1) out_channels = (v[0] if isinstance(v, tuple) else v) conv2d = nn.Conv2d(in_channels, out_channels, kernel_size=3, padding=padding) if batch_norm: layers += [conv2d, nn.BatchNorm2d(out_channels, affine=False), nn.ReLU()] else: layers += [conv2d, nn.ReLU()] in_channels = out_channels return nn.Sequential(*layers)
class Conv2dIndepBeta(_DeepIndepBeta): def __init__(self, backbone: nn.Module, hidden_channels: int=1, out_channels: int=1): super().__init__(backbone=backbone, alpha_head=nn.Conv2d(hidden_channels, out_channels=out_channels, kernel_size=1), beta_head=nn.Conv2d(hidden_channels, out_channels=out_channels, kernel_size=1))
def prune(model, amount=0.3): import torch.nn.utils.prune as prune print('Pruning model... ', end='') for (name, m) in model.named_modules(): if isinstance(m, nn.Conv2d): prune.l1_unstructured(m, name='weight', amount=amount) prune.remove(m, 'weight') print((' %.3g global sparsity' % sparsity(model)))
class MelFilterBank(): def __init__(self, specSize, numCoefficients, sampleRate): numBands = int(numCoefficients) minMel = 0 maxMel = self.freqToMel((sampleRate / 2.0)) melStep = ((maxMel - minMel) / (numBands + 1)) melFilterEdges = (np.arange(0, (numBands + 2)) * melStep) centerIndices = list(map((lambda x: self.freqToBin(math.floor(self.melToFreq(x)), sampleRate, specSize)), melFilterEdges)) filterMatrix = np.zeros((numBands, specSize)) for i in range(numBands): (start, center, end) = centerIndices[i:(i + 3)] k1 = np.float((center - start)) k2 = np.float((end - center)) up = ((np.array(range(start, center)) - start) / k1) down = ((end - np.array(range(center, end))) / k2) filterMatrix[i][start:center] = up filterMatrix[i][center:end] = down self.melMatrix = filterMatrix.transpose() self.melMatrix = self.makeNormal((self.melMatrix / self.normSum(self.melMatrix))) self.melInvMatrix = self.melMatrix.transpose() self.melInvMatrix = self.makeNormal((self.melInvMatrix / self.normSum(self.melInvMatrix))) def normSum(self, x): retSum = np.sum(x, axis=0) retSum[np.where((retSum == 0))] = 1.0 return retSum def fuzz(self, x): return (x + 1e-07) def freqToBin(self, freq, sampleRate, specSize): return int(math.floor(((freq / (sampleRate / 2.0)) * specSize))) def freqToMel(self, freq): return (2595.0 * math.log10((1.0 + (freq / 700.0)))) def melToFreq(self, mel): return (700.0 * (math.pow(10.0, (mel / 2595.0)) - 1.0)) def toMelScale(self, spectrogram): return np.dot(spectrogram, self.melMatrix) def fromMelScale(self, melSpectrogram): return np.dot(melSpectrogram, self.melInvMatrix) def makeNormal(self, x): nanIdx = np.isnan(x) x[nanIdx] = 0 infIdx = np.isinf(x) x[infIdx] = 0 return x def toMels(self, spectrogram): return self.toMelScale(spectrogram) def fromMels(self, melSpectrogram): return self.fromMelScale(melSpectrogram) def toLogMels(self, spectrogram): return self.makeNormal(np.log(self.fuzz(self.toMelScale(spectrogram)))) def fromLogMels(self, melSpectrogram): return self.makeNormal(self.fromMelScale(np.exp(melSpectrogram)))
class Softmax(Module): def __init__(self, dim): super().__init__() self.dim = dim def forward(self, input): return input.softmax(self.dim) def from_onnx(parameters=None, attributes=None): if (attributes is None): attributes = {} return Softmax(attributes['axis'])
def tf2th(conv_weights): if (conv_weights.ndim == 4): conv_weights = conv_weights.transpose([3, 2, 0, 1]) return torch.from_numpy(conv_weights)
def add_args(parser): parser.add_argument('--num_steps', type=int, default=(10 ** 6), help='number of steps in training') parser.add_argument('--transitions_per_step', type=int, default=1, help='env transitions per training step. Defaults to 1, but will need to be set higher for repaly ratios < 1') parser.add_argument('--max_episode_steps', type=int, default=100000, help='maximum steps per episode') parser.add_argument('--batch_size', type=int, default=512, help='training batch size') parser.add_argument('--actor_lr', type=float, default=0.0001, help='actor learning rate') parser.add_argument('--critic_lr', type=float, default=0.0001, help='critic learning rate') parser.add_argument('--gamma', type=float, default=0.99, help='gamma, the discount factor') parser.add_argument('--buffer_size', type=int, default=1000000, help='replay buffer size') parser.add_argument('--eval_interval', type=int, default=5000, help='how often to test the agent without exploration (in episodes)') parser.add_argument('--eval_episodes', type=int, default=10, help='how many episodes to run for when testing') parser.add_argument('--warmup_steps', type=int, default=1000, help='warmup length, in steps') parser.add_argument('--render', action='store_true', help='flag to enable env rendering during training') parser.add_argument('--actor_clip', type=float, default=None, help='gradient clipping for actor updates') parser.add_argument('--critic_clip', type=float, default=None, help='gradient clipping for critic updates') parser.add_argument('--name', type=str, default='tsr_caql_run', help='dir name for saves') parser.add_argument('--actor_l2', type=float, default=0.0, help='L2 regularization coeff for actor network') parser.add_argument('--critic_l2', type=float, default=0.0, help='L2 regularization coeff for critic network') parser.add_argument('--save_interval', type=int, default=100000, help='How many steps to go between saving the agent params to disk') parser.add_argument('--verbosity', type=int, default=1, help='verbosity > 0 displays a progress bar during training') parser.add_argument('--max_critic_updates_per_step', type=int, default=10, help='Max critic updates to make per step. The GRAC paper calls this K') parser.add_argument('--prioritized_replay', action='store_true', help='flag that enables use of prioritized experience replay') parser.add_argument('--skip_save_to_disk', action='store_true', help='flag to skip saving agent params to disk during training') parser.add_argument('--skip_log_to_disk', action='store_true', help='flag to skip saving agent performance logs to disk during training') parser.add_argument('--log_std_low', type=float, default=(- 10), help='Lower bound for log std of action distribution.') parser.add_argument('--log_std_high', type=float, default=2, help='Upper bound for log std of action distribution.') parser.add_argument('--critic_target_improvement_init', type=float, default=0.7, help='Stop critic updates when loss drops by this factor. The GRAC paper calls this alpha') parser.add_argument('--critic_target_improvement_final', type=float, default=0.9, help='Stop critic updates when loss drops by this factor. The GRAC paper calls this alpha') parser.add_argument('--debug_logs', action='store_true')
