code
stringlengths
101
5.91M
def hed(): img_input = Input(shape=(480, 480, 3), name='input') x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv1')(img_input) x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv2')(x) b1 = side_branch(x, 1) x = MaxPooling2D((2, 2), strides=(2, 2), padding='same', name='block1_pool')(x) x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv1')(x) x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv2')(x) b2 = side_branch(x, 2) x = MaxPooling2D((2, 2), strides=(2, 2), padding='same', name='block2_pool')(x) x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv1')(x) x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv2')(x) x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv3')(x) b3 = side_branch(x, 4) x = MaxPooling2D((2, 2), strides=(2, 2), padding='same', name='block3_pool')(x) x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv1')(x) x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv2')(x) x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv3')(x) b4 = side_branch(x, 8) x = MaxPooling2D((2, 2), strides=(2, 2), padding='same', name='block4_pool')(x) x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv1')(x) x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv2')(x) x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv3')(x) b5 = side_branch(x, 16) fuse = Concatenate(axis=(- 1))([b1, b2, b3, b4, b5]) fuse = Conv2D(1, (1, 1), padding='same', use_bias=False, activation=None)(fuse) o1 = Activation('sigmoid', name='o1')(b1) o2 = Activation('sigmoid', name='o2')(b2) o3 = Activation('sigmoid', name='o3')(b3) o4 = Activation('sigmoid', name='o4')(b4) o5 = Activation('sigmoid', name='o5')(b5) ofuse = Activation('sigmoid', name='ofuse')(fuse) model = Model(inputs=[img_input], outputs=[o1, o2, o3, o4, o5, ofuse]) filepath = './models/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5' model.compile(loss={'o1': cross_entropy_balanced, 'o2': cross_entropy_balanced, 'o3': cross_entropy_balanced, 'o4': cross_entropy_balanced, 'o5': cross_entropy_balanced, 'ofuse': cross_entropy_balanced}, metrics={'ofuse': ofuse_pixel_error}, optimizer='adam') return model
class Net(nn.Module): def __init__(self): super(Net, self).__init__() self.conv1 = nn.Conv2d(1, 20, 5, 1) self.conv2 = nn.Conv2d(20, 50, 5, 1) self.fc1 = nn.Linear(((4 * 4) * 50), 500) self.fc2 = nn.Linear(500, 10) def forward(self, x): x = F.relu(self.conv1(x)) x = F.max_pool2d(x, 2, 2) x = F.relu(self.conv2(x)) x = F.max_pool2d(x, 2, 2) x = x.view((- 1), ((4 * 4) * 50)) x = F.relu(self.fc1(x)) x = self.fc2(x) return F.log_softmax(x, dim=1)
class ModelEvaluator(Evaluator): def __init__(self, dataset: Dataset, batch_size: int, num_workers: int, mixed_precision: bool=True): self.dataset = dataset self.mixed_precision = mixed_precision self.loader = DataLoader(dataset, batch_size, shuffle=False, num_workers=num_workers, drop_last=False) def num_batches(self): return len(self.loader) def evaluate(self, model: Classifier, device: Optional[Union[(torch.device, str)]]=None) -> Evaluator.Result: return expand_generator(self.evaluate_iter(model, device), return_only=True) def evaluate_iter(self, model: Classifier, device: Optional[Union[(torch.device, str)]]=None) -> Generator[(dict, None, Evaluator.Result)]: with model.as_eval(), torch.no_grad(), torch.cuda.amp.autocast(enabled=self.mixed_precision): mean_accuracy = 0.0 mean_log_loss = 0.0 for (i, (x, y)) in enumerate(self.loader): x = x.to(device) y = y.to(device) logits = model(x) correct = torch.sum((logits.argmax((- 1)) == y)).item() log_loss = F.cross_entropy(logits, y, reduction='sum').item() mean_accuracy += (correct / len(self.dataset)) mean_log_loss += (log_loss / len(self.dataset)) (yield dict(batch=i)) return self.Result(accuracy=mean_accuracy, log_loss=mean_log_loss)
