code
stringlengths
101
5.91M
def dense_layer(inp: int, out: int, activation: str, p: float, bn: bool, linear_first: bool): act_fn = get_activation_fn(activation) layers = ([nn.BatchNorm1d((out if linear_first else inp))] if bn else []) if (p != 0): layers.append(nn.Dropout(p)) lin = [nn.Linear(inp, out, bias=(not bn)), act_fn] layers = ((lin + layers) if linear_first else (layers + lin)) return nn.Sequential(*layers)
def shared_convl1_lrelu(shape, nb_filters, kernel, stride=(1, 1), **kwargs): c = Convolution2D(nb_filters, kernel, padding='same', kernel_initializer='he_uniform', kernel_regularizer=l1(0.01), strides=(stride, stride), input_shape=shape) l = LeakyReLU() return Sequential([c, l], **kwargs)
class MostVisitedPositiveExtract(AbstractExtract): def __call__(self, node): nodes = [node] while ((not node.terminal) and (len(node.children) > 0)): if (max([child.avg_reward for child in node.children]) > 0): allowed = (lambda child: (child.avg_reward > 0)) else: allowed = (lambda child: True) visits = [(i, child.n_visits) for (i, child) in enumerate(node.children) if allowed(child)] (max_idx, max_num_visits) = max(visits, key=operator.itemgetter(1)) node = node.children[max_idx] nodes.append(node) return nodes
def parse_args(): parser = argparse.ArgumentParser(description='TangoBERT') parser.add_argument('--task_name', type=str, help='Name of the GLUE task.', choices=list(task_to_keys.keys()), required=True) parser.add_argument('--small_model_name_or_path', type=str, help='Path to the small pretrained model or model identifier from huggingface.co/models.', required=True) parser.add_argument('--big_model_name_or_path', type=str, help='Path to the big pretrained model or model identifier from huggingface.co/models.', required=True) parser.add_argument('--device_small', default=0, help='Defines the device (e.g., "cpu", "cuda:1", "mps", or a GPU ordinal rank like 1) on which small model pipeline will be allocated. ') parser.add_argument('--device_big', default=0, help='Defines the device (e.g., "cpu", "cuda:1", "mps", or a GPU ordinal rank like 1) on which this big model pipeline will be allocated. ') parser.add_argument('--per_device_eval_batch_size-small', type=int, default=32, help='Batch size (per device) for small model inference.') parser.add_argument('--per_device_eval_batch_size-big', type=int, default=32, help='Batch size (per device) for big model inference.') parser.add_argument('--confidence_threshold', type=float, default=0.85, action=range_action(0.5, 1.0), help='Confidence threshold for small model prediction (must be in range 0.5..1.0).') args = parser.parse_args() return args
def y_pred_header(outcome, underscore=False): return ((str(outcome) + ('_' if underscore else '-')) + 'y_pred1')
class HypergraphConv(MessagePassing): def __init__(self, in_channels, out_channels, symdegnorm=False, use_attention=False, heads=1, concat=True, negative_slope=0.2, dropout=0, bias=True, **kwargs): kwargs.setdefault('aggr', 'add') super(HypergraphConv, self).__init__(node_dim=0, **kwargs) self.in_channels = in_channels self.out_channels = out_channels self.use_attention = use_attention self.symdegnorm = symdegnorm if self.use_attention: self.heads = heads self.concat = concat self.negative_slope = negative_slope self.dropout = dropout self.weight = Parameter(torch.Tensor(in_channels, (heads * out_channels))) self.att = Parameter(torch.Tensor(1, heads, (2 * out_channels))) else: self.heads = 1 self.concat = True self.weight = Parameter(torch.Tensor(in_channels, out_channels)) if (bias and concat): self.bias = Parameter(torch.Tensor((heads * out_channels))) elif (bias and (not concat)): self.bias = Parameter(torch.Tensor(out_channels)) else: self.register_parameter('bias', None) self.reset_parameters() def reset_parameters(self): glorot(self.weight) if self.use_attention: glorot(self.att) zeros(self.bias) def forward(self, x: Tensor, hyperedge_index: Tensor, hyperedge_weight: Optional[Tensor]=None) -> Tensor: (num_nodes, num_edges) = (x.size(0), 0) if (hyperedge_index.numel() > 0): num_edges = (int(hyperedge_index[1].max()) + 1) if (hyperedge_weight is None): hyperedge_weight = x.new_ones(num_edges) x = torch.matmul(x, self.weight) alpha = None if self.use_attention: assert (num_edges <= num_edges) x = x.view((- 1), self.heads, self.out_channels) (x_i, x_j) = (x[hyperedge_index[0]], x[hyperedge_index[1]]) alpha = (torch.cat([x_i, x_j], dim=(- 1)) * self.att).sum(dim=(- 1)) alpha = F.leaky_relu(alpha, self.negative_slope) alpha = softmax(alpha, hyperedge_index[0], num_nodes=x.size(0)) alpha = F.dropout(alpha, p=self.dropout, training=self.training) if (not self.symdegnorm): D = scatter_add(hyperedge_weight[hyperedge_index[1]], hyperedge_index[0], dim=0, dim_size=num_nodes) D = (1.0 / D) D[(D == float('inf'))] = 0 B = scatter_add(x.new_ones(hyperedge_index.size(1)), hyperedge_index[1], dim=0, dim_size=num_edges) B = (1.0 / B) B[(B == float('inf'))] = 0 self.flow = 'source_to_target' out = self.propagate(hyperedge_index, x=x, norm=B, alpha=alpha, size=(num_nodes, num_edges)) self.flow = 'target_to_source' out = self.propagate(hyperedge_index, x=out, norm=D, alpha=alpha, size=(num_edges, num_nodes)) else: D = scatter_add(hyperedge_weight[hyperedge_index[1]], hyperedge_index[0], dim=0, dim_size=num_nodes) D = (1.0 / (D ** 0.5)) D[(D == float('inf'))] = 0 B = scatter_add(x.new_ones(hyperedge_index.size(1)), hyperedge_index[1], dim=0, dim_size=num_edges) B = (1.0 / B) B[(B == float('inf'))] = 0 x = (D.unsqueeze((- 1)) * x) self.flow = 'source_to_target' out = self.propagate(hyperedge_index, x=x, norm=B, alpha=alpha, size=(num_nodes, num_edges)) self.flow = 'target_to_source' out = self.propagate(hyperedge_index, x=out, norm=D, alpha=alpha, size=(num_edges, num_nodes)) if (self.concat is True): out = out.view((- 1), (self.heads * self.out_channels)) else: out = out.mean(dim=1) if (self.bias is not None): out = (out + self.bias) return out def message(self, x_j: Tensor, norm_i: Tensor, alpha: Tensor) -> Tensor: (H, F) = (self.heads, self.out_channels) out = (norm_i.view((- 1), 1, 1) * x_j.view((- 1), H, F)) if (alpha is not None): out = (alpha.view((- 1), self.heads, 1) * out) return out def __repr__(self): return '{}({}, {})'.format(self.__class__.__name__, self.in_channels, self.out_channels)
def train_epoch(encoder, classifier, classifiers, batch_trains, dev, test, optimizer_encoder, optimizer_classifier, start, I): all_time = dev_time = all_tagged = this_tagged = this_loss = 0 mtl_criterion = nn.CrossEntropyLoss() moe_criterion = nn.NLLLoss() domain_encs = None for (ind, batch) in enumerate(batch_trains): optimizer_encoder.zero_grad() optimizer_classifier.zero_grad() loss_mtl = [] loss_moe = [] ms_outputs = [] hiddens = [] train_labels = [] for (source_ind, s) in enumerate(batch): I += 1 (words, golds) = zip(*s) hidden = encoder(words) outputs = [] hiddens.append(hidden) for sthi in classifiers: output = sthi(hidden) outputs.append(output) ms_outputs.append(outputs) train_labels.append(encoder.get_var(torch.LongTensor([encoder.vt.w2i[t] for t in golds]))) preds = ms_outputs[source_ind][source_ind] loss = mtl_criterion(preds, train_labels[(- 1)]) loss_mtl.append(loss) loss_mtl = sum(loss_mtl) loss = loss_mtl loss.backward() optimizer_encoder.step() optimizer_classifier.step() print('\n\nEnded last epoch.\n') print(I) return (encoder, classifier, optimizer_encoder, optimizer_classifier)
def densenet161(num_classes=1000, pretrained='imagenet'): model = models.densenet161(pretrained=False) if (pretrained is not None): settings = pretrained_settings['densenet161'][pretrained] model = load_pretrained(model, num_classes, settings) model = modify_densenets(model) return model
class EffiInvResUnit(nn.Module): def __init__(self, in_channels, out_channels, kernel_size, stride, expansion_factor, bn_eps, activation, tf_mode): super(EffiInvResUnit, self).__init__() self.kernel_size = kernel_size self.stride = stride self.tf_mode = tf_mode self.residual = ((in_channels == out_channels) and (stride == 1)) mid_channels = (in_channels * expansion_factor) dwconv_block_fn = (dwconv3x3_block if (kernel_size == 3) else (dwconv5x5_block if (kernel_size == 5) else None)) self.conv1 = conv1x1_block(in_channels=in_channels, out_channels=mid_channels, bn_eps=bn_eps, activation=activation) self.conv2 = dwconv_block_fn(in_channels=mid_channels, out_channels=mid_channels, stride=stride, padding=(0 if tf_mode else (kernel_size // 2)), bn_eps=bn_eps, activation=activation) self.se = SEBlock(channels=mid_channels, reduction=24, activation=activation) self.conv3 = conv1x1_block(in_channels=mid_channels, out_channels=out_channels, bn_eps=bn_eps, activation=None) def forward(self, x): if self.residual: identity = x x = self.conv1(x) if self.tf_mode: x = F.pad(x, pad=calc_tf_padding(x, kernel_size=self.kernel_size, stride=self.stride)) x = self.conv2(x) x = self.se(x) x = self.conv3(x) if self.residual: x = (x + identity) return x
def matching_cascade(distance_metric, max_distance, cascade_depth, tracks, detections, track_indices=None, detection_indices=None): if (track_indices is None): track_indices = list(range(len(tracks))) if (detection_indices is None): detection_indices = list(range(len(detections))) unmatched_detections = detection_indices matches = [] for level in range(cascade_depth): if (len(unmatched_detections) == 0): break track_indices_l = [k for k in track_indices if (tracks[k].time_since_update == (1 + level))] if (len(track_indices_l) == 0): continue (matches_l, _, unmatched_detections) = min_cost_matching(distance_metric, max_distance, tracks, detections, track_indices_l, unmatched_detections) matches += matches_l unmatched_tracks = list((set(track_indices) - set((k for (k, _) in matches)))) return (matches, unmatched_tracks, unmatched_detections)