def read_reco2vol(volumes_file): volumes = {} with open(volumes_file) as volume_reader: for line in volume_reader.readlines(): if (len(line.strip()) == 0): continue parts = line.strip().split() if (len(parts) != 2): raise RuntimeError('Unable to parse the line {0} in file {1}.'.format(line.strip(), volumes_file)) volumes[parts[0]] = float(parts[1]) return volumes
def gelu(x): x = tf.convert_to_tensor(x) cdf = (0.5 * (1.0 + tf.math.erf((x / tf.math.sqrt(2.0))))) return (x * cdf)
_registry(operator_type='LayerNorm') class LayerNorm(Operator): def __init__(self): super().__init__() def set_attr(self, framework, node): if (framework == 'torch'): if (node.inputsSize() > 4): self._attr['epsilon'] = node.inputsAt(4).toIValue() self._attr['normalized_shape'] = list2str(parseTorchListConstruct(node.inputsAt(1)))
def get_train_type(train_type, checkpoint): exist_status = (checkpoint and os.path.exists(checkpoint)) if (train_type == 'NORMAL'): return train_type if ((train_type == 'FPD') and exist_status): return 'FPD' if ((train_type == 'FPD') and (not exist_status)): exit('ERROR: teacher checkpoint is not existed.') else: exit('ERROR: please change train type {} to NORMAL or FPD.'.format(train_type))
class Indexer(): def __init__(self, symbols=['<pad>', '<unk>', '<s>', '</s>']): self.vocab = defaultdict(int) self.PAD = symbols[0] self.UNK = symbols[1] self.BOS = symbols[2] self.EOS = symbols[3] self.d = {self.PAD: 0, self.UNK: 1, self.BOS: 2, self.EOS: 3} self.idx2word = {} def add_w(self, ws): for w in ws: if (w not in self.d): self.d[w] = len(self.d) def convert(self, w): return (self.d[w] if (w in self.d) else self.d[self.UNK]) def convert_sequence(self, ls): return [self.convert(l) for l in ls] def write(self, outfile): out = open(outfile, 'w') items = [(v, k) for (k, v) in self.d.items()] items.sort() for (v, k) in items: out.write((' '.join([k, str(v)]) + '\n')) out.close() def prune_vocab(self, k, cnt=False): vocab_list = [(word, count) for (word, count) in self.vocab.items()] if cnt: self.pruned_vocab = {pair[0]: pair[1] for pair in vocab_list if (pair[1] > k)} else: vocab_list.sort(key=(lambda x: x[1]), reverse=True) k = min(k, len(vocab_list)) self.pruned_vocab = {pair[0]: pair[1] for pair in vocab_list[:k]} for word in self.pruned_vocab: if (word not in self.d): self.d[word] = len(self.d) for (word, idx) in self.d.items(): self.idx2word[idx] = word def load_vocab(self, vocab_file): self.d = {} for line in open(vocab_file, 'r'): (v, k) = line.strip().split() self.d[v] = int(k) for (word, idx) in self.d.items(): self.idx2word[idx] = word
_module() class Sharpness(object): def __init__(self, magnitude, prob=0.5, random_negative_prob=0.5): assert isinstance(magnitude, (int, float)), f'The magnitude type must be int or float, but got {type(magnitude)} instead.' assert (0 <= prob <= 1.0), f'The prob should be in range [0,1], got {prob} instead.' assert (0 <= random_negative_prob <= 1.0), f'The random_negative_prob should be in range [0,1], got {random_negative_prob} instead.' self.magnitude = magnitude self.prob = prob self.random_negative_prob = random_negative_prob def __call__(self, results): if (np.random.rand() > self.prob): return results magnitude = random_negative(self.magnitude, self.random_negative_prob) for key in results.get('img_fields', ['strong']): img = results[key] img_sharpened = mmcv.adjust_sharpness(img, factor=(1 + magnitude)) results[key] = img_sharpened.astype(img.dtype) return results def __repr__(self): repr_str = self.__class__.__name__ repr_str += f'(magnitude={self.magnitude}, ' repr_str += f'prob={self.prob}, ' repr_str += f'random_negative_prob={self.random_negative_prob})' return repr_str
def parse_args(): parser = argparse.ArgumentParser(description='Create periodicity threshold figure') parser.add_argument('--names', required=True, nargs='+', help='Corresponding labels for each evaluation') parser.add_argument('--evaluations', type=Path, required=True, nargs='+', help='The evaluations to plot') parser.add_argument('--output_file', type=Path, required=True, help='The output jpg file') return parser.parse_known_args()[0]
def get_messages_tokens(messages): cnt = 0 for message in messages: cnt += count_tokens(message['content']) return cnt
def D_logistic(G, D, opt, training_set, minibatch_size, reals, labels): _ = (opt, training_set) latents = tf.random_normal(([minibatch_size] + G.input_shapes[0][1:])) fake_images_out = G.get_output_for(latents, labels, is_training=True) real_scores_out = D.get_output_for(reals, labels, is_training=True) fake_scores_out = D.get_output_for(fake_images_out, labels, is_training=True) real_scores_out = autosummary('Loss/scores/real', real_scores_out) fake_scores_out = autosummary('Loss/scores/fake', fake_scores_out) loss = tf.nn.softplus(fake_scores_out) loss += tf.nn.softplus((- real_scores_out)) return (loss, None)
def run(api_key, api_url, index): es = Elasticsearch(hosts=[ELASTICSEARCH_HOST]) arxivdigest_connector = ArxivdigestConnector(api_key, api_url) if (not es.indices.exists(index=index)): logger.info('Creating index') init_index(es, index) logger.info('Indexing articles from arXivDigest API.') run_indexing(es, index, arxivdigest_connector) recommend(es, arxivdigest_connector, index) logger.info('\nFinished recommending articles.')