def videodata_kwargs(cfg): return {'root': cfg.data.root, 'root_targets': cfg.data.root_targets, 'sources': cfg.data.sources, 'targets': cfg.data.targets, 'height': cfg.data.height, 'width': cfg.data.width, 'transforms': cfg.data.transforms, 'norm_mean': cfg.data.norm_mean, 'norm_std': cfg.data.norm_std, 'use_gpu': cfg.use_gpu, 'split_id': cfg.data.split_id, 'combineall': cfg.data.combineall, 'batch_size_train': cfg.train.batch_size, 'batch_size_test': cfg.test.batch_size, 'workers': cfg.data.workers, 'num_instances': cfg.sampler.num_instances, 'num_cams': cfg.sampler.num_cams, 'num_datasets': cfg.sampler.num_datasets, 'train_sampler': cfg.sampler.train_sampler, 'seq_len': cfg.video.seq_len, 'sample_method': cfg.video.sample_method}
def make_vector_field(eval_func, x_bounds, y_bounds, *, resolution=10, info=None): if (info is None): info = {} x_values = np.linspace(*x_bounds, num=resolution) y_values = np.linspace(*y_bounds, num=resolution) values = np.zeros((resolution, resolution)) dx_values = np.zeros((resolution, resolution)) dy_values = np.zeros((resolution, resolution)) for x in range(resolution): for y in range(resolution): (value, dx, dy) = eval_func(x_values[x], y_values[y]) values[(x, y)] = value dx_values[(x, y)] = dx dy_values[(x, y)] = dy return VectorField(values=values, dx_values=dx_values, dy_values=dy_values, x_values=x_values, y_values=y_values, info=info)
class Attention2d(torch.nn.Module): def __init__(self, in_channels: int, out_channels: int, num_kernels: int, kernel_size: Tuple[(int, int)], padding_size: Tuple[(int, int)]): super(Attention2d, self).__init__() self.conv_depth = torch.nn.Conv2d(in_channels=in_channels, out_channels=(in_channels * num_kernels), kernel_size=kernel_size, padding=padding_size, groups=in_channels) self.conv_point = torch.nn.Conv2d(in_channels=(in_channels * num_kernels), out_channels=out_channels, kernel_size=(1, 1)) self.bn = torch.nn.BatchNorm2d(num_features=out_channels) self.activation = torch.nn.Sigmoid() def forward(self, x: torch.Tensor, size: torch.Size) -> torch.Tensor: x = F.adaptive_max_pool2d(x, size) x = self.conv_depth(x) x = self.conv_point(x) x = self.bn(x) x = self.activation(x) return x
class A000255(ExtremesOfPermanentsSequence): def __init__(self): SloaneSequence.__init__(self, offset=0) self._b = [] self._a0a1d = (1, 1, 1) self._precompute(2) def _repr_(self): return 'a(n) = n*a(n-1) + (n-1)*a(n-2), a(0) = 1, a(1) = 1.'
class BasicAggModel(nn.Module): def __init__(self, include_ff=True, include_res_ln=True, dropout=0.0, d_inner=2048, d_model=768, return_att_weights=False, n_head=8, d_k=96, n_rules=63, device='cpu', is_dense_bias=True): super(BasicAggModel, self).__init__() self.include_ff = include_ff self.include_res_ln = include_res_ln self.d_inner = d_inner self.d_model = d_model self.n_rules = n_rules self.device = device self.return_att_weights = return_att_weights self.mha = MultiHeadAttention(n_head=n_head, d_k=d_k, d_v=d_k, dropout=dropout, return_att_weights=return_att_weights, include_res_ln=include_res_ln) self.ff = PositionwiseFeedForward(self.d_model, self.d_inner, dropout=dropout) self.layer_norm = nn.LayerNorm(d_model, eps=1e-06) self.dropout = nn.Dropout(p=dropout) self.rule_encoder = nn.Embedding(self.n_rules, self.d_model) if is_dense_bias: self.dense_proj = nn.Linear(d_model, n_rules) else: self.dense_proj = nn.Linear(d_model, n_rules, bias=False) for (name, p) in self.named_parameters(): if ((name != 'dense_proj.weight') and (p.dim() > 1)): nn.init.xavier_uniform_(p) def forward(self, query, keys, mask): query = torch.unsqueeze(query, 1) bs = query.size(0) mask = torch.sum(mask, dim=(- 1)) mask = mask.unsqueeze(1) values = keys query = self.layer_norm(query) keys = self.layer_norm(keys) values = self.layer_norm(values) (pred, att_weights) = self.mha(query, keys, values, mask) if self.include_ff: pred = self.ff(pred) pred = self.dense_proj(pred) if self.return_att_weights: return (pred, att_weights) else: return (pred, None)