def convert_tokens_seq(eval_file, qa_id, symbols, probs, id2word, map_to_orig): def _get_penalty(syms): trigrams = [tuple(syms[i:(i + 3)]) for i in range((len(syms) - 2))] repeat_trigram = (list(filter((lambda x: (x > 1)), list(Counter(trigrams).values()))) != []) return repeat_trigram answer_dict = {} for (qid, prob, bsyms) in zip(qa_id, probs, zip(*symbols)): answers = [] try: bsyms = zip(*bsyms) except: bsyms = [bsyms] for (p, syms) in zip(prob, bsyms): context_tokens = eval_file[str(qid)]['paragraph'] if (102 in syms): syms = syms[:syms.index(102)] syms = [(id2word[sym] if (sym in id2word) else context_tokens[(sym - len(id2word))]) for sym in syms] tokens = convert_to_words(map_to_orig, syms) answer = u' '.join(tokens) (lp, penalty) = ((len(tokens) + 1), 0.0) if _get_penalty(tokens): penalty = 1.0 answers.append((((p / lp) - penalty), answer)) answer_dict[str(qid)] = answers return answer_dict
def test_hash(): class Hashable(object): def __init__(self, value): self.value = value def __hash__(self): return self.value class Unhashable(object): __hash__ = None assert (m.hash_function(Hashable(42)) == 42) with pytest.raises(TypeError): m.hash_function(Unhashable())
def build_batchers(data_dir, batch_size): def coll(batch): (art_batch, abs_batch) = unzip(batch) art_sents = list(filter(bool, map(tokenize(None), art_batch))) abs_sents = list(filter(bool, map(tokenize(None), abs_batch))) return (art_sents, abs_sents) loader = DataLoader(RLDataset('train', data_dir), batch_size=batch_size, shuffle=True, num_workers=4, collate_fn=coll) val_loader = DataLoader(RLDataset('val', data_dir), batch_size=batch_size, shuffle=False, num_workers=4, collate_fn=coll) return (cycle(loader), val_loader)
def parser_sample(parser): parser.add_argument('-in', '--input_img', type=str, required=True, help='path of input image') parser.add_argument('--sigma', type=float, default=0.75, required=False, help='noise level to adjust the variatonality of the new sample. default is 0.75 (float)') parser.add_argument('--alpha', type=float, default=0.005, required=False, help='alpha parameter of the normalizing distance matrix. small alpha encourages completeness. default is 0.005 (float)') parser.add_argument('--task', type=str, default='random_sample') return parser
def fit(model, loss, opt, train_dataset, epochs, train_batch_size, max_steps=None): pbar = tqdm(train_dataset) for (i, batch) in enumerate(pbar): with tf.GradientTape() as tape: (inputs, targets) = batch outputs = model(batch) loss_value = loss(targets, outputs.logits) if SDP_ENABLED: tape = sdp.DistributedGradientTape(tape, sparse_as_dense=True) grads = tape.gradient(loss_value, model.trainable_variables) opt.apply_gradients(zip(grads, model.trainable_variables)) pbar.set_description(f'Loss: {loss_value:.4f}') if (SDP_ENABLED and (i == 0)): sdp.broadcast_variables(model.variables, root_rank=0) sdp.broadcast_variables(opt.variables(), root_rank=0) if (max_steps and (i >= max_steps)): break train_results = {'loss': loss_value.numpy()} return train_results
class FormatterNoInfo(logging.Formatter): def __init__(self, fmt='%(levelname)s: %(message)s'): logging.Formatter.__init__(self, fmt) def format(self, record): if (record.levelno == logging.INFO): return str(record.getMessage()) return logging.Formatter.format(self, record)
def test_nasfcos_fpn(): NASFCOS_FPN(in_channels=[8, 16, 32, 64], out_channels=8, start_level=0, end_level=3, num_outs=4) NASFCOS_FPN(in_channels=[8, 16, 32, 64], out_channels=8, start_level=0, end_level=(- 1), num_outs=4) with pytest.raises(AssertionError): NASFCOS_FPN(in_channels=[8, 16, 32, 64], out_channels=8, start_level=1, end_level=4, num_outs=2) with pytest.raises(AssertionError): NASFCOS_FPN(in_channels=[8, 16, 32, 64], out_channels=8, start_level=1, end_level=2, num_outs=3)
def generate_configs(base_config, dest_dir, embeddings, grid, refresh, ckpts_path, target): with open(base_config, 'r') as f: base = json.load(f) with open(ckpts_path, 'r') as f: ckpts = json.load(f) model_family = {'smallnet': {'preproc': {'crop': 15, 'imwidth': 100}, 'name': 'SmallNet'}, 'hourglass': {'preproc': {'crop': 20, 'imwidth': 136}, 'name': 'HourglassNet'}} for model_name in embeddings: tokens = model_name.split('-') if (tokens[(- 1)] == 'dve'): tokens.pop() (model_type, embedding_dim) = (tokens[(- 2)], int(tokens[(- 1)][:(- 1)])) preproc_kwargs = model_family[model_type]['preproc'] hparam_vals = [x for x in grid.values()] grid_vals = list(itertools.product(*hparam_vals)) hparams = list(grid.keys()) if ('-ft-keypoints' in target): prefix = target.replace('-keypoints', '') prefix = prefix.replace('-limit-annos', '') prefix = prefix.replace('-no-aug', '') ckpt_name = f'{prefix}-{model_name}' else: ckpt_name = model_name epoch = ckpts[ckpt_name]['epoch'] for cfg_vals in grid_vals: config = copy.deepcopy(base) for (hparam, val) in zip(hparams, cfg_vals): if (hparam == 'smax'): config['keypoint_regressor']['softmaxarg_mul'] = val elif (hparam == 'lr'): config['optimizer']['args']['lr'] = val elif (hparam == 'bs'): val = int(val) config['batch_size'] = val elif (hparam == 'upsample'): val = bool(val) config['keypoint_regressor_upsample'] = val elif (hparam == 'annos'): config['restrict_annos'] = int(val) else: raise ValueError(f'unknown hparam: {hparam}') ckpt = f'checkpoint-epoch{epoch}.pth' timestamp = ckpts[ckpt_name]['timestamp'] ckpt_path = (((Path('data/saved/models') / ckpt_name) / timestamp) / ckpt) config['arch']['type'] = model_family[model_type]['name'] config['arch']['args']['num_output_channels'] = embedding_dim config['dataset']['args'].update(preproc_kwargs) config['finetune_from'] = str(ckpt_path) if (('-ft' in str(dest_dir)) and ('-keypoints' not in str(dest_dir))): loss = 'dense_correlation_loss' if ('dve' in model_name): loss += '_dve' config['loss'] = loss if ('hourglass' in model_name): config['batch_size'] = 16 if ('annos' in grid): model_name_ = f"{config['restrict_annos']}-annos-{model_name}" else: model_name_ = model_name if (len(grid['lr']) > 1): model_name_ = f"{model_name_}-lr{config['optimizer']['args']['lr']}" if (len(grid['bs']) > 1): model_name_ = f"{model_name_}-bs{config['batch_size']}" dest_path = (Path(dest_dir) / f'{model_name_}.json') dest_path.parent.mkdir(exist_ok=True, parents=True) if ((not dest_path.exists()) or refresh): with open(str(dest_path), 'w') as f: json.dump(config, f, indent=4, sort_keys=False) else: print(f'config file at {str(dest_path)} exists, skipping....') print(f'Wrote {len(grid_vals)} configs to disk')
def test_funcall_kwarg(): run_cell('\n def f(y):\n return 2 * y + 8\n ') run_cell('x = 7') run_cell('z = f(y=x)') run_cell('x = 8') run_cell('logging.info(z)') assert_detected('`z` depends on stale `x`')
def check_models_are_tested(module, test_file): defined_models = get_models(module) tested_models = find_tested_models(test_file) if (tested_models is None): if (test_file in TEST_FILES_WITH_NO_COMMON_TESTS): return return [((f'{test_file} should define `all_model_classes` to apply common tests to the models it tests. ' + 'If this intentional, add the test filename to `TEST_FILES_WITH_NO_COMMON_TESTS` in the file ') + '`utils/check_repo.py`.')] failures = [] for (model_name, _) in defined_models: if ((model_name not in tested_models) and (model_name not in IGNORE_NON_TESTED)): failures.append((((f'{model_name} is defined in {module.__name__} but is not tested in ' + f'{os.path.join(PATH_TO_TESTS, test_file)}. Add it to the all_model_classes in that file.') + 'If common tests should not applied to that model, add its name to `IGNORE_NON_TESTED`') + 'in the file `utils/check_repo.py`.')) return failures
class InfoGen(nn.Module): def __init__(self, t_emb, output_size): super(InfoGen, self).__init__() self.tconv1 = nn.ConvTranspose2d(t_emb, 512, 3, 2, padding=1, bias=False) self.bn1 = nn.BatchNorm2d(512) self.tconv2 = nn.ConvTranspose2d(512, 128, 3, 2, padding=1, bias=False) self.bn2 = nn.BatchNorm2d(128) self.tconv3 = nn.ConvTranspose2d(128, 64, 3, 2, padding=1, bias=False) self.bn3 = nn.BatchNorm2d(64) self.tconv4 = nn.ConvTranspose2d(64, output_size, 3, (2, 1), padding=(1, 0), bias=False) self.bn4 = nn.BatchNorm2d(output_size) def forward(self, t_embedding): x = F.relu(self.bn1(self.tconv1(t_embedding))) x = F.relu(self.bn2(self.tconv2(x))) x = F.relu(self.bn3(self.tconv3(x))) x = F.relu(self.bn4(self.tconv4(x))) return x
class Hadron(Ion): def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) assert self.is_hadron()
def get_datasets(): result = [] for items in CONFIG_GROUPS.values(): result += items['datasets'] return result
.parametrize('spin, spin_angle, launch_angle, launch_direction_angle, expected', [(0, 1, 1, 1, (0, 0, 0))]) def test_spin_components(spin, spin_angle, launch_angle, launch_direction_angle, expected): (wx, wy, wz) = spin_components(spin, spin_angle, launch_angle, launch_direction_angle) for (a, b) in zip((wx, wy, wz), expected): assert (a == pytest.approx(b))
def imagenet_vgg16_pretrained(output_dim): model = torchvision.models.vgg16(pretrained=True) return _vgg_replace_fc(model, output_dim)