class nnUNetTrainerV2_lReLU_biasInSegOutput(nnUNetTrainerV2): def initialize_network(self): if self.threeD: conv_op = nn.Conv3d dropout_op = nn.Dropout3d norm_op = nn.InstanceNorm3d else: conv_op = nn.Conv2d dropout_op = nn.Dropout2d norm_op = nn.InstanceNorm2d norm_op_kwargs = {'eps': 1e-05, 'affine': True} dropout_op_kwargs = {'p': 0, 'inplace': True} net_nonlin = nn.LeakyReLU net_nonlin_kwargs = {'negative_slope': 0.01, 'inplace': True} self.network = Generic_UNet(self.num_input_channels, self.base_num_features, self.num_classes, len(self.net_num_pool_op_kernel_sizes), self.conv_per_stage, 2, conv_op, norm_op, norm_op_kwargs, dropout_op, dropout_op_kwargs, net_nonlin, net_nonlin_kwargs, True, False, (lambda x: x), InitWeights_He(0), self.net_num_pool_op_kernel_sizes, self.net_conv_kernel_sizes, False, True, True, seg_output_use_bias=True) if torch.cuda.is_available(): self.network.cuda() self.network.inference_apply_nonlin = softmax_helper
def iso_recal(exp_props, obs_props): exp_props = exp_props.flatten() obs_props = obs_props.flatten() min_obs = torch.min(obs_props) max_obs = torch.max(obs_props) iso_model = IsotonicRegression(increasing=True, out_of_bounds='clip') try: assert (torch.min(obs_props) == 0.0) assert (torch.max(obs_props) == 1.0) except: print('Obs props not ideal: from {} to {}'.format(min_obs, max_obs)) exp_0_idx = get_q_idx(exp_props, 0.0) exp_1_idx = get_q_idx(exp_props, 1.0) within_01 = obs_props[exp_0_idx:(exp_1_idx + 1)] (beg_idx, end_idx) = (None, None) min_obs_below = torch.min(obs_props[:exp_0_idx]) min_obs_within = torch.min(within_01) if (min_obs_below < min_obs_within): i = (exp_0_idx - 1) while (obs_props[i] > min_obs_below): i -= 1 beg_idx = i elif (torch.sum((within_01 == min_obs_within).float()) > 1): i = (exp_1_idx - 1) while (obs_props[i] > min_obs_within): i -= 1 beg_idx = i elif (torch.sum((within_01 == min_obs_within).float()) == 1): beg_idx = (torch.argmin(within_01) + exp_0_idx) else: import pudb pudb.set_trace() max_obs_above = torch.max(obs_props[(exp_1_idx + 1):]) max_obs_within = torch.max(within_01) if (max_obs_above > max_obs_within): i = (exp_1_idx + 1) while (obs_props[i] < max_obs_above): i += 1 end_idx = (i + 1) elif (torch.sum((within_01 == max_obs_within).float()) > 1): i = beg_idx while (obs_props[i] < max_obs_within): i += 1 end_idx = (i + 1) elif (torch.sum((within_01 == max_obs_within).float()) == 1): end_idx = ((exp_0_idx + torch.argmax(within_01)) + 1) else: import pudb pudb.set_trace() assert (end_idx > beg_idx) filtered_obs_props = obs_props[beg_idx:end_idx] filtered_exp_props = exp_props[beg_idx:end_idx] try: iso_model = iso_model.fit(filtered_obs_props, filtered_exp_props) except: import pudb pudb.set_trace() return iso_model
class AdapterResnetBlock(nn.Module): def __init__(self, channels: int): super().__init__() self.block1 = nn.Conv2d(channels, channels, kernel_size=3, padding=1) self.act = nn.ReLU() self.block2 = nn.Conv2d(channels, channels, kernel_size=1) def forward(self, x: torch.Tensor) -> torch.Tensor: h = self.act(self.block1(x)) h = self.block2(h) return (h + x)
class CrossEntropyLoss(nn.CrossEntropyLoss): def __init__(self, weight=None, ignore_index=(- 100), reduction='mean', smooth_eps=None, smooth_dist=None): super(CrossEntropyLoss, self).__init__(weight=weight, ignore_index=ignore_index, reduction=reduction) self.smooth_eps = smooth_eps self.smooth_dist = smooth_dist def forward(self, input, target, smooth_dist=None): if (smooth_dist is None): smooth_dist = self.smooth_dist return cross_entropy(input, target, self.weight, self.ignore_index, self.reduction, self.smooth_eps, smooth_dist)
def getModelFiles(): return [dict(name=os.path.basename(p), mtime=int(os.path.getmtime(p))) for p in glob.glob(os.path.join(models_dir, '*.tflite'))]
def test_poisson_time_generator(): gen = PoissonTimeGenerator(lambda_time=2, random_generator=np.random.RandomState(seed=1)) for _ in range(10): print(gen.next())
def test_capsule(capture): pytest.gc_collect() with capture: a = m.return_capsule_with_destructor() del a pytest.gc_collect() assert (capture.unordered == '\n creating capsule\n destructing capsule\n ') with capture: a = m.return_capsule_with_destructor_2() del a pytest.gc_collect() assert (capture.unordered == '\n creating capsule\n destructing capsule: 1234\n ') with capture: a = m.return_capsule_with_name_and_destructor() del a pytest.gc_collect() assert (capture.unordered == "\n created capsule (1234, 'pointer type description')\n destructing capsule (1234, 'pointer type description')\n ")
_model('s2spect2_conformer') class S2SpecT2ConformerModel(S2SpecTConformerModel): def add_args(parser): S2SpecTConformerModel.add_args(parser) parser.add_argument('--translation-decoder-layers', type=int, default=4, metavar='N', help='num decoder layers in the first-pass translation module') parser.add_argument('--synthesizer', default='transformer', choices=['transformer'], help='') parser.add_argument('--synthesizer-encoder-layers', type=int, default=0, metavar='N', help='num encoder layers in the second-pass synthesizer module') def build_multitask_decoder(cls, args, tgt_dict, in_dim, is_mt_decoder, decoder_layers, decoder_embed_dim, decoder_attention_heads): decoder_args = args.decoder_args decoder_args.encoder_embed_dim = in_dim if (args.decoder_type == 'transformer'): if is_mt_decoder: multitask_text_transformer_decoder_arch(decoder_args, decoder_layers, decoder_embed_dim, decoder_attention_heads) else: base_multitask_text_transformer_decoder_arch(decoder_args) task_decoder = TransformerDecoder(decoder_args, tgt_dict, embed_tokens=TransformerModelBase.build_embedding(decoder_args, tgt_dict, decoder_args.decoder_embed_dim)) elif (args.decoder_type == 'ctc'): task_decoder = CTCDecoder(dictionary=tgt_dict, in_dim=in_dim) else: raise NotImplementedError("currently only support multitask decoder_type 'transformer', 'ctc'") return task_decoder def build_decoder(cls, args): _args = copy.deepcopy(args) _args.encoder_embed_dim = args.decoder_embed_dim if (args.synthesizer == 'transformer'): return TTSTransformerDecoder(_args, None, padding_idx=1) else: raise NotImplementedError(args.synthesizer) def build_model(cls, args, task): encoder = cls.build_encoder(args) decoder = cls.build_decoder(args) base_model = cls(encoder, decoder) base_model.mt_task_name = None base_model.multitask_decoders = {} has_first_pass_decoder = False for (task_name, task_obj) in task.multitask_tasks.items(): if task_obj.is_first_pass_decoder: has_first_pass_decoder = True base_model.mt_task_name = task_name in_dim = (args.encoder_embed_dim if (task_obj.args.input_from == 