.skipif((not limits.can_set_time_limit()), reason='Cannot set time limits on this system') def test_hard_time_limit(): def preexec_fn(): limits.set_time_limit(10) driver = [sys.executable, 'fast-downward.py'] parameters = ['--translate', '--translate-time-limit', '10s', 'misc/tests/benchmarks/gripper/prob01.pddl'] subprocess.check_call((driver + parameters), preexec_fn=preexec_fn, cwd=REPO_ROOT_DIR) parameters = ['--translate', '--translate-time-limit', '20s', 'misc/tests/benchmarks/gripper/prob01.pddl'] with pytest.raises(subprocess.CalledProcessError) as exception_info: subprocess.check_call((driver + parameters), preexec_fn=preexec_fn, cwd=REPO_ROOT_DIR) assert (exception_info.value.returncode == returncodes.DRIVER_INPUT_ERROR)
class KR_type_Dn_twistedElement(KirillovReshetikhinGenericCrystalElement): def e0(self): n = (self.parent().cartan_type().rank() - 1) s = self.parent().s() [b, l] = self.lift().to_highest_weight(index_set=list(range(2, (n + 1)))) pm = self.parent().from_highest_weight_vector_to_pm_diagram(b) [l1, l2] = pm.pm_diagram[(n - 1)] l3 = pm.pm_diagram[(n - 2)][0] if ((((l1 + l2) + l3) == s) and (l1 == 0)): return None if (((l1 + l2) + l3) < s): pm.pm_diagram[(n - 1)][1] = (l2 + 2) pm.pm_diagram[n][0] -= 2 elif (l1 > 1): pm.pm_diagram[(n - 1)][0] = (l1 - 2) pm.pm_diagram[n][0] += 2 elif (l1 == 1): pm.pm_diagram[(n - 1)][0] = 0 pm.pm_diagram[(n - 1)][1] = (l2 + 1) pm = PMDiagram(pm.pm_diagram) b = self.parent().from_pm_diagram_to_highest_weight_vector(pm) b = b.f_string(reversed(l)) return self.parent().retract(b) def f0(self): n = (self.parent().cartan_type().rank() - 1) s = self.parent().s() [b, l] = self.lift().to_highest_weight(index_set=list(range(2, (n + 1)))) pm = self.parent().from_highest_weight_vector_to_pm_diagram(b) [l1, l2] = pm.pm_diagram[(n - 1)] l3 = pm.pm_diagram[(n - 2)][0] if ((((l1 + l2) + l3) == s) and (l2 == 0)): return None if (((l1 + l2) + l3) < s): pm.pm_diagram[(n - 1)][0] = (l1 + 2) pm.pm_diagram[n][0] -= 2 elif (l2 > 1): pm.pm_diagram[(n - 1)][1] = (l2 - 2) pm.pm_diagram[n][0] += 2 elif (l2 == 1): pm.pm_diagram[(n - 1)][1] = 0 pm.pm_diagram[(n - 1)][0] = (l1 + 1) pm = PMDiagram(pm.pm_diagram) b = self.parent().from_pm_diagram_to_highest_weight_vector(pm) b = b.f_string(reversed(l)) return self.parent().retract(b) def epsilon0(self): n = (self.parent().cartan_type().rank() - 1) [b, l] = self.lift().to_highest_weight(index_set=list(range(2, (n + 1)))) pm = self.parent().from_highest_weight_vector_to_pm_diagram(b) l1 = pm.pm_diagram[(n - 1)][0] l4 = pm.pm_diagram[n][0] return (l1 + (l4 // 2)) def phi0(self): n = (self.parent().cartan_type().rank() - 1) (b, l) = self.lift().to_highest_weight(index_set=list(range(2, (n + 1)))) pm = self.parent().from_highest_weight_vector_to_pm_diagram(b) l2 = pm.pm_diagram[(n - 1)][1] l4 = pm.pm_diagram[n][0] return (l2 + (l4 // 2))
.parametrize('dtype', [ti.i32, ti.f32, ti.i64, ti.f64]) .parametrize('shape', [(8,), (6, 12)]) .parametrize('offset', [0, (- 4), 4]) .parametrize('m, n', [(3, 4)]) _utils.test(arch=get_host_arch_list()) def test_matrix_to_numpy_with_offset(dtype, shape, offset, m, n): import numpy as np x = ti.Matrix.field(dtype=dtype, m=m, n=n, shape=shape, offset=([offset] * len(shape))) x.fill(1.0) numpy_dtypes = {ti.i32: np.int32, ti.f32: np.float32, ti.f64: np.float64, ti.i64: np.int64} numpy_shape = (((shape,) if isinstance(shape, int) else shape) + (n, m)) arr = x.to_numpy() assert np.allclose(arr, np.ones(numpy_shape, dtype=numpy_dtypes[dtype]))