class InitiateNewTrainStage(BaseCallback): def __init__(self, n_envs: int=1, treshhold_type: str='succ', upper_threshold: float=0, lower_threshold: float=0, task_mode: str='staged', verbose=0): super(InitiateNewTrainStage, self).__init__(verbose=verbose) self.n_envs = n_envs self.threshhold_type = treshhold_type assert ((self.threshhold_type == 'rew') or (self.threshhold_type == 'succ')), "given theshhold type neither 'rew' or 'succ'" if ((self.threshhold_type == 'rew') and (upper_threshold == 0)): self.upper_threshold = 13 self.lower_threshold = 7 elif ((self.threshhold_type == 'succ') and (upper_threshold == 0)): self.upper_threshold = 0.85 self.lower_threshold = 0.6 else: self.upper_threshold = upper_threshold self.lower_threshold = lower_threshold assert (self.upper_threshold > self.lower_threshold), 'upper threshold has to be bigger than lower threshold' assert ((self.upper_threshold >= 0) and (self.lower_threshold >= 0)), 'upper/lower threshold have to be positive numbers' if (self.threshhold_type == 'succ'): assert ((self.upper_threshold <= 1) and (self.lower_threshold >= 0)), 'succ thresholds have to be between [1.0, 0.0]' self.verbose = verbose self.activated = bool((task_mode == 'staged')) if self.activated: rospy.set_param('/last_stage_reached', False) self._instantiate_publishers() self._trigger = Bool() self._trigger.data = True def _instantiate_publishers(self): self._publishers_next = [] self._publishers_previous = [] self._publishers_next.append(rospy.Publisher(f'/eval_sim/next_stage', Bool, queue_size=1)) self._publishers_previous.append(rospy.Publisher(f'/eval_sim/previous_stage', Bool, queue_size=1)) for env_num in range(self.n_envs): self._publishers_next.append(rospy.Publisher(f'/sim_{(env_num + 1)}/next_stage', Bool, queue_size=1)) self._publishers_previous.append(rospy.Publisher(f'/sim_{(env_num + 1)}/previous_stage', Bool, queue_size=1)) def _on_step(self, EvalObject: EvalCallback) -> bool: assert isinstance(EvalObject, EvalCallback), f'InitiateNewTrainStage must be called within EvalCallback' if self.activated: if (EvalObject.n_eval_episodes < 20): warnings.warn(('Only %d evaluation episodes considered for threshold monitoring,results might not represent agent performance well' % EvalObject.n_eval_episodes)) if (((self.threshhold_type == 'rew') and (EvalObject.best_mean_reward <= self.lower_threshold)) or ((self.threshhold_type == 'succ') and (EvalObject.last_success_rate <= self.lower_threshold))): for (i, pub) in enumerate(self._publishers_previous): pub.publish(self._trigger) if (i == 0): self.log_curr_stage(EvalObject.logger) if (((self.threshhold_type == 'rew') and (EvalObject.best_mean_reward >= self.upper_threshold)) or ((self.threshhold_type == 'succ') and (EvalObject.last_success_rate >= self.upper_threshold))): if (not rospy.get_param('/last_stage_reached')): EvalObject.best_mean_reward = (- np.inf) EvalObject.last_success_rate = (- np.inf) for (i, pub) in enumerate(self._publishers_next): pub.publish(self._trigger) if (i == 0): self.log_curr_stage(EvalObject.logger) def log_curr_stage(self, logger): time.sleep(1) curr_stage = rospy.get_param('/curr_stage', (- 1)) logger.record('train_stage/stage_idx', curr_stage)
def ChunkedSourceIterator(source_items: List, num_instances: int=1, instance_rank: int=0): chunk_size = math.ceil((len(source_items) / num_instances)) chunk = source_items[(instance_rank * chunk_size):((instance_rank + 1) * chunk_size)] return NativeCheckpointableIterator(chunk)
class keypoint_outputs(nn.Module): def __init__(self, dim_in): super().__init__() self.upsample_heatmap = (cfg.KRCNN.UP_SCALE > 1) if cfg.KRCNN.USE_DECONV: self.deconv = nn.ConvTranspose2d(dim_in, cfg.KRCNN.DECONV_DIM, cfg.KRCNN.DECONV_KERNEL, 2, padding=(int((cfg.KRCNN.DECONV_KERNEL / 2)) - 1)) dim_in = cfg.KRCNN.DECONV_DIM if cfg.KRCNN.USE_DECONV_OUTPUT: self.classify = nn.ConvTranspose2d(dim_in, cfg.KRCNN.NUM_KEYPOINTS, cfg.KRCNN.DECONV_KERNEL, 2, padding=int(((cfg.KRCNN.DECONV_KERNEL / 2) - 1))) else: self.classify = nn.Conv2d(dim_in, cfg.KRCNN.NUM_KEYPOINTS, 1, 1, padding=0) if self.upsample_heatmap: self.upsample = mynn.BilinearInterpolation2d(cfg.KRCNN.NUM_KEYPOINTS, cfg.KRCNN.NUM_KEYPOINTS, cfg.KRCNN.UP_SCALE) self._init_weights() def _init_weights(self): if cfg.KRCNN.USE_DECONV: init.normal_(self.deconv.weight, std=0.01) init.constant_(self.deconv.bias, 0) if (cfg.KRCNN.CONV_INIT == 'GaussianFill'): init.normal_(self.classify.weight, std=0.001) elif (cfg.KRCNN.CONV_INIT == 'MSRAFill'): mynn.init.MSRAFill(self.classify.weight) else: raise ValueError(cfg.KRCNN.CONV_INIT) init.constant_(self.classify.bias, 0) def detectron_weight_mapping(self): detectron_weight_mapping = {} if cfg.KRCNN.USE_DECONV: detectron_weight_mapping.update({'deconv.weight': 'kps_deconv_w', 'deconv.bias': 'kps_deconv_b'}) if self.upsample_heatmap: blob_name = 'kps_score_lowres' detectron_weight_mapping.update({'upsample.upconv.weight': None, 'upsample.upconv.bias': None}) else: blob_name = 'kps_score' detectron_weight_mapping.update({'classify.weight': (blob_name + '_w'), 'classify.bias': (blob_name + '_b')}) return (detectron_weight_mapping, []) def forward(self, x): if cfg.KRCNN.USE_DECONV: x = F.relu(self.deconv(x), inplace=True) x = self.classify(x) if self.upsample_heatmap: x = self.upsample(x) return x
class TestClassAssociationRule(unittest.TestCase): def test_compare(self): row1 = [1, 1, 0] header1 = ['A', 'B', 'C'] transaction1 = Transaction(row1, header1, ('Class', 0)) item1 = Item('A', 1) item2 = Item('B', 1) item3 = Item('C', 0) item4 = Item('B', 5) ant1 = Antecedent([item1, item2]) ant2 = Antecedent([item2]) ant3 = Antecedent([item3]) ant4 = Antecedent([item4]) cons1 = Consequent('Y', 1) cons2 = Consequent('Y', 2) cons3 = Consequent('Y', 3) car1 = ClassAssocationRule(ant1, cons1, 0.5, 0.9) car2 = ClassAssocationRule(ant2, cons2, 0.5, 0.9) car3 = ClassAssocationRule(ant3, cons3, 0.5, 0.9) car4 = ClassAssocationRule(ant4, cons3, 0.5, 1) sorted_cars = sorted([car1, car2, car3, car4], reverse=True) assert (car1 < car2) assert (car2 > car3) assert (car3 < car2) assert (car4 > car3) assert (car1.antecedent <= transaction1) assert (car2.antecedent <= transaction1) assert (car3.antecedent <= transaction1) assert (not (car4.antecedent <= transaction1)) assert (sorted_cars[0] == car4) def test_len(self): item1 = Item('A', 1) item2 = Item('B', 1) ant1 = Antecedent([item1, item2]) cons1 = Consequent('Y', 1) car1 = ClassAssocationRule(ant1, cons1, 0.5, 0.9) assert (len(car1) == 3)
class SparseMM(torch.autograd.Function): def forward(self, matrix1, matrix2): self.save_for_backward(matrix1, matrix2) return torch.mm(matrix1, matrix2) def backward(self, grad_output): (matrix1, matrix2) = self.saved_tensors grad_matrix1 = grad_matrix2 = None if self.needs_input_grad[0]: grad_matrix1 = torch.mm(grad_output, matrix2.t()) if self.needs_input_grad[1]: grad_matrix2 = torch.mm(matrix1.t(), grad_output) return (grad_matrix1, grad_matrix2)
def test_chroma_cqt(waveform): chroma_cqt = waveform.chroma_cqt() assert isinstance(chroma_cqt, np.ndarray)
def train(model): model = model.cuda() optimizer = torch.optim.Adam(model.parameters(), lr=0.0001) criterion = nn.L1Loss().cuda() losses = [] for _ in range(500): (x, y) = gen_data(batch_size=(2 ** 10), max_length=10) (x, y) = (torch.from_numpy(x).float().cuda(), torch.from_numpy(y).float().cuda()) loss = criterion(model(x), y) optimizer.zero_grad() loss.backward() optimizer.step() losses.append(loss.item()) return losses
class AttrDict(dict): IMMUTABLE = '__immutable__' def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) self.__dict__[AttrDict.IMMUTABLE] = False def __getattr__(self, name): if (name in self.__dict__): return self.__dict__[name] elif (name in self): return self[name] else: raise AttributeError(name) def __setattr__(self, name, value): if (not self.__dict__[AttrDict.IMMUTABLE]): if (name in self.__dict__): self.__dict__[name] = value else: self[name] = value else: raise AttributeError('Attempted to set "{}" to "{}", but AttrDict is immutable'.format(name, value)) def immutable(self, is_immutable): self.__dict__[AttrDict.IMMUTABLE] = is_immutable for v in self.__dict__.values(): if isinstance(v, AttrDict): v.immutable(is_immutable) for v in self.values(): if isinstance(v, AttrDict): v.immutable(is_immutable) def is_immutable(self): return self.__dict__[AttrDict.IMMUTABLE]
class SingleLinearClassifier(nn.Module): def __init__(self, hidden_size, num_label): super(SingleLinearClassifier, self).__init__() self.num_label = num_label self.classifier = nn.Linear(hidden_size, num_label) def forward(self, input_features): features_output = self.classifier(input_features) return features_output
def _groupByClip(dict_text: Dict[(str, str)]): sentence_ids = list(dict_text.keys()) sentence_ids.sort(key=natural_keys) dict_text_video = {} for utt_id in sentence_ids: if (utt_id[:11] in dict_text_video): dict_text_video[utt_id[:11]] += dict_text[utt_id].replace('\n', ' ') else: dict_text_video[utt_id[:11]] = dict_text[utt_id].replace('\n', ' ') return dict_text_video
class ConcatLayer(MergeLayer): def __init__(self, incomings, axis=1, cropping=None, **kwargs): super(ConcatLayer, self).__init__(incomings, **kwargs) self.axis = axis if (cropping is not None): cropping = list(cropping) cropping[axis] = None self.cropping = cropping def get_output_shape_for(self, input_shapes): input_shapes = autocrop_array_shapes(input_shapes, self.cropping) output_shape = [next((s for s in sizes if (s is not None)), None) for sizes in zip(*input_shapes)] def match(shape1, shape2): return ((len(shape1) == len(shape2)) and all((((i == self.axis) or (s1 is None) or (s2 is None) or (s1 == s2)) for (i, (s1, s2)) in enumerate(zip(shape1, shape2))))) if (not all((match(shape, output_shape) for shape in input_shapes))): raise ValueError('Mismatch: input shapes must be the same except in the concatenation axis') sizes = [input_shape[self.axis] for input_shape in input_shapes] concat_size = (None if any(((s is None) for s in sizes)) else sum(sizes)) output_shape[self.axis] = concat_size return tuple(output_shape) def get_output_for(self, inputs, **kwargs): inputs = autocrop(inputs, self.cropping) return T.concatenate(inputs, axis=self.axis)