'encoder') else args.decoder_embed_dim) task_decoder = cls.build_multitask_decoder(task_obj.args, task_obj.target_dictionary, in_dim, task_obj.is_first_pass_decoder, getattr(args, 'translation_decoder_layers', 4), getattr(args, 'decoder_embed_dim', 256), getattr(args, 'decoder_attention_heads', 4)) setattr(base_model, f'{task_name}_decoder', task_decoder) decoder_model_cls = (FairseqEncoderModel if (task_obj.args.decoder_type == 'ctc') else FairseqLanguageModel) base_model.multitask_decoders[task_name] = decoder_model_cls(getattr(base_model, f'{task_name}_decoder')) assert has_first_pass_decoder, 'set at least one intermediate non-CTC decoder' if (getattr(args, 'synthesizer_encoder_layers', 0) > 0): base_model.synthesizer_encoder = cls.build_text_encoder(args) else: base_model.synthesizer_encoder = None return base_model def build_text_encoder(cls, args): _args = copy.deepcopy(args) _args.encoder_layers = args.synthesizer_encoder_layers _args.encoder_embed_dim = args.decoder_embed_dim _args.encoder_ffn_embed_dim = args.decoder_ffn_embed_dim _args.encoder_attention_heads = args.decoder_attention_heads _args.encoder_normalize_before = True return TransformerEncoderNoEmb(_args) def forward(self, src_tokens, src_lengths, prev_output_tokens, prev_output_tokens_mt, tgt_speaker=None, incremental_state=None, target_lengths=None, speaker=None, return_all_hiddens=False): encoder_out = self.encoder(src_tokens, src_lengths=src_lengths, tgt_speaker=tgt_speaker, return_all_hiddens=return_all_hiddens) mt_decoder = getattr(self, f'{self.mt_task_name}_decoder') mt_decoder_out = mt_decoder(prev_output_tokens_mt, encoder_out=encoder_out) x = mt_decoder_out[1]['inner_states'][(- 1)] if (mt_decoder.layer_norm is not None): x = mt_decoder.layer_norm(x) mt_decoder_padding_mask = None if prev_output_tokens_mt.eq(mt_decoder.padding_idx).any(): mt_decoder_padding_mask = prev_output_tokens_mt.eq(mt_decoder.padding_idx) if (self.synthesizer_encoder is not None): tts_encoder_out = self.synthesizer_encoder(x, mt_decoder_padding_mask, return_all_hiddens=return_all_hiddens) else: tts_encoder_out = {'encoder_out': [x], 'encoder_padding_mask': [mt_decoder_padding_mask]} decoder_out = self.decoder(prev_output_tokens, encoder_out=tts_encoder_out, incremental_state=incremental_state, target_lengths=target_lengths, speaker=speaker) if return_all_hiddens: decoder_out[(- 1)]['encoder_states'] = encoder_out['encoder_states'] decoder_out[(- 1)]['encoder_padding_mask'] = encoder_out['encoder_padding_mask'] decoder_out[(- 1)]['mt_decoder_out'] = mt_decoder_out return decoder_out
def resnet18(pretrained=False, filter_size=1, pool_only=True, **kwargs): model = ResNet(BasicBlock, [2, 2, 2, 2], filter_size=filter_size, pool_only=pool_only, **kwargs) if pretrained: model.load_state_dict(model_zoo.load_url(model_urls['resnet18'])) return model
def test_digits_corr_two_stage_init(): model = SaturatedCoverageSelection(100, 'corr', optimizer='two-stage', initial_subset=digits_corr_ranking[:5]) model.fit(X_digits) assert_array_equal(model.ranking[:(- 5)], digits_corr_ranking[5:]) assert_array_almost_equal(model.gains[:(- 5)], digits_corr_gains[5:], 4) assert_array_almost_equal(model.subset, X_digits[model.ranking])
.parametrize('as_frame', [True, False]) def test_load_movielens100k(as_frame): (df_data, df_users, df_items) = load_movielens100k(as_frame=as_frame) if as_frame: assert ((df_data.shape, df_users.shape, df_items.shape, type(df_data), type(df_users), type(df_items)) == ((100000, 4), (943, 5), (1682, 24), pd.DataFrame, pd.DataFrame, pd.DataFrame)) else: assert ((df_data.shape, df_users.shape, df_items.shape, type(df_data), type(df_users), type(df_items)) == ((100000, 4), (943, 5), (1682, 24), np.ndarray, np.ndarray, np.ndarray))
def objective(trial): param = {'min_samples_leaf': trial.suggest_int('min_samples_leaf', 2, 20), 'min_samples_split': trial.suggest_int('min_samples_split', 2, 20), 'max_depth': trial.suggest_int('max_depth', 2, 32), 'n_estimators': trial.suggest_int('n_estimators', 100, 1000, step=100)} clf = RandomForestRegressor(**param).fit(X_train, y_train) pred = clf.predict(X_test) r2 = r2_score(y_test, pred) return r2
class MyDataloader(data.Dataset): modality_names = ['rgb', 'rgbd', 'd'] color_jitter = transforms.ColorJitter(0.4, 0.4, 0.4) def __init__(self, root, type, sparsifier=None, modality='rgb', loader=h5_loader): (classes, class_to_idx) = find_classes(root) imgs = make_dataset(root, class_to_idx) assert (len(imgs) > 0), (('Found 0 images in subfolders of: ' + root) + '\n') print('Found {} images in {} folder.'.format(len(imgs), type)) self.root = root self.imgs = imgs self.classes = classes self.class_to_idx = class_to_idx if (type == 'train'): self.transform = self.train_transform elif (type == 'val'): self.transform = self.val_transform else: raise RuntimeError((('Invalid dataset type: ' + type) + '\nSupported dataset types are: train, val')) self.loader = loader self.sparsifier = sparsifier assert (modality in self.modality_names), (((('Invalid modality type: ' + modality) + '\n') + 'Supported dataset types are: ') + ''.join(self.modality_names)) self.modality = modality def train_transform(self, rgb, depth): raise RuntimeError('train_transform() is not implemented. ') def val_transform(rgb, depth): raise RuntimeError('val_transform() is not implemented.') def create_sparse_depth(self, rgb, depth): if (self.sparsifier is None): return depth else: mask_keep = self.sparsifier.dense_to_sparse(rgb, depth) sparse_depth = np.zeros(depth.shape) sparse_depth[mask_keep] = depth[mask_keep] return sparse_depth def create_rgbd(self, rgb, depth): sparse_depth = self.create_sparse_depth(rgb, depth) rgbd = np.append(rgb, np.expand_dims(sparse_depth, axis=2), axis=2) return rgbd def __getraw__(self, index): (path, target) = self.imgs[index] (rgb, depth) = self.loader(path) return (rgb, depth) def __getitem__(self, index): (rgb, depth) = self.__getraw__(index) if (self.transform is not None): (rgb_np, depth_np) = self.transform(rgb, depth) else: raise RuntimeError('transform not defined') if (self.modality == 'rgb'): input_np = rgb_np elif (self.modality == 'rgbd'): input_np = self.create_rgbd(rgb_np, depth_np) elif (self.modality == 'd'): input_np = self.create_sparse_depth(rgb_np, depth_np) input_tensor = to_tensor(input_np) while (input_tensor.dim() < 3): input_tensor = input_tensor.unsqueeze(0) depth_tensor = to_tensor(depth_np) depth_tensor = depth_tensor.unsqueeze(0) return (input_tensor, depth_tensor) def __len__(self): return len(self.imgs)