class BasicBlock(nn.Module): expansion = 1 def __init__(self, inplanes, planes, shortcut, bn_aff, stride=1, downsample=None, groups=1, base_width=64, dilation=1, norm_layer=None): super(BasicBlock, self).__init__() if (norm_layer is None): norm_layer = nn.BatchNorm2d if ((groups != 1) or (base_width != 64)): raise ValueError('BasicBlock only supports groups=1 and base_width=64') if (dilation > 1): raise NotImplementedError('Dilation > 1 not supported in BasicBlock') self.bn_aff = bn_aff self.shortcut = shortcut self.conv1 = conv3x3(inplanes, planes, stride) self.bn1 = norm_layer(planes, affine=self.bn_aff) self.relu = nn.ReLU(inplace=True) self.conv2 = conv3x3(planes, planes) self.bn2 = norm_layer(planes, affine=self.bn_aff) self.downsample = downsample self.stride = stride def forward(self, x): identity = x out = self.conv1(x) out = self.bn1(out) out = self.relu(out) out = self.conv2(out) out = self.bn2(out) if (self.downsample is not None): identity = self.downsample(x) if self.shortcut: out += identity out = self.relu(out) return out
(allow_output_mutation=True) def load_models(): if LOAD_DENSE_INDEX: qar_tokenizer = AutoTokenizer.from_pretrained('yjernite/retribert-base-uncased') qar_model = AutoModel.from_pretrained('yjernite/retribert-base-uncased').to('cuda:0') _ = qar_model.eval() else: (qar_tokenizer, qar_model) = (None, None) if (MODEL_TYPE == 'bart'): s2s_tokenizer = AutoTokenizer.from_pretrained('yjernite/bart_eli5') s2s_model = AutoModelForSeq2SeqLM.from_pretrained('yjernite/bart_eli5').to('cuda:0') save_dict = torch.load('seq2seq_models/eli5_bart_model_blm_2.pth') s2s_model.load_state_dict(save_dict['model']) _ = s2s_model.eval() else: (s2s_tokenizer, s2s_model) = make_qa_s2s_model(model_name='t5-small', from_file='seq2seq_models/eli5_t5_model_1024_4.pth', device='cuda:0') return (qar_tokenizer, qar_model, s2s_tokenizer, s2s_model)
class DPTImageProcessingTester(unittest.TestCase): def __init__(self, parent, batch_size=7, num_channels=3, image_size=18, min_resolution=30, max_resolution=400, do_resize=True, size=None, do_normalize=True, image_mean=[0.5, 0.5, 0.5], image_std=[0.5, 0.5, 0.5]): size = (size if (size is not None) else {'height': 18, 'width': 18}) self.parent = parent self.batch_size = batch_size self.num_channels = num_channels self.image_size = image_size self.min_resolution = min_resolution self.max_resolution = max_resolution self.do_resize = do_resize self.size = size self.do_normalize = do_normalize self.image_mean = image_mean self.image_std = image_std def prepare_image_processor_dict(self): return {'image_mean': self.image_mean, 'image_std': self.image_std, 'do_normalize': self.do_normalize, 'do_resize': self.do_resize, 'size': self.size}
def adb_devices(): serialnos = [] p = re.compile('(\\S+)\\s+device') for line in split_stdout(sh.adb('devices')): m = p.match(line) if m: serialnos.append(m.group(1)) return serialnos
class PrefixConstrainedLogitsProcessor(metaclass=DummyObject): _backends = ['torch'] def __init__(self, *args, **kwargs): requires_backends(self, ['torch'])
def get_view_a2j_parser(): parser = argparse.ArgumentParser() parser.add_argument('--seed', default=0, type=int) parser.add_argument('--dataset', default='nyu') parser.add_argument('--phase', default='train') parser.add_argument('--split', default=1, type=int, help='Divide the train dataset into s parts, each as an epoch.') parser.add_argument('--batch_size', default=16, type=int) parser.add_argument('--num_epoch', default=20, type=int) parser.add_argument('--gpus', type=int, nargs='+') parser.add_argument('--optim', default='Adam', help='Must be Adam or SGD.') parser.add_argument('--learning_rate', default=0.001, type=float) parser.add_argument('--model_saved_path', default='./checkpoint/base') parser.add_argument('--pre_a2j', default=None) parser.add_argument('--pre_model_path', default=None) parser.add_argument('--log_dir', default='./logs/base') parser.add_argument('--resume_training', default=False, type=bool, help='If resume_training is False, log dir will not be remove.') parser.add_argument('--learning_decay_rate', default=0.8, help='Learning decay rate each epoch.') parser.add_argument('--reg_weight', default=0.001, type=float, help='Regularization weight.') parser.add_argument('--level', default=3, type=int, help='Specify the number of virtual views. Levels 1, 2, 3, 4 have 3, 9, 15 and 25 views, respectively.') parser.add_argument('--n_head', default=1, type=int) parser.add_argument('--d_attn', default=256, type=int) parser.add_argument('--d_k', default=64, type=int) parser.add_argument('--d_v', default=64, type=int) parser.add_argument('--d_inner', default=256, type=int) parser.add_argument('--dropout_rate', default=0.5, type=float) parser.add_argument('--num_select', default=3, type=int) parser.add_argument('--alpha', default=0.5, type=float) parser.add_argument('--conf_factor', default=100.0, type=float) parser.add_argument('--random_select', default=False, type=bool) parser.add_argument('--num_worker', default=4, type=int, help='Number worker of Dataloader.') parser.add_argument('--max_jitter', default=10.0, type=float, help='Sigma of jittering center of mass.') parser.add_argument('--depth_sigma', default=1.0, type=float, help='Sigma of jittering depth of pixel.') parser.add_argument('--cube_len', default=None, type=float) parser.add_argument('--min_scale', default=1.0, type=float) parser.add_argument('--max_scale', default=1.0, type=float) parser.add_argument('--offset', default=20.0, type=float, help='Offset of bounding box.') parser.add_argument('--hand_thickness', default=20.0, type=float) parser.add_argument('--random_flip', default=False, type=bool, help='Whether flip image randomly.') parser.add_argument('--save_result', default=False, type=bool) parser.add_argument('--use_joint', default=False, type=bool) parser.add_argument('--adjust_cube', default=False, type=bool) parser.add_argument('--config', default='./config/nyu/train_a2j.yaml') return parser
_model def nfnet_f7s(pretrained=False, **kwargs): return _create_normfreenet('nfnet_f7s', pretrained=pretrained, **kwargs)
def main(): parser = get_parser() args = parser.parse_args() source_path = osp.join(args.source, args.split) data_poth = ((source_path + '_unfiltered') if args.unfiltered else source_path) print(f'data path: {data_poth}') features = np.load((data_poth + '.npy'), mmap_mode='r') pca_A = torch.from_numpy(np.load((args.pca_path + '_A.npy'))).cuda() pca_b = torch.from_numpy(np.load((args.pca_path + '_b.npy'))).cuda() os.makedirs(args.save_dir, exist_ok=True) save_path = osp.join(args.save_dir, args.split) copyfile((source_path + '.tsv'), (save_path + '.tsv')) copyfile((data_poth + '.lengths'), (save_path + '.lengths')) if osp.exists((source_path + '.phn')): copyfile((source_path + '.phn'), (save_path + '.phn')) if osp.exists((source_path + '.wrd')): copyfile((source_path + '.wrd'), (save_path + '.wrd')) if osp.exists((save_path + '.npy')): os.remove((save_path + '.npy')) npaa = NpyAppendArray((save_path + '.npy')) batches = math.ceil((features.shape[0] / args.batch_size)) with torch.no_grad(): for b in tqdm.trange(batches): start = (b * args.batch_size) end = (start + args.batch_size) x = torch.from_numpy(features[start:end]).cuda() x = (torch.matmul(x, pca_A) + pca_b) npaa.append(x.cpu().numpy())
def write_uchars(fd, values, fmt='>{:d}B'): fd.write(struct.pack(fmt.format(len(values)), *values)) return (len(values) * 1)
def print_stats(args, epoch, num_samples, trainloader, metrics): if ((num_samples % args.log_interval) == 1): print('Epoch:{:2d}\tSample:{:5d}/{:5d}\tLoss:{:.4f}\tAccuracy:{:.2f}\tPPV:{:.3f}\tsensitivity{:.3f}'.format(epoch, num_samples, (len(trainloader) * args.batch_size), metrics.avg('loss'), metrics.avg('accuracy'), metrics.avg('ppv'), metrics.avg('sensitivity')))
class DataLoader(object): def __init__(self, batch_size, seq_len, dataset_name, task_name, data_dir, tokenizer_dir): self.batch_size = batch_size dataset = load_dataset(dataset_name, cache_dir=data_dir, split='validation') tokenizer = AutoTokenizer.from_pretrained(tokenizer_dir) self.dataset = dataset.map((lambda e: tokenizer(e['text'], truncation=True, padding='max_length', max_length=seq_len)), batched=True) def __getitem__(self, idx): start = (idx * self.batch_size) end = (start + self.batch_size) if (end > len(self.dataset)): input_ids_data = self.dataset[start:]['input_ids'] input_mask_data = self.dataset[start:]['attention_mask'] label_data = self.dataset[start:]['label'] else: input_ids_data = self.dataset[start:end]['input_ids'] input_mask_data = self.dataset[start:end]['attention_mask'] label_data = self.dataset[start:end]['label'] sample_size = (len(input_ids_data) if isinstance(input_ids_data, list) else 1) return ([np.array(input_ids_data).reshape(sample_size, (- 1)).astype('int32'), np.array(input_mask_data).reshape(sample_size, (- 1)).astype('int32')], np.array(label_data).reshape(sample_size, (- 1)).astype('int32')) def __len__(self): return math.ceil((len(self.dataset) / self.batch_size))
def get_confusion_matrix(correct_seg, segmentation, elements=None): (height, width) = correct_seg.shape if (elements is None): elements = set(np.unique(correct_seg)) elements = elements.union(set(np.unique(segmentation))) logging.debug('elements parameter not given to get_confusion_matrix') logging.debug(" assume '%s'", elements) confusion_matrix = {} for i in elements: confusion_matrix[i] = {} for j in elements: confusion_matrix[i][j] = 0 for x in range(width): for y in range(height): confusion_matrix[correct_seg[y][x]][segmentation[y][x]] += 1 return confusion_matrix
def get_l32_config(): config = get_l16_config() config.patches.size = (32, 32) return config