class Config(): def __init__(self): self.det_head = 'pip' self.net_stride = 32 self.batch_size = 16 self.init_lr = 0.0001 self.num_epochs = 60 self.decay_steps = [30, 50] self.input_size = 256 self.backbone = 'resnet18' self.pretrained = True self.criterion_cls = 'l2' self.criterion_reg = 'l1' self.cls_loss_weight = 10 self.reg_loss_weight = 1 self.num_lms = 68 self.save_interval = self.num_epochs self.num_nb = 10 self.use_gpu = True self.gpu_id = 2
def static_baseline(num, probas): logits = [np.log((p + 1e-19)) for p in probas] return torch.tensor(([logits] * num)).float()
def build_optim(model, optim_opt): optim = nmt.Optim(optim_opt.optim_method, optim_opt.learning_rate, optim_opt.max_grad_norm, optim_opt.learning_rate_decay, optim_opt.weight_decay, optim_opt.start_decay_at) optim.set_parameters(model.parameters()) return optim
def DPT_Hybrid(pretrained=True, **kwargs): model = DPTDepthModel(path=None, backbone='vitb_rn50_384', non_negative=True) if pretrained: checkpoint = ' state_dict = torch.hub.load_state_dict_from_url(checkpoint, map_location=torch.device('cpu'), progress=True, check_hash=True) model.load_state_dict(state_dict) return model
def identify_and_tag_authors(line, authors_kb): (re_auth, re_auth_near_miss) = get_author_regexps() for (pattern, repl) in authors_kb: line = line.replace(pattern, repl) output_line = line line = strip_tags(output_line) matched_authors = list(re_auth.finditer(line)) unidecoded_line = strip_tags(unidecode(output_line)) matched_authors_unidecode = list(re_auth.finditer(unidecoded_line)) if (len(matched_authors_unidecode) > len(matched_authors)): output_line = unidecode(output_line) matched_authors = matched_authors_unidecode if matched_authors: matched_positions = [] preceeding_text_string = line preceeding_text_start = 0 for (auth_no, match) in enumerate(matched_authors): if (line[match.start():match.end()].find('_') == (- 1)): matched_positions.append({'start': match.start(), 'end': match.end(), 'etal': (match.group('et') or match.group('et2')), 'ed_start': match.group('es'), 'ed_end': match.group('ee'), 'multi_auth': match.group('multi_auth'), 'multi_surs': match.group('multi_surs'), 'text_before': preceeding_text_string[preceeding_text_start:match.start()], 'auth_no': auth_no, 'author_names': match.group('author_names')}) preceeding_text_start = match.end() matched_positions.reverse() for m in matched_positions: dump_in_misc = False start = m['start'] end = m['end'] lower_text_before = m['text_before'].strip().lower() for e in etal_matches: if lower_text_before.endswith(e): dump_in_misc = True break if ((not dump_in_misc) and (not (m['multi_auth'] or m['multi_surs'])) and lower_text_before.endswith(' and')): weaker_match = re_auth_near_miss.match(m['text_before']) if (weaker_match and (not (weaker_match.group('es') or weaker_match.group('ee')))): start = (start - (len(m['text_before']) - weaker_match.start())) else: dump_in_misc = True add_to_misc = '' if (len(output_line) > m['end']): if (output_line[m['end']].strip(' ,.') == ';'): add_to_misc = ';' tmp_output_line = re.sub(re_ed_notation, '(ed.)', output_line[start:end], re.IGNORECASE) tmp_output_line = re.sub(re_etal, 'et al.', tmp_output_line, re.IGNORECASE) tmp_output_line = tmp_output_line.lstrip('.').strip(',:;- [](') if (not tmp_output_line.endswith('(ed.)')): tmp_output_line = tmp_output_line.strip(')') if (m['etal'] and (not (m['ed_start'] or m['ed_end'] or dump_in_misc))): output_line = (((((output_line[:start] + '<cds.AUTHetal>') + tmp_output_line) + CFG_REFEXTRACT_MARKER_CLOSING_AUTHOR_ETAL) + add_to_misc) + output_line[end:]) elif (not (m['ed_start'] or m['ed_end'] or dump_in_misc)): output_line = (((((output_line[:start] + '<cds.AUTHstnd>') + tmp_output_line) + CFG_REFEXTRACT_MARKER_CLOSING_AUTHOR_STND) + add_to_misc) + output_line[end:]) elif (m['ed_start'] or m['ed_end']): ed_notation = ' (eds.)' tmp_output_line = re.sub(re_etal, 'et al.', m['author_names'], re.IGNORECASE) output_line = ((((((output_line[:start] + '<cds.AUTHincl>') + tmp_output_line.strip(',:;- [](')) + ed_notation) + CFG_REFEXTRACT_MARKER_CLOSING_AUTHOR_INCL) + add_to_misc) + output_line[end:]) return output_line
def build_feature_extractor(cfg): (model_name, backbone_name) = cfg.MODEL.NAME.split('_') if backbone_name.startswith('resnetv2'): backbone = resnet_feature_extractor_v2(backbone_name.replace('v2', ''), pretrained_weights=cfg.MODEL.WEIGHTS, aux=False, pretrained_backbone=True, eval_bn=cfg.MODEL.EVAL_BN) elif backbone_name.startswith('resnet'): backbone = resnet_feature_extractor(backbone_name, pretrained_weights=cfg.MODEL.WEIGHTS, aux=False, pretrained_backbone=True, freeze_bn=cfg.MODEL.FREEZE_BN) elif backbone_name.startswith('vgg'): backbone = vgg_feature_extractor(backbone_name, pretrained_weights=cfg.MODEL.WEIGHTS, aux=False, pretrained_backbone=True, freeze_bn=cfg.MODEL.FREEZE_BN) else: raise NotImplementedError return backbone
def GetPlanToJointStateService(): return GetService('/costar/PlanToJointState', ServoToJointState)
class LengthGroupedSampler(Sampler): def __init__(self, batch_size: int, world_size: int, lengths: Optional[List[int]]=None, generator=None, group_by_modality: bool=False): if (lengths is None): raise ValueError('Lengths must be provided.') self.batch_size = batch_size self.world_size = world_size self.lengths = lengths self.generator = generator self.group_by_modality = group_by_modality def __len__(self): return len(self.lengths) def __iter__(self): if self.group_by_modality: indices = get_modality_length_grouped_indices(self.lengths, self.batch_size, self.world_size, generator=self.generator) else: indices = get_length_grouped_indices(self.lengths, self.batch_size, self.world_size, generator=self.generator) return iter(indices)
def get_split_time(num_domain=2, mode='pre_process', data_file=None, station=None, dis_type='coral'): spilt_time = {'2': [('2013-3-6 0:0', '2015-5-31 23:0'), ('2015-6-2 0:0', '2016-6-30 23:0')]} if (mode == 'pre_process'): return spilt_time[str(num_domain)] if (mode == 'tdc'): return TDC(num_domain, data_file, station, dis_type=dis_type) else: print('error in mode')
(scope='session') def t2(dummy: ep.Tensor) -> ep.Tensor: return ep.arange(dummy, 7, 17, 2).float32()
def fake_output_machine(t, beam_size): assert (beam_size >= 2) if (t >= 0): values = ([0.5, 0.3] + [0.001 for _ in range((beam_size - 2))]) indices = (random.sample([1, 2, 3, 4], k=2) + [0 for _ in range((beam_size - 2))]) return (values, indices)
def partition_data(datadir, partition, n_nets, alpha, logger): logger.info('partition data') (X_train, y_train, X_test, y_test) = load_tiny_data(datadir) n_train = X_train.shape[0] if (partition == 'homo'): total_num = n_train idxs = np.random.permutation(total_num) batch_idxs = np.array_split(idxs, n_nets) net_dataidx_map = {i: batch_idxs[i] for i in range(n_nets)} elif (partition == 'hetero'): min_size = 0 K = 200 N = y_train.shape[0] logger.info(('N = ' + str(N))) net_dataidx_map = {} while (min_size < 200): idx_batch = [[] for _ in range(n_nets)] for k in range(K): idx_k = np.where((y_train == k))[0] np.random.shuffle(idx_k) proportions = np.random.dirichlet(np.repeat(alpha, n_nets)) proportions = np.array([(p * (len(idx_j) < (N / n_nets))) for (p, idx_j) in zip(proportions, idx_batch)]) proportions = (proportions / proportions.sum()) proportions = (np.cumsum(proportions) * len(idx_k)).astype(int)[:(- 1)] idx_batch = [(idx_j + idx.tolist()) for (idx_j, idx) in zip(idx_batch, np.split(idx_k, proportions))] min_size = min([len(idx_j) for idx_j in idx_batch]) for j in range(n_nets): np.random.shuffle(idx_batch[j]) net_dataidx_map[j] = idx_batch[j] elif (partition == 'n_cls'): n_client = n_nets n_cls = 200 n_data_per_clnt = (len(y_train) / n_client) clnt_data_list = np.random.lognormal(mean=np.log(n_data_per_clnt), sigma=0, size=n_client) clnt_data_list = ((clnt_data_list / np.sum(clnt_data_list)) * len(y_train)).astype(int) cls_priors = np.zeros(shape=(n_client, n_cls)) for i in range(n_client): cls_priors[i][random.sample(range(n_cls), int(alpha))] = (1.0 / alpha) prior_cumsum = np.cumsum(cls_priors, axis=1) idx_list = [np.where((y_train == i))[0] for i in range(n_cls)] cls_amount = [len(idx_list[i]) for i in range(n_cls)] net_dataidx_map = {} for j in range(n_client): net_dataidx_map[j] = [] while (np.sum(clnt_data_list) != 0): curr_clnt = np.random.randint(n_client) if (clnt_data_list[curr_clnt] <= 0): continue clnt_data_list[curr_clnt] -= 1 curr_prior = prior_cumsum[curr_clnt] while True: cls_label = np.argmax((np.random.uniform() <= curr_prior)) if (cls_amount[cls_label] <= 0): cls_amount[cls_label] = np.random.randint(0, len(idx_list[cls_label])) continue cls_amount[cls_label] -= 1 net_dataidx_map[curr_clnt].append(idx_list[cls_label][cls_amount[cls_label]]) break elif (partition == 'dir'): n_client = n_nets n_cls = 200 n_data_per_clnt = (len(y_train) / n_client) clnt_data_list = np.random.lognormal(mean=np.log(n_data_per_clnt), sigma=0, size=n_client) clnt_data_list = ((clnt_data_list / np.sum(clnt_data_list)) * len(y_train)).astype(int) cls_priors = np.random.dirichlet(alpha=([alpha] * n_cls), size=n_client) prior_cumsum = np.cumsum(cls_priors, axis=1) idx_list = [np.where((y_train == i))[0] for i in range(n_cls)] cls_amount = [len(idx_list[i]) for i in range(n_cls)] net_dataidx_map = {} for j in range(n_client): net_dataidx_map[j] = [] while (np.sum(clnt_data_list) != 0): curr_clnt = np.random.randint(n_client) if (clnt_data_list[curr_clnt] <= 0): continue clnt_data_list[curr_clnt] -= 1 curr_prior = prior_cumsum[curr_clnt] while True: cls_label = np.argmax((np.random.uniform() <= curr_prior)) if (cls_amount[cls_label] <= 0): continue cls_amount[cls_label] -= 1 net_dataidx_map[curr_clnt].append(idx_list[cls_label][cls_amount[cls_label]]) break elif (partition == 'my_part'): n_shards = alpha n_client = n_nets n_cls = 200 n_data_per_clnt = (len(y_train) / n_client) clnt_data_list = np.random.lognormal(mean=np.log(n_data_per_clnt), sigma=0, size=n_client) clnt_data_list = ((clnt_data_list / np.sum(clnt_data_list)) * len(y_train)).astype(int) cls_priors = np.zeros(shape=(n_client, n_cls)) cls_priors_tmp = np.random.dirichlet(alpha=([0.3] * n_cls), size=int(n_shards)) for i in range(n_client): cls_priors[i] = cls_priors_tmp[int((i / n_shards))] prior_cumsum = np.cumsum(cls_priors, axis=1) idx_list = [np.where((y_train == i))[0] for i in range(n_cls)] cls_amount = [len(idx_list[i]) for i in range(n_cls)] net_dataidx_map = {} for j in range(n_client): net_dataidx_map[j] = [] while (np.sum(clnt_data_list) != 0): curr_clnt = np.random.randint(n_client) if (clnt_data_list[curr_clnt] <= 0): continue clnt_data_list[curr_clnt] -= 1 curr_prior = prior_cumsum[curr_clnt] while True: cls_label = np.argmax((np.random.uniform() <= curr_prior)) if (cls_amount[cls_label] <= 0): cls_amount[cls_label] = np.random.randint(0, len(idx_list[cls_label])) continue cls_amount[cls_label] -= 1 net_dataidx_map[curr_clnt].append(idx_list[cls_label][cls_amount[cls_label]]) break traindata_cls_counts = record_net_data_stats(y_train, net_dataidx_map) return (X_train, y_train, X_test, y_test, net_dataidx_map, traindata_cls_counts)
_module() class BaseConvBboxHead(nn.Module): def __init__(self, in_channels=0, shared_conv_channels=(), cls_conv_channels=(), num_cls_out_channels=0, reg_conv_channels=(), num_reg_out_channels=0, conv_cfg=dict(type='Conv1d'), norm_cfg=dict(type='BN1d'), act_cfg=dict(type='ReLU'), bias='auto', *args, **kwargs): super(BaseConvBboxHead, self).__init__(*args, **kwargs) assert (in_channels > 0) assert (num_cls_out_channels > 0) assert (num_reg_out_channels > 0) self.in_channels = in_channels self.shared_conv_channels = shared_conv_channels self.cls_conv_channels = cls_conv_channels self.num_cls_out_channels = num_cls_out_channels self.reg_conv_channels = reg_conv_channels self.num_reg_out_channels = num_reg_out_channels self.conv_cfg = conv_cfg self.norm_cfg = norm_cfg self.act_cfg = act_cfg self.bias = bias if (len(self.shared_conv_channels) > 0): self.shared_convs = self._add_conv_branch(self.in_channels, self.shared_conv_channels) out_channels = self.shared_conv_channels[(- 1)] else: out_channels = self.in_channels prev_channel = out_channels if (len(self.cls_conv_channels) > 0): self.cls_convs = self._add_conv_branch(prev_channel, self.cls_conv_channels) prev_channel = self.cls_conv_channels[(- 1)] self.conv_cls = build_conv_layer(conv_cfg, in_channels=prev_channel, out_channels=num_cls_out_channels, kernel_size=1) prev_channel = out_channels if (len(self.reg_conv_channels) > 0): self.reg_convs = self._add_conv_branch(prev_channel, self.reg_conv_channels) prev_channel = self.reg_conv_channels[(- 1)] self.conv_reg = build_conv_layer(conv_cfg, in_channels=prev_channel, out_channels=num_reg_out_channels, kernel_size=1) def _add_conv_branch(self, in_channels, conv_channels): conv_spec = ([in_channels] + list(conv_channels)) conv_layers = nn.Sequential() for i in range((len(conv_spec) - 1)): conv_layers.add_module(f'layer{i}', ConvModule(conv_spec[i], conv_spec[(i + 1)], kernel_size=1, padding=0, conv_cfg=self.conv_cfg, norm_cfg=self.norm_cfg, act_cfg=self.act_cfg, bias=self.bias, inplace=True)) return conv_layers def init_weights(self): pass def forward(self, feats): if (len(self.shared_conv_channels) > 0): x = self.shared_convs(feats) x_cls = x x_reg = x if (len(self.cls_conv_channels) > 0): x_cls = self.cls_convs(x_cls) cls_score = self.conv_cls(x_cls) if (len(self.reg_conv_channels) > 0): x_reg = self.reg_convs(x_reg) bbox_pred = self.conv_reg(x_reg) return (cls_score, bbox_pred)