_LAYERS.register_module() class ExampleConv(nn.Module): def __init__(self, in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, norm_cfg=None): super(ExampleConv, self).__init__() self.in_channels = in_channels self.out_channels = out_channels self.kernel_size = kernel_size self.stride = stride self.padding = padding self.dilation = dilation self.groups = groups self.bias = bias self.norm_cfg = norm_cfg self.output_padding = (0, 0, 0) self.transposed = False self.conv0 = nn.Conv2d(in_channels, out_channels, kernel_size) self.init_weights() def forward(self, x): x = self.conv0(x) return x def init_weights(self): nn.init.constant_(self.conv0.weight, 0)
def at_hat_loss(model, x, y, optimizer, step_size=0.007, epsilon=0.031, perturb_steps=10, h=3.5, beta=1.0, gamma=1.0, attack='linf-pgd', hr_model=None, label_smoothing=0.1): criterion_ce_smooth = SmoothCrossEntropyLoss(reduction='mean', smoothing=label_smoothing) criterion_ce = nn.CrossEntropyLoss() model.train() track_bn_stats(model, False) attack = create_attack(model, criterion_ce, attack, epsilon, perturb_steps, step_size) with ctx_noparamgrad_and_eval(model): (x_adv, _) = attack.perturb(x, y) model.train() x_hr = (x + (h * (x_adv - x))) with ctx_noparamgrad_and_eval(hr_model): y_hr = hr_model(x_adv).argmax(dim=1) optimizer.zero_grad() track_bn_stats(model, True) (out_clean, out_adv, out_help) = (model(x), model(x_adv), model(x_hr)) loss_clean = criterion_ce_smooth(out_clean, y) loss_adv = criterion_ce(out_adv, y) loss_help = F.cross_entropy(out_help, y_hr, reduction='mean') loss = ((loss_clean + (beta * loss_adv)) + (gamma * loss_help)) batch_metrics = {'loss': loss.item()} batch_metrics.update({'adversarial_acc': accuracy(y, out_adv.detach()), 'helper_acc': accuracy(y_hr, out_help.detach())}) batch_metrics.update({'clean_acc': accuracy(y, out_clean.detach())}) return (loss, batch_metrics)
.mujoco .no_cover .timeout(30) def test_maml_vpg(): assert (subprocess.run([str((EXAMPLES_ROOT_DIR / 'torch/maml_vpg_half_cheetah_dir.py')), '--epochs', '1', '--rollouts_per_task', '1', '--meta_batch_size', '1'], check=False).returncode == 0)
class Linear(torch.nn.Linear): def forward(self, x): if ((x.numel() == 0) and obsolete_torch_version(TORCH_VERSION, (1, 5))): out_shape = [x.shape[0], self.out_features] empty = NewEmptyTensorOp.apply(x, out_shape) if self.training: dummy = (sum((x.view((- 1))[0] for x in self.parameters())) * 0.0) return (empty + dummy) else: return empty return super().forward(x)
class WideBasic(nn.Module): def __init__(self, in_planes, planes, stride=1): super(WideBasic, self).__init__() self.bn1 = nn.BatchNorm2d(in_planes) self.dropout = nn.Dropout(p=P) self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, padding=1, bias=True) self.bn2 = nn.BatchNorm2d(planes) self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride, padding=1, bias=True) self.shortcut = nn.Sequential() if ((stride != 1) or (in_planes != planes)): self.shortcut = nn.Sequential(nn.Conv2d(in_planes, planes, kernel_size=1, stride=stride, bias=True)) def forward(self, x): out = self.dropout(self.conv1(F.relu(self.bn1(x)))) out = self.conv2(F.relu(self.bn2(out))) out += self.shortcut(x) return out
def circle_path(obs_all, radius, k): global tracked global direction global old_distance if (k == 1): tracked = False direction = (- 1) actions = np.array([[[]]]) for obs_env in obs_all: for obs in obs_env: agent_pos = np.matrix([obs[2], obs[3]]).T agent_ang = np.arctan2(obs.item(1), obs.item(0)) if (agent_ang < 0.0): agent_ang += (2 * np.pi) landmark_pos = np.matrix([obs[4], obs[5]]).T angle_agent_landmark = np.arctan2(landmark_pos.item(1), landmark_pos.item(0)) if (angle_agent_landmark < 0.0): angle_agent_landmark += (2 * np.pi) distance = np.sqrt(((landmark_pos.item(0) ** 2) + (landmark_pos.item(1) ** 2))) alpha_radius = (64.181 * (radius ** (- 0.872))) radius_circ = (radius / 1000.0) if ((distance > (radius_circ * 2)) or (k < 4)): const = 0.5 if (((agent_ang - angle_agent_landmark) > np.pi) and (agent_ang > angle_agent_landmark)): direction = 1.0 angle = direction elif (((agent_ang - angle_agent_landmark) < np.pi) and (agent_ang > angle_agent_landmark)): direction = (- 1.0) if ((agent_ang - angle_agent_landmark) < ((np.pi / 2.0) / 2.0)): angle = 0.0 else: angle = direction elif (((angle_agent_landmark - agent_ang) > np.pi) and (agent_ang < angle_agent_landmark)): direction = (- 1.0) angle = direction elif (((angle_agent_landmark - agent_ang) < np.pi) and (agent_ang < angle_agent_landmark)): direction = 1.0 if ((angle_agent_landmark - agent_ang) < ((np.pi / 2.0) / 2.0)): angle = 0.0 else: angle = direction else: angle = 0.0 else: aux = abs((agent_ang - angle_agent_landmark)) vector_1 = np.array([obs.item(1), obs.item(0)]) vector_2 = np.array([landmark_pos.item(1), landmark_pos.item(0)]) unit_vector_1 = (vector_1 / np.linalg.norm(vector_1)) unit_vector_2 = (vector_2 / np.linalg.norm(vector_2)) dot_product = np.dot(unit_vector_1, unit_vector_2) aux = np.arccos(dot_product) angle = ((0.4 * (aux + 0.7)) * direction) angle -= (1 - alpha_radius) if (k < 2): angle = (- 1.0) return np.array([[[angle]]])
def evaluate(features, support, labels, mask, placeholders): t_test = time.time() feed_dict_val = construct_feed_dict(features, support, labels, mask, placeholders) outs_val = sess.run([model.loss, model.accuracy], feed_dict=feed_dict_val) return (outs_val[0], outs_val[1], (time.time() - t_test))
def frame_generator(frame_duration_ms, audio, sample_rate): n = int(((sample_rate * (frame_duration_ms / 1000.0)) * 2)) offset = 0 timestamp = 0.0 duration = ((float(n) / sample_rate) / 2.0) while ((offset + n) < len(audio)): (yield Frame(audio[offset:(offset + n)], timestamp, duration)) timestamp += duration offset += n
def _run_process(children, report_queue, global_t, is_stopped, is_paused, env_fn): try: if ((not hasattr(children, 'create_env')) or (children.create_env is None)): children.create_env = env_fn(children) if hasattr(children, 'initialize'): children.initialize() def _process(process, *args, **kwargs): result = process(*args, **kwargs) report_queue.put(result) return result def _run(process, **kwargs): children._global_t = 0 children._is_stopped = False while (not children._is_stopped): children._global_t = global_t.value (tdiff, _, _) = process(mode='train', context=dict()) children._global_t += tdiff if is_paused.value: report_queue.put('paused') while (is_paused.value and (not is_stopped.value)): sleep(0.005) children._global_t = global_t.value children._is_stopped = is_stopped.value return None return _run(process=partial(_process, process=children.process)) except Exception as e: report_queue.put('error') raise e
class IfTimeout(): def __init__(self, timeout): self.start_time = time.time() self.ignored_time = 0.0 if (timeout is None): self.target_time = None else: self.target_time = (self.start_time + timeout) self.interval = None def is_timeout(self): if (self.target_time is None): return False else: cur_time = time.time() if (((cur_time - self.target_time) - self.ignored_time) > 0): if (self.interval is None): self.interval = ((cur_time - self.start_time) - self.ignored_time) return True else: return False def add_ignored_time(self, time_amount): self.ignored_time += time_amount
def convert_examples_to_features(examples, label2id, max_seq_length, tokenizer, special_tokens, mode='text'): def get_special_token(w): if (w not in special_tokens): special_tokens[w] = ('[unused%d]' % (len(special_tokens) + 1)) return special_tokens[w] num_tokens = 0 num_fit_examples = 0 num_shown_examples = 0 features = [] for (ex_index, example) in enumerate(examples): if ((ex_index % 10000) == 0): logger.info(('Writing example %d of %d' % (ex_index, len(examples)))) tokens = [CLS] SUBJECT_START = get_special_token('SUBJ_START') SUBJECT_END = get_special_token('SUBJ_END') OBJECT_START = get_special_token('OBJ_START') OBJECT_END = get_special_token('OBJ_END') SUBJECT_NER = get_special_token(('SUBJ=%s' % example.ner1)) OBJECT_NER = get_special_token(('OBJ=%s' % example.ner2)) if mode.startswith('text'): for (i, token) in enumerate(example.sentence): if (i == example.span1[0]): tokens.append(SUBJECT_START) if (i == example.span2[0]): tokens.append(OBJECT_START) for sub_token in tokenizer.tokenize(token): tokens.append(sub_token) if (i == example.span1[1]): tokens.append(SUBJECT_END) if (i == example.span2[1]): tokens.append(OBJECT_END) if (mode == 'text_ner'): tokens = (tokens + [SEP, SUBJECT_NER, SEP, OBJECT_NER, SEP]) else: tokens.append(SEP) else: subj_tokens = [] obj_tokens = [] for (i, token) in enumerate(example.sentence): if (i == example.span1[0]): tokens.append(SUBJECT_NER) if (i == example.span2[0]): tokens.append(OBJECT_NER) if ((i >= example.span1[0]) and (i <= example.span1[1])): for sub_token in tokenizer.tokenize(token): subj_tokens.append(sub_token) elif ((i >= example.span2[0]) and (i <= example.span2[1])): for sub_token in tokenizer.tokenize(token): obj_tokens.append(sub_token) else: for sub_token in tokenizer.tokenize(token): tokens.append(sub_token) if (mode == 'ner_text'): tokens.append(SEP) for sub_token in subj_tokens: tokens.append(sub_token) tokens.append(SEP) for sub_token in obj_tokens: tokens.append(sub_token) tokens.append(SEP) num_tokens += len(tokens) if (len(tokens) > max_seq_length): tokens = tokens[:max_seq_length] else: num_fit_examples += 1 segment_ids = ([0] * len(tokens)) input_ids = tokenizer.convert_tokens_to_ids(tokens) input_mask = ([1] * len(input_ids)) padding = ([0] * (max_seq_length - len(input_ids))) input_ids += padding input_mask += padding segment_ids += padding label_id = label2id[example.label] assert (len(input_ids) == max_seq_length) assert (len(input_mask) == max_seq_length) assert (len(segment_ids) == max_seq_length) if (num_shown_examples < 20): if ((ex_index < 5) or (label_id > 0)): num_shown_examples += 1 logger.info('*** Example ***') logger.info(('guid: %s' % example.guid)) logger.info(('tokens: %s' % ' '.join([str(x) for x in tokens]))) logger.info(('input_ids: %s' % ' '.join([str(x) for x in input_ids]))) logger.info(('input_mask: %s' % ' '.join([str(x) for x in input_mask]))) logger.info(('segment_ids: %s' % ' '.join([str(x) for x in segment_ids]))) logger.info(('label: %s (id = %d)' % (example.label, label_id))) features.append(InputFeatures(input_ids=input_ids, input_mask=input_mask, segment_ids=segment_ids, label_id=label_id)) logger.info(('Average #tokens: %.2f' % ((num_tokens * 1.0) / len(examples)))) logger.info(('%d (%.2f %%) examples can fit max_seq_length = %d' % (num_fit_examples, ((num_fit_examples * 100.0) / len(examples)), max_seq_length))) return features