class Scale(object): def __init__(self, size): self.size = size def __call__(self, sample): (image, depth) = (sample['image'], sample['depth']) image = self.changeScale(image, self.size) depth = self.changeScale(depth, self.size, Image.NEAREST) return {'image': image, 'depth': depth} def changeScale(self, img, size, interpolation=Image.BILINEAR): if (not _is_pil_image(img)): raise TypeError('img should be PIL Image. Got {}'.format(type(img))) if (not (isinstance(size, int) or (isinstance(size, collections.Iterable) and (len(size) == 2)))): raise TypeError('Got inappropriate size arg: {}'.format(size)) if isinstance(size, int): (w, h) = img.size if (((w <= h) and (w == size)) or ((h <= w) and (h == size))): return img if (w < h): ow = size oh = int(((size * h) / w)) return img.resize((ow, oh), interpolation) else: oh = size ow = int(((size * w) / h)) return img.resize((ow, oh), interpolation) else: return img.resize(size[::(- 1)], interpolation)
class HierarchicalDecoder(nn.Module): def __init__(self, num_convolutions=3, filters=(256, 128, 64, 32, 16), latent_dim=100, output_size=(1, 128, 128), upconv=False, use_weight_norm=False, use_spectral_norm=False, hierarchical_layers=(1, 3, 5), context_dim=4, div_factor=8): super().__init__() self.num_convolutions = num_convolutions self.filters = filters self.upconv = upconv if (use_weight_norm and use_spectral_norm): raise ValueError('Cannot use both weight norm and spectral norm.') self.use_weight_norm = use_weight_norm self.use_spectral_norm = use_spectral_norm self.hierarchical_layers = hierarchical_layers hierarchical_layers_ = [h for h in hierarchical_layers if (h != len(filters))] self.context_dim = context_dim self.div_factor = div_factor self.resolution_layers = nn.ModuleList([]) self.up_layers = nn.ModuleList([]) self.intermediate_shapes = [] self.context_attention = nn.ModuleList([]) cur_channels = filters[0] self.start_context_attention = self._attn(cur_channels) self.start_up_layer = nn.Sequential(*self._upsample_layer(cur_channels, cur_channels)) if (len(filters) in hierarchical_layers): self.intermediate_shapes.append((np.array(output_size) // (2 ** len(filters)))) self.intermediate_shapes[(- 1)][0] = cur_channels for (i, c) in enumerate(filters[1:], 1): resolution_layer = [] i = (len(filters) - i) input_layer = (i in hierarchical_layers_) in_channels = max((cur_channels // div_factor), 1) for j in range(0, (num_convolutions - 1)): ci = ((in_channels + cur_channels) if ((j == 0) and input_layer) else cur_channels) resolution_layer += self._conv_layer(ci, cur_channels) self.resolution_layers.append(nn.Sequential(*resolution_layer)) self.context_attention.append(self._attn(cur_channels)) self.up_layers.append(nn.Sequential(*self._upsample_layer(cur_channels, c))) if input_layer: self.intermediate_shapes.append((np.array(output_size) // (2 ** i))) self.intermediate_shapes[(- 1)][0] = in_channels cur_channels = c final_layer = self._conv_layer(cur_channels, cur_channels) final_layer.append(self._conv(cur_channels, output_size[0], 1, 1, bias=True)) self.final_layer = nn.Sequential(*final_layer) self.fc = nn.Sequential(nn.Linear(latent_dim, np.prod(self.intermediate_shapes[0]), bias=False), nn.BatchNorm1d(np.prod(self.intermediate_shapes[0])), nn.LeakyReLU(0.1, inplace=True)) def _conv(self): return partial(Conv2d, use_weight_norm=self.use_weight_norm, use_spectral_norm=self.use_spectral_norm) def _conv_transpose(self): return partial(ConvTranspose2d, use_weight_norm=self.use_weight_norm, use_spectral_norm=self.use_spectral_norm) def _conv_layer(self, ci, co): return [self._conv(ci, co, 3, 1, 1, bias=False), nn.BatchNorm2d(co, momentum=0.05), nn.LeakyReLU(0.1, inplace=True)] def _upsample_layer(self, ci, co): if self.upconv: layer = [nn.Upsample(scale_factor=2, mode='nearest'), self._conv(ci, co, kernel_size=5, stride=1, padding=2, bias=False)] else: layer = [self._conv_transpose(ci, co, kernel_size=4, stride=2, padding=1, bias=False)] layer += [nn.BatchNorm2d(co, momentum=0.05), nn.LeakyReLU(0.1, inplace=True)] return layer def _attn(self, co): hidden_dim = max((co // 4), self.context_dim) return nn.Sequential(nn.Linear(self.context_dim, hidden_dim), nn.LeakyReLU(0.1, inplace=True), nn.Linear(hidden_dim, co), nn.Sigmoid()) def forward(self, x, ctx): assert (x[0].size(0) == ctx.size(0)) batch_size = ctx.size(0) layers = zip(self.resolution_layers, self.up_layers, self.context_attention) ctx_attn = self.start_context_attention(ctx).view(batch_size, (- 1), 1, 1) y = self.fc(x.pop()).view((- 1), *self.intermediate_shapes[0]) y = (self.start_up_layer(y) * ctx_attn) for (i, (conv, up, attn)) in enumerate(layers, 1): i = (len(self.filters) - i) output_layer = (i in self.hierarchical_layers) ctx_attn = attn(ctx).view(batch_size, (- 1), 1, 1) if output_layer: y = torch.cat([y, x.pop()], 1) y = (conv(y) * ctx_attn) y = up(y) y = self.final_layer(y) return y
def fit_to_block_size(sequence, block_size, pad_token_id): if (len(sequence) > block_size): return sequence[:block_size] else: sequence.extend(([pad_token_id] * (block_size - len(sequence)))) return sequence
class SparseMaxPool2d(SparseMaxPool): def __init__(self, kernel_size, stride=1, padding=0, dilation=1): super(SparseMaxPool2d, self).__init__(2, kernel_size, stride, padding, dilation)
class GCKNet(nn.Module): def __init__(self, nclass, input_size, hidden_sizes, path_sizes, kernel_funcs=None, kernel_args_list=None, pooling='mean', global_pooling='sum', heads=1, out_size=3, max_iter=100, eps=0.1, aggregation=False, weight_decay=0.0, batch_norm=False, **kwargs): super().__init__() self.features = GCKNetFeature(input_size, hidden_sizes, path_sizes, kernel_funcs, kernel_args_list, pooling, global_pooling, heads, out_size, max_iter, eps, aggregation, **kwargs) self.output_size = self.features.output_size self.nclass = nclass self.batch_norm = batch_norm if batch_norm: self.bn_layer = nn.BatchNorm1d(self.output_size) self.classifier = Linear(self.output_size, nclass, weight_decay) def reset_parameters(self): for layer in self.children(): layer.reset_parameters() def representation(self, input, paths_indices, other_info): return self.features(input, paths_indices, other_info) def forward(self, input, paths_indices, other_info): features = self.representation(input, paths_indices, other_info) if self.batch_norm: features = self.bn_layer(features) return self.classifier(features) def unsup_train(self, data_loader, n_sampling_paths=100000, init=None, use_cuda=False): self.features.unsup_train(data_loader=data_loader, n_sampling_paths=n_sampling_paths, init=init, use_cuda=use_cuda) def unsup_train_classifier(self, data_loader, criterion, use_cuda=False): (encoded_data, labels) = self.features.predict(data_loader, use_cuda) print(encoded_data.shape) self.classifier.fit(encoded_data, labels, criterion)