class ModelArguments(): save_adapter_weights: bool = field(default=True, metadata={'help': 'Save the weights for the task-specific adapter.'}) load_adapter_weights: bool = field(default=False, metadata={'help': 'Load the weights used to task-sepcific adapters.'}) adapter_dir: str = field(default=None, metadata={'help': 'Path to load task-specific adapters'}) model_name_or_path: str = field(default=None, metadata={'help': 'Path to pretrained model or model identifier from huggingface.co/models'}) config_name: Optional[str] = field(default=None, metadata={'help': 'Pretrained config name or path if not the same as model_name'}) tokenizer_name: Optional[str] = field(default=None, metadata={'help': 'Pretrained tokenizer name or path if not the same as model_name'}) cache_dir: Optional[str] = field(default=None, metadata={'help': 'Where to store the pretrained models downloaded from huggingface.co'}) use_fast_tokenizer: bool = field(default=True, metadata={'help': 'Whether to use one of the fast tokenizer (backed by the tokenizers library) or not.'}) model_revision: str = field(default='main', metadata={'help': 'The specific model version to use (can be a branch name, tag name or commit id).'}) use_auth_token: bool = field(default=False, metadata={'help': 'Will use the token generated when running `transformers-cli login` (necessary to use this script with private models).'}) load_prefix_embeddings: bool = field(default=False, metadata={'help': 'load prefix embeddings or not'}) save_prefix_only: bool = field(default=False, metadata={'help': 'save prefix embeddings only'}) prompt_embedding_path: Optional[List[str]] = field(default=None, metadata={'help': 'A list of the paths to prefix embeddings'}) target_prompt_embedding_path: Optional[str] = field(default=None, metadata={'help': 'a path to the target prompt embedding'}) attn_prefix_tuning: bool = field(default=False, metadata={'help': 'Set true if you try ATTEMPT.'}) attn_method: Optional[str] = field(default='sub', metadata={'help': 'Attention model for attn_prefix. We currently support the following methods: linear, sub (our main method), and constant (gives the constant and equal weights to all of the prompts.)'}) shared_attn: bool = field(default=False, metadata={'help': 'shared attention'}) load_attention: bool = field(default=False, metadata={'help': 'Set true if you want to load pre-trained attention weights'}) attn_path: Optional[str] = field(default=None, metadata={'help': 'path to attention weights (linear attentions). '}) attn_path_sub: Optional[List[str]] = field(default=None, metadata={'help': 'list of the path to attention weights (sub attentions). [path_to_down_projection_weights, path_to_up_projection_weights]'}) ignore_target: bool = field(default=False, metadata={'help': 'Whether to ignore the new target tokens. Mainly for ablation.'}) fix_attention: bool = field(default=False, metadata={'help': 'this will make the attention weights frozen during training. Mainly for ablation.'}) temperature: float = field(default=2000, metadata={'help': 'set the soft max temperature of ATTEMPT.'}) attn_learning_rate: float = field(default=None, metadata={'help': 'set the learning rate for the attention modules.'}) load_layer_norm: bool = field(default=False, metadata={'help': 'Set true if you want to load pre-trained layer-norm weight and biases.'}) layer_norm_dir: Optional[List[str]] = field(default=None, metadata={'help': 'Layer norm dir. [path_to_layer_norm_weight.pt, path_to_layer_norm_bias.pt]'}) prefix_num: Optional[int] = field(default=1, metadata={'help': 'the number of prefix'})
class DIV2K(multiscalesrdata.SRData): def __init__(self, args, name='DIV2K', train=True, benchmark=False): super(DIV2K, self).__init__(args, name=name, train=train, benchmark=benchmark) def _scan(self): (names_hr, names_lr) = super(DIV2K, self)._scan() names_hr = names_hr[(self.begin - 1):self.end] names_lr = [n[(self.begin - 1):self.end] for n in names_lr] return (names_hr, names_lr) def _set_filesystem(self, dir_data): super(DIV2K, self)._set_filesystem(dir_data) self.dir_hr = os.path.join(self.apath, 'HR') self.dir_lr = os.path.join(self.apath, 'LR_bicubic')
class NCBIDataset(BaseDataset): def __init__(self, dataset): super().__init__(dataset=dataset)
class UnrolledAdam(UnrolledOptimizer): 'The Adam optimizer defined in _State = collections.namedtuple('State', ['t', 'm', 'u']) def __init__(self, lr=0.1, lr_fn=None, beta1=0.9, beta2=0.999, epsilon=1e-09, colocate_gradients_with_ops=False): super(UnrolledAdam, self).__init__(colocate_gradients_with_ops=colocate_gradients_with_ops) self._lr = lr self._lr_fn = ((lambda l, i: l) if (lr_fn is None) else lr_fn) self._beta1 = beta1 self._beta2 = beta2 self._epsilon = epsilon def init_state(self, x): return self._State(t=tf.constant(0, dtype=tf.float32), m=[tf.zeros_like(v) for v in x], u=[tf.zeros_like(v) for v in x]) def _apply_gradients(self, grads, x, optim_state): lr = self._lr_fn(self._lr, optim_state.t) new_optim_state = self._State(t=(optim_state.t + 1.0), m=([None] * len(x)), u=([None] * len(x))) t = new_optim_state.t new_x = ([None] * len(x)) for i in range(len(x)): g = grads[i] m_old = optim_state.m[i] u_old = optim_state.u[i] new_optim_state.m[i] = ((self._beta1 * m_old) + ((1.0 - self._beta1) * g)) new_optim_state.u[i] = ((self._beta2 * u_old) + (((1.0 - self._beta2) * g) * g)) m_hat = (new_optim_state.m[i] / (1.0 - tf.pow(self._beta1, t))) u_hat = (new_optim_state.u[i] / (1.0 - tf.pow(self._beta2, t))) new_x[i] = (x[i] - ((lr * m_hat) / (tf.sqrt(u_hat) + self._epsilon))) return (new_x, new_optim_state) def minimize(self, loss, x, optim_state): grads = tf.gradients(loss, x, colocate_gradients_with_ops=self._colocate_gradients_with_ops) return self._apply_gradients(grads, x, optim_state)
_methods class Model(HPOMixin, tf.keras.Model): def __init__(self, **kwargs): super().__init__() self.model_class = tf.keras.Model self.kwargs = kwargs self.lazyinputs_ = kwargs.get('inputs', None) self.lazyoutputs_ = kwargs.get('outputs', None) def _model_init_args(self, trial): (in_tensors, out_tensors) = CallCache.execute(self.lazyinputs_, self.lazyoutputs_, trial, self.backend) self.kwargs['inputs'] = in_tensors self.kwargs['outputs'] = out_tensors return self.kwargs def _get_model_init_args_func_kwargs(self): return {'lazyinputs': self.lazyinputs_, 'lazyoutputs': self.lazyoutputs_, 'kwargs': self.kwargs, 'backend': self.backend} def _model_init_args_func(trial, lazyinputs, lazyoutputs, kwargs, backend): (in_tensors, out_tensors) = CallCache.execute(lazyinputs, lazyoutputs, trial, backend) kwargs['inputs'] = in_tensors kwargs['outputs'] = out_tensors return kwargs
def _get_nerf_inner(hparams: Namespace, appearance_count: int, layer_dim: int, xyz_dim: int, weight_key: str) -> nn.Module: if (hparams.container_path is not None): container = torch.jit.load(hparams.container_path, map_location='cpu') if (xyz_dim == 3): return MegaNeRF([getattr(container, 'sub_module_{}'.format(i)) for i in range(len(container.centroids))], container.centroids, hparams.boundary_margin, False, container.cluster_2d) else: return MegaNeRF([getattr(container, 'bg_sub_module_{}'.format(i)) for i in range(len(container.centroids))], container.centroids, hparams.boundary_margin, True, container.cluster_2d) elif hparams.use_cascade: nerf = Cascade(_get_single_nerf_inner(hparams, appearance_count, (layer_dim if (xyz_dim == 4) else layer_dim), xyz_dim), _get_single_nerf_inner(hparams, appearance_count, layer_dim, xyz_dim)) elif (hparams.train_mega_nerf is not None): centroid_metadata = torch.load(hparams.train_mega_nerf, map_location='cpu') centroids = centroid_metadata['centroids'] nerf = MegaNeRF([_get_single_nerf_inner(hparams, appearance_count, layer_dim, xyz_dim) for _ in range(len(centroids))], centroids, 1, (xyz_dim == 4), centroid_metadata['cluster_2d'], True) else: nerf = _get_single_nerf_inner(hparams, appearance_count, layer_dim, xyz_dim) if (hparams.ckpt_path is not None): state_dict = torch.load(hparams.ckpt_path, map_location='cpu')[weight_key] consume_prefix_in_state_dict_if_present(state_dict, prefix='module.') model_dict = nerf.state_dict() model_dict.update(state_dict) nerf.load_state_dict(model_dict) return nerf
def make_cosine_lr_schedule(init_lr, total_steps): def schedule(step): t = (step / total_steps) return ((0.5 * init_lr) * (1 + jnp.cos((t * onp.pi)))) return schedule
def reduce_tensor(tensor, n_gpus): rt = tensor.clone() dist.all_reduce(rt, op=dist.reduce_op.SUM) rt /= n_gpus return rt
def eval_nmi(pred, gold): from sklearn.metrics import normalized_mutual_info_score res = dict() res['nmi'] = normalized_mutual_info_score(pred, gold) return res
def get_chamfer_average(model_id, pre_sampled=True, cat_desc=None, **kwargs): import os from shapenet.core import cat_desc_to_id manager = get_chamfer_manager(model_id, pre_sampled, **kwargs) values = None if os.path.isfile(manager.path): with manager.get_saving_dataset('r') as ds: values = np.array(tuple(ds.values())) if ((values is None) or (len(values) == 0)): manager.save_all() with manager.get_saving_dataset('r') as ds: if (cat_desc is not None): if (not isinstance(cat_desc, (list, tuple, set))): cat_desc = [cat_desc] cat_id = set(cat_desc_to_id(cat_desc)) ds = ds.filter_keys((lambda key: (key[0] in cat_id))) values = np.array(tuple(ds.values())) return np.mean(values)
class WideResNet(nn.Module): def __init__(self, conv_layer, linear_layer, depth=34, num_classes=10, widen_factor=10, dropRate=0.0): super(WideResNet, self).__init__() nChannels = [16, (16 * widen_factor), (32 * widen_factor), (64 * widen_factor)] assert (((depth - 4) % 6) == 0) n = ((depth - 4) / 6) block = BasicBlock self.conv1 = conv_layer(3, nChannels[0], kernel_size=3, stride=1, padding=1, bias=False) self.block1 = NetworkBlock(n, nChannels[0], nChannels[1], block, conv_layer, 1, dropRate) self.sub_block1 = NetworkBlock(n, nChannels[0], nChannels[1], block, conv_layer, 1, dropRate) self.block2 = NetworkBlock(n, nChannels[1], nChannels[2], block, conv_layer, 2, dropRate) self.block3 = NetworkBlock(n, nChannels[2], nChannels[3], block, conv_layer, 2, dropRate) self.bn1 = nn.BatchNorm2d(nChannels[3]) self.relu = nn.ReLU(inplace=True) self.fc = linear_layer(nChannels[3], num_classes) self.nChannels = nChannels[3] for m in self.modules(): if isinstance(m, nn.Conv2d): n = ((m.kernel_size[0] * m.kernel_size[1]) * m.out_channels) m.weight.data.normal_(0, math.sqrt((2.0 / n))) elif isinstance(m, nn.BatchNorm2d): m.weight.data.fill_(1) m.bias.data.zero_() elif isinstance(m, linear_layer): m.bias.data.zero_() def forward(self, x): out = self.conv1(x) out = self.block1(out) out = self.block2(out) out = self.block3(out) out = self.relu(self.bn1(out)) out = F.avg_pool2d(out, 8) out = out.view((- 1), self.nChannels) return self.fc(out)
def DistributedFairseqModel(args, model): assert isinstance(model, BaseFairseqModel) if (args.ddp_backend == 'c10d'): ddp_class = parallel.DistributedDataParallel init_kwargs = dict(module=model, device_ids=[args.device_id], output_device=args.device_id, broadcast_buffers=False, bucket_cap_mb=args.bucket_cap_mb) if ('check_reduction' in inspect.getargspec(ddp_class)[0]): init_kwargs['check_reduction'] = True if ('find_unused_parameters' in inspect.getargspec(ddp_class)[0]): init_kwargs['find_unused_parameters'] = args.find_unused_parameters elif (args.ddp_backend == 'no_c10d'): ddp_class = LegacyDistributedDataParallel init_kwargs = dict(module=model, world_size=args.distributed_world_size, buffer_size=(2 ** 28)) else: raise ValueError(('Unknown --ddp-backend: ' + args.ddp_backend)) class _DistributedFairseqModel(ddp_class): def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) def __getattr__(self, name): wrapped_module = super().__getattr__('module') if hasattr(wrapped_module, name): return getattr(wrapped_module, name) return super().__getattr__(name) return _DistributedFairseqModel(**init_kwargs)
def make_position_amplitude_gaussian_proposal(model_apply: ModelApply[P], get_std_move: Callable[([PositionAmplitudeData], chex.Scalar)]) -> Callable[([P, PositionAmplitudeData, PRNGKey], Tuple[(PositionAmplitudeData, PRNGKey)])]: def proposal_fn(params: P, data: PositionAmplitudeData, key: PRNGKey): std_move = get_std_move(data) (proposed_position, key) = metropolis.gaussian_proposal(data['walker_data']['position'], std_move, key) proposed_amplitude = model_apply(params, proposed_position) return (make_position_amplitude_data(proposed_position, proposed_amplitude, data['move_metadata']), key) return proposal_fn
def data_to_extract(username, args): labels = {} labels['title'] = 'PPI Link Prediction' labels['x_label'] = 'Iterations' labels['y_label'] = 'Percent' if args.local: param_str = 'Local' else: param_str = 'Global' labels['train_metric_auc'] = (('Train_' + param_str) + '_Graph_') labels['train_metric_ap'] = (('Train_' + param_str) + '_Graph_') labels['test_metric_auc'] = (('Test_' + param_str) + '_Graph_') labels['test_metric_ap'] = (('Test_' + param_str) + '_Graph_') if (username == 'joeybose'): labels['experiments_key'] = [[args.one_maml], [args.two_maml], [args.random_exp], [args.no_finetune], [args.finetune], [args.adamic_adar]] if args.local: labels['experiments_name'] = ['1-MAML', '2-MAML', 'NoFinetune', 'Finetune', 'Adamic-Adar'] else: labels['experiments_name'] = ['1-MAML', '2-MAML', 'Random', 'NoFinetune', 'Finetune'] return labels
def shufflenet_v2_mpncov_x2_0(pretrained=False, progress=True, **kwargs): return _shufflenetv2_mpncov('shufflenetv2_mpncov_x2.0', pretrained, progress, [4, 8, 4], [24, 244, 488, 976, 2048], **kwargs)
def season_game_logs(season): GH_TOKEN = os.getenv('GH_TOKEN', '') g = Github(GH_TOKEN) repo = g.get_repo('chadwickbureau/retrosheet') gamelogs = [f.path[(f.path.rfind('/') + 1):] for f in repo.get_contents('gamelog')] file_name = f'GL{season}.TXT' if (file_name not in gamelogs): raise ValueError(f'Season game logs not available for {season}') s = get_text_file(gamelog_url.format(season)) data = pd.read_csv(StringIO(s), header=None, sep=',', quotechar='"') data.columns = gamelog_columns return data
def find_average_velocity(df): df['vbar'] = three_comp_average(df['vxbar'], df['vybar'], df['vzbar']) return df
def validate(model, loader, loss_fn, args, amp_autocast=suppress, log_suffix=''): batch_time_m = AverageMeter() losses_m = AverageMeter() top1_m = AverageMeter() top5_m = AverageMeter() model.eval() end = time.time() last_idx = (len(loader) - 1) with torch.no_grad(): for (batch_idx, (input, target)) in enumerate(loader): last_batch = (batch_idx == last_idx) if (not args.prefetcher): input = input.cuda() target = target.cuda() if args.channels_last: input = input.contiguous(memory_format=torch.channels_last) with amp_autocast(): output = model(input) if isinstance(output, (tuple, list)): output = output[0] reduce_factor = args.tta if (reduce_factor > 1): output = output.unfold(0, reduce_factor, reduce_factor).mean(dim=2) target = target[0:target.size(0):reduce_factor] loss = loss_fn(output, target) (acc1, acc5) = accuracy(output, target, topk=(1, 5)) if args.distributed: reduced_loss = reduce_tensor(loss.data, args.world_size) acc1 = reduce_tensor(acc1, args.world_size) acc5 = reduce_tensor(acc5, args.world_size) else: reduced_loss = loss.data torch.cuda.synchronize() losses_m.update(reduced_loss.item(), input.size(0)) top1_m.update(acc1.item(), output.size(0)) top5_m.update(acc5.item(), output.size(0)) batch_time_m.update((time.time() - end)) end = time.time() if ((args.local_rank == 0) and (last_batch or ((batch_idx % args.log_interval) == 0))): log_name = ('Test' + log_suffix) _logger.info('{0}: [{1:>4d}/{2}] Time: {batch_time.val:.3f} ({batch_time.avg:.3f}) Loss: {loss.val:>7.4f} ({loss.avg:>6.4f}) : {top1.val:>7.4f} ({top1.avg:>7.4f}) : {top5.val:>7.4f} ({top5.avg:>7.4f})'.format(log_name, batch_idx, last_idx, batch_time=batch_time_m, loss=losses_m, top1=top1_m, top5=top5_m)) metrics = OrderedDict([('loss', losses_m.avg), ('top1', top1_m.avg), ('top5', top5_m.avg)]) return metrics
class CustomJSONEncoder(JSONEncoder): def default(self, obj): try: if isinstance(obj, datetime): return obj.strftime('%Y-%m-%d %H:%M:%S') elif isinstance(obj, date): return obj.strftime('%Y-%m-%d') iterable = iter(obj) except TypeError: pass else: return list(iterable) return JSONEncoder.default(self, obj)
class BaseEncoder(nn.Module): def __init__(self): super(BaseEncoder, self).__init__() def forward(self, inputs, inputs_mask, **kargs): raise NotImplementedError def inference(self, inputs, inputs_mask, cache=None, **kargs): raise NotImplementedError
class IBNConvBlock(nn.Module): def __init__(self, in_channels, out_channels, kernel_size, stride, padding, dilation=1, groups=1, bias=False, use_ibn=False, activate=True): super(IBNConvBlock, self).__init__() self.activate = activate self.use_ibn = use_ibn self.conv = nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=kernel_size, stride=stride, padding=padding, dilation=dilation, groups=groups, bias=bias) if self.use_ibn: self.ibn = IBN(channels=out_channels) else: self.bn = nn.BatchNorm2d(num_features=out_channels) if self.activate: self.activ = nn.ReLU(inplace=True) def forward(self, x): x = self.conv(x) if self.use_ibn: x = self.ibn(x) else: x = self.bn(x) if self.activate: x = self.activ(x) return x
def _renorm(x, dim=0, inplace=False, eps=1e-12): if (not inplace): x = x.clone() return x.div_(_norm_exclude_dim(x, dim, keepdim=True))
def plot_acc(acc, val_acc, epochs, val_epochs, save_path, plot_name): plt.figure() plt.plot(epochs, acc, label='training') plt.plot(val_epochs, val_acc, label='validation') plt.title('Training and validation acc') plt.ylabel('acc') plt.xlabel('epoch') plt.xlim(min(epochs), max(epochs)) plt.ylim(0, 100) plt.legend(loc='upper right') plt.savefig((save_path + '{}.png'.format(plot_name))) plt.show()
def data_type_dict(): return {'float16': torch.float16, 'float32': torch.float32, 'float64': torch.float64, 'uint8': torch.uint8, 'int8': torch.int8, 'int16': torch.int16, 'int32': torch.int32, 'int64': torch.int64, 'bool': torch.bool}
class KNNSearch(torch.nn.Module): def __init__(self, metric='L2', ignore_query_point=False, return_distances=False, index_dtype=torch.int32, **kwargs): self.metric = metric self.ignore_query_point = ignore_query_point self.return_distances = return_distances assert (index_dtype in [torch.int32, torch.int64]) self.index_dtype = index_dtype super().__init__() def forward(self, points, queries, k, points_row_splits=None, queries_row_splits=None): if (points_row_splits is None): points_row_splits = torch.LongTensor([0, points.shape[0]]) if (queries_row_splits is None): queries_row_splits = torch.LongTensor([0, queries.shape[0]]) result = ops.knn_search(ignore_query_point=self.ignore_query_point, return_distances=self.return_distances, metric=self.metric, points=points, queries=queries, k=k, points_row_splits=points_row_splits, queries_row_splits=queries_row_splits, index_dtype=self.index_dtype) return result