class Epanechnikov(torch.nn.Module): def __init__(self): super().__init__() self.support = ((- 1.0), 1.0) def forward(self, u): return (((((- 1.0) <= u) * (u <= 1.0)) * (1.0 - (u ** 2))) * 0.75)
def get_mask_paths(metadata): mask_paths = {} ignore_paths = {} with open(metadata.localization) as f: for line in f.readlines(): (image_id, mask_path, ignore_path) = line.strip('\n').split(',') if (image_id in mask_paths): mask_paths[image_id].append(mask_path) assert (len(ignore_path) == 0) else: mask_paths[image_id] = [mask_path] ignore_paths[image_id] = ignore_path return (mask_paths, ignore_paths)
def new_func(*args, **kwds): default_in_kw = kwds.get('default', None) default_in_args = (args[2] if (len(args) > 2) else None) default = (default_in_kw or default_in_args) try: return config_tree_get_ori(*args, **kwds) except ConfigMissingException: logger.info(f'key {args[1]}, def {default}') return default
def parse_args(): parser = argparse.ArgumentParser(description='Finetune a transformers model on a text classification task') parser.add_argument('--task_name', type=str, default=None, help='The name of the glue task to train on.', choices=list(task_to_keys.keys())) parser.add_argument('--train_file', type=str, default=None, help='A csv or a json file containing the training data.') parser.add_argument('--validation_file', type=str, default=None, help='A csv or a json file containing the validation data.') parser.add_argument('--max_length', type=int, default=128, help='The maximum total input sequence length after tokenization. Sequences longer than this will be truncated, sequences shorter will be padded if `--pad_to_max_length` is passed.') parser.add_argument('--pad_to_max_length', action='store_true', help='If passed, pad all samples to `max_length`. Otherwise, dynamic padding is used.') parser.add_argument('--model_name_or_path', type=str, help='Path to pretrained model or model identifier from huggingface.co/models.', required=True) parser.add_argument('--use_slow_tokenizer', action='store_true', help='If passed, will use a slow tokenizer (not backed by the Tokenizers library).') parser.add_argument('--per_device_train_batch_size', type=int, default=8, help='Batch size (per device) for the training dataloader.') parser.add_argument('--per_device_eval_batch_size', type=int, default=8, help='Batch size (per device) for the evaluation dataloader.') parser.add_argument('--learning_rate', type=float, default=5e-05, help='Initial learning rate (after the potential warmup period) to use.') parser.add_argument('--weight_decay', type=float, default=0.0, help='Weight decay to use.') parser.add_argument('--num_train_epochs', type=int, default=3, help='Total number of training epochs to perform.') parser.add_argument('--max_train_steps', type=int, default=None, help='Total number of training steps to perform. If provided, overrides num_train_epochs.') parser.add_argument('--gradient_accumulation_steps', type=int, default=1, help='Number of updates steps to accumulate before performing a backward/update pass.') parser.add_argument('--lr_scheduler_type', type=SchedulerType, default='linear', help='The scheduler type to use.', choices=['linear', 'cosine', 'cosine_with_restarts', 'polynomial', 'constant', 'constant_with_warmup']) parser.add_argument('--num_warmup_steps', type=int, default=0, help='Number of steps for the warmup in the lr scheduler.') parser.add_argument('--output_dir', type=str, default=None, help='Where to store the final model.') parser.add_argument('--seed', type=int, default=None, help='A seed for reproducible training.') parser.add_argument('--push_to_hub', action='store_true', help='Whether or not to push the model to the Hub.') parser.add_argument('--hub_model_id', type=str, help='The name of the repository to keep in sync with the local `output_dir`.') parser.add_argument('--hub_token', type=str, help='The token to use to push to the Model Hub.') parser.add_argument('--checkpointing_steps', type=str, default=None, help="Whether the various states should be saved at the end of every n steps, or 'epoch' for each epoch.") parser.add_argument('--resume_from_checkpoint', type=str, default=None, help='If the training should continue from a checkpoint folder.') parser.add_argument('--with_tracking', action='store_true', help='Whether to enable experiment trackers for logging.') parser.add_argument('--report_to', type=str, default='all', help='The integration to report the results and logs to. Supported platforms are `"tensorboard"`, `"wandb"`, `"comet_ml"` and `"clearml"`. Use `"all"` (default) to report to all integrations.Only applicable when `--with_tracking` is passed.') parser.add_argument('--ignore_mismatched_sizes', action='store_true', help='Whether or not to enable to load a pretrained model whose head dimensions are different.') args = parser.parse_args() if ((args.task_name is None) and (args.train_file is None) and (args.validation_file is None)): raise ValueError('Need either a task name or a training/validation file.') else: if (args.train_file is not None): extension = args.train_file.split('.')[(- 1)] assert (extension in ['csv', 'json']), '`train_file` should be a csv or a json file.' if (args.validation_file is not None): extension = args.validation_file.split('.')[(- 1)] assert (extension in ['csv', 'json']), '`validation_file` should be a csv or a json file.' if args.push_to_hub: assert (args.output_dir is not None), 'Need an `output_dir` to create a repo when `--push_to_hub` is passed.' return args
def _find_stack(stacks, item): for (stack_id, stack) in enumerate(stacks): if (stack[0] == item): return (stack, stack_id) return (None, None)
class FlaxElectraForMaskedLM(metaclass=DummyObject): _backends = ['flax'] def __init__(self, *args, **kwargs): requires_backends(self, ['flax'])
def loadmat(file): f = h5py.File(file) arr = {k: np.array(v) for (k, v) in f.items()} return arr
_model_architecture('transformer_lm', 'transformer_lm_gpt3_2_7') def transformer_lm_gpt3_2_7(args): args.decoder_layers = getattr(args, 'decoder_layers', 32) args.decoder_embed_dim = getattr(args, 'decoder_embed_dim', 2560) args.decoder_attention_heads = getattr(args, 'decoder_attention_heads', 32) base_gpt3_architecture(args)