def pyramidnet110_a270_cifar100(num_classes=100, **kwargs): return get_pyramidnet_cifar(num_classes=num_classes, blocks=110, alpha=270, bottleneck=False, model_name='pyramidnet110_a270_cifar100', **kwargs)
class DebertaTokenizeTransform(TokenizeTransform): def __init__(self, train): super().__init__(tokenizer=DebertaV2Tokenizer.from_pretrained('microsoft/deberta-v3-base')) del train
def make_mirror(src_module, dst_module): def setattr_recursive(mod, key, value): (key, *next_key) = key.split('.', maxsplit=1) if (not next_key): setattr(mod, key, value) else: setattr_recursive(getattr(mod, key), next_key[0], value) for (name, param) in src_module.named_parameters(): setattr_recursive(dst_module, name, param)
def _add_file_handler(logger, path, level='INFO'): for h in logger.handlers: if isinstance(h, logging.FileHandler): if (os.path.abspath(path) == h.baseFilename): return if os.path.exists(path): assert os.path.isfile(path) warnings.warn('log already exists in {}'.format(path)) dirname = os.path.abspath(os.path.dirname(path)) os.makedirs(dirname, exist_ok=True) file_handler = logging.FileHandler(path, mode='a') file_handler.setLevel(_get_level(level)) file_formatter = logging.Formatter(fmt='%(asctime)s - %(module)s - [%(levelname)s] - %(message)s', datefmt='%Y/%m/%d %H:%M:%S') file_handler.setFormatter(file_formatter) logger.addHandler(file_handler)
class SampleCountingLoader(): def __init__(self, loader): self.loader = loader def __iter__(self): it = iter(self.loader) storage = get_event_storage() while True: try: batch = next(it) num_inst_per_dataset = {} for data in batch: dataset_name = data['dataset'] if (dataset_name not in num_inst_per_dataset): num_inst_per_dataset[dataset_name] = 0 num_inst = len(data['instances']) num_inst_per_dataset[dataset_name] += num_inst for dataset_name in num_inst_per_dataset: storage.put_scalar(f'batch/{dataset_name}', num_inst_per_dataset[dataset_name]) (yield batch) except StopIteration: break
def test_run_emits_events_if_successful(run): run() observer = run.observers[0] assert observer.started_event.called assert observer.heartbeat_event.called assert observer.completed_event.called assert (not observer.interrupted_event.called) assert (not observer.failed_event.called)
def torch_nn_conv1d(self, input): l_in = input.shape[(- 1)] shape = None padding = self.padding if (padding == 'valid'): padding = (0, 0) if (padding == 'same'): shape = list(input.shape) if (shape is None): shape = list(input.shape) l_out = math.floor((((((l_in + (2 * padding[0])) - (self.dilation[0] * (self.kernel_size[0] - 1))) - 1) / self.stride[0]) + 1)) shape[(- 1)] = l_out shape[(- 2)] = self.out_channels return torch.empty(shape, device=_DEVICE)
def create_model(model_type: str, deterministic: bool, enc_blocks: int, conv_type: str, dataset_type: str, decoder: str, r: float, temperature: float, *args: Any, **kwargs: Any): if (dataset_type in ['mnist', 'bdp', 'pbt']): if (model_type == 'euclidean'): return FeedForwardVAE(*args, **kwargs) else: return FeedForwardHyperboloidVAE(*args, **kwargs) elif (dataset_type in ['cora', 'pubmed', 'ppi', 'disease_lp', 'lobster', 'grid', 'prufer', 'csphd', 'phylo', 'diseases', 'wordnet-noun', 'wordnet-mammal']): if (model_type == 'euclidean'): return VGAE(Encoder(deterministic, enc_blocks, conv_type, *args, **kwargs), decoder, r, temperature) else: return HyperboloidVGAE(HyperboloidEncoder(deterministic, enc_blocks, conv_type, *args, **kwargs), decoder, r, temperature) else: raise ValueError(f"Unknown dataset type: '{dataset_type}'.")
def get_input_data(input_file, seq_length, batch_size, num_labels): def parser(record): name_to_features = {'input_ids': tf.FixedLenFeature([seq_length], tf.int64), 'input_mask': tf.FixedLenFeature([seq_length], tf.int64), 'segment_ids': tf.FixedLenFeature([seq_length], tf.int64), 'label_ids': tf.FixedLenFeature([num_labels], tf.int64)} example = tf.parse_single_example(record, features=name_to_features) input_ids = example['input_ids'] input_mask = example['input_mask'] segment_ids = example['segment_ids'] labels = example['label_ids'] return (input_ids, input_mask, segment_ids, labels) dataset = tf.data.TFRecordDataset(input_file) dataset = dataset.map(parser).repeat().batch(batch_size).shuffle(buffer_size=1000) iterator = dataset.make_one_shot_iterator() (input_ids, input_mask, segment_ids, labels) = iterator.get_next() return (input_ids, input_mask, segment_ids, labels)
class SpatialEncoder(nn.Module): def __init__(self, backbone='resnet18', pretrained=True, num_layers=3, index_interp='bilinear', index_padding='zeros', upsample_interp='bilinear', feature_scale=1.0, use_first_pool=True, norm_type='batch'): super().__init__() if (norm_type != 'batch'): assert (not pretrained) self.use_custom_resnet = (backbone == 'custom') self.feature_scale = feature_scale self.use_first_pool = use_first_pool norm_layer = get_norm_layer(norm_type) pretrained = cfg.pretrained print('Using torchvision', backbone, 'encoder') print(('Pretrained: ' + str(pretrained))) self.model = getattr(torchvision.models, backbone)(pretrained=pretrained, norm_layer=norm_layer) self.model.fc = nn.Sequential() self.model.avgpool = nn.Sequential() self.latent_size = [0, 64, 128, 256, 512, 1024][num_layers] if (cfg.img_feat_size == 128): num_layers = 2 elif (cfg.img_feat_size == 256): num_layers = 3 elif (cfg.img_feat_size == 512): num_layers = 4 self.reduction_layer = nn.Conv2d(cfg.img_feat_size, cfg.embed_size, 1) self.num_layers = num_layers self.index_interp = index_interp self.index_padding = index_padding self.upsample_interp = upsample_interp def forward(self, x): if (self.feature_scale != 1.0): x = F.interpolate(x, scale_factor=self.feature_scale, mode=('bilinear' if (self.feature_scale > 1.0) else 'area'), align_corners=(True if (self.feature_scale > 1.0) else None), recompute_scale_factor=True) pixel_feat_map = None holder_feat_map = None if self.use_custom_resnet: self.latent = self.model(x) else: x = self.model.conv1(x) x = self.model.bn1(x) x = self.model.relu(x) latents = [x] if (self.num_layers > 1): if self.use_first_pool: x = self.model.maxpool(x) x = self.model.layer1(x) latents.append(x) if (self.num_layers > 2): x = self.model.layer2(x) latents.append(x) if (self.num_layers > 3): x = self.model.layer3(x) latents.append(x) if (self.num_layers > 4): x = self.model.layer4(x) latents.append(x) align_corners = (None if (self.index_interp == 'nearest ') else True) latent_sz = latents[0].shape[(- 2):] for i in range(len(latents)): latents[i] = F.interpolate(latents[i], latent_sz, mode=self.upsample_interp, align_corners=align_corners) pixel_feat_map = torch.cat(latents, dim=1) pixel_feat_scale = np.array([pixel_feat_map.shape[(- 1)], pixel_feat_map.shape[(- 2)]]) pixel_feat_scale = ((pixel_feat_scale / (pixel_feat_scale - 1)) * 2.0) holder_feat_map = self.reduction_layer(pixel_feat_map) holder_feat_scale = np.array([holder_feat_map.shape[(- 1)], holder_feat_map.shape[(- 2)]]) holder_feat_scale = ((holder_feat_scale / (holder_feat_scale - 1)) * 2.0) return (holder_feat_map, holder_feat_scale, pixel_feat_map, pixel_feat_scale) def from_conf(cls, conf): return cls(conf.get_string('backbone'), pretrained=conf.get_bool('pretrained', True), num_layers=conf.get_int('num_layers', 4), index_interp=conf.get_string('index_interp', 'bilinear'), index_padding=conf.get_string('index_padding', 'border'), upsample_interp=conf.get_string('upsample_interp', 'bilinear'), feature_scale=conf.get_float('feature_scale', 1.0), use_first_pool=conf.get_bool('use_first_pool', True))
def proxylessnas_cpu(**kwargs): return get_proxylessnas(version='cpu', model_name='proxylessnas_cpu', **kwargs)
def get_site(): m = re.search('([^.]+)\\.brainpp\\.cn$', socket.getfqdn()) if m: return m.group(1)
def GetService(name, config_class): service = None while (service is None): rospy.wait_for_service(name) service = rospy.ServiceProxy(name, config_class) return service
def board8x8() -> Board: board = jnp.array([[2, 1, 1, 0, 2, 1, 1, 0], [2, 3, 1, 2, 2, 1, 1, 0], [2, 3, 1, 2, 2, 1, 1, 0], [2, 3, 0, 0, 2, 1, 1, 0], [1, 1, 0, 0, 0, 1, 1, 2], [0, 1, 1, 2, 2, 3, 0, 0], [1, 0, 0, 2, 1, 2, 3, 5], [2, 0, 0, 0, 2, 1, 1, 5]]) return board
def inception_v2(inputs, num_classes=1000, is_training=True, dropout_keep_prob=0.8, min_depth=16, depth_multiplier=1.0, prediction_fn=slim.softmax, spatial_squeeze=True, reuse=None, scope='InceptionV2'): if (depth_multiplier <= 0): raise ValueError('depth_multiplier is not greater than zero.') with tf.variable_scope(scope, 'InceptionV2', [inputs, num_classes], reuse=reuse) as scope: with slim.arg_scope([slim.batch_norm, slim.dropout], is_training=is_training): (net, end_points) = inception_v2_base(inputs, scope=scope, min_depth=min_depth, depth_multiplier=depth_multiplier) with tf.variable_scope('Logits'): kernel_size = _reduced_kernel_size_for_small_input(net, [7, 7]) net = slim.avg_pool2d(net, kernel_size, padding='VALID', scope='AvgPool_1a_{}x{}'.format(*kernel_size)) net = slim.dropout(net, keep_prob=dropout_keep_prob, scope='Dropout_1b') logits = slim.conv2d(net, num_classes, [1, 1], activation_fn=None, normalizer_fn=None, scope='Conv2d_1c_1x1') if spatial_squeeze: logits = tf.squeeze(logits, [1, 2], name='SpatialSqueeze') end_points['Logits'] = logits end_points['Predictions'] = prediction_fn(logits, scope='Predictions') return (logits, end_points)