def feature_column_json_hook(obj): if isinstance(obj, dict): typ = obj.get('type') if (typ in SUPPORTED_CONCRETE_FEATURE_COLUMNS): return FeatureColumn.from_dict_or_feature_column(obj) return obj
class DecoderClassifier(nn.Module): def __init__(self, config, embedding_weights): super(DecoderClassifier, self).__init__() self.cls = BertOnlyMLMHead(config, embedding_weights) def forward(self, hidden_states): cls_scores = self.cls(hidden_states) return cls_scores
_utils.test(require=ti.extension.assertion, debug=True, gdb_trigger=False) def test_not_out_of_bound_with_offset(): x = ti.field(ti.i32, shape=(8, 16), offset=((- 4), (- 8))) def func(): x[((- 4), (- 8))] = 1 x[(3, 7)] = 2 func()
def read_frames_cv2_charades(video_path, num_frames, sample, start_sec=None, end_sec=None): cap = cv2.VideoCapture(video_path) assert cap.isOpened() vlen = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) fps = cap.get(5) if ((not start_sec) and (not end_sec)): frame_idxs = sample_frames(num_frames, vlen, sample=sample) else: start_f = max(0, int((start_sec * fps))) end_f = min(int((end_sec * fps)), vlen) frame_idxs = sample_frames_start_end(num_frames, start_f, end_f, sample=sample) frames = [] success_idxs = [] for index in frame_idxs: cap.set(cv2.CAP_PROP_POS_FRAMES, (index - 1)) (ret, frame) = cap.read() if ret: frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) frame = torch.from_numpy(frame) frame = frame.permute(2, 0, 1) frames.append(frame) success_idxs.append(index) else: pass frames = (torch.stack(frames).float() / 255) cap.release() return (frames, success_idxs)
class ETagResponseMixin(object): def cache_control(self): def on_update(cache_control): if ((not cache_control) and ('cache-control' in self.headers)): del self.headers['cache-control'] elif cache_control: self.headers['Cache-Control'] = cache_control.to_header() return parse_cache_control_header(self.headers.get('cache-control'), on_update, ResponseCacheControl) def _wrap_response(self, start, length): if (self.status_code == 206): self.response = _RangeWrapper(self.response, start, length) def _is_range_request_processable(self, environ): return ((('HTTP_IF_RANGE' not in environ) or (not is_resource_modified(environ, self.headers.get('etag'), None, self.headers.get('last-modified'), ignore_if_range=False))) and ('HTTP_RANGE' in environ)) def _process_range_request(self, environ, complete_length=None, accept_ranges=None): from ..exceptions import RequestedRangeNotSatisfiable if ((accept_ranges is None) or (complete_length is None) or (not self._is_range_request_processable(environ))): return False parsed_range = parse_range_header(environ.get('HTTP_RANGE')) if (parsed_range is None): raise RequestedRangeNotSatisfiable(complete_length) range_tuple = parsed_range.range_for_length(complete_length) content_range_header = parsed_range.to_content_range_header(complete_length) if ((range_tuple is None) or (content_range_header is None)): raise RequestedRangeNotSatisfiable(complete_length) content_length = (range_tuple[1] - range_tuple[0]) self.headers['Content-Length'] = content_length self.headers['Accept-Ranges'] = accept_ranges self.content_range = content_range_header self.status_code = 206 self._wrap_response(range_tuple[0], content_length) return True def make_conditional(self, request_or_environ, accept_ranges=False, complete_length=None): environ = _get_environ(request_or_environ) if (environ['REQUEST_METHOD'] in ('GET', 'HEAD')): if ('date' not in self.headers): self.headers['Date'] = accept_ranges = _clean_accept_ranges(accept_ranges) is206 = self._process_range_request(environ, complete_length, accept_ranges) if ((not is206) and (not is_resource_modified(environ, self.headers.get('etag'), None, self.headers.get('last-modified')))): if parse_etags(environ.get('HTTP_IF_MATCH')): self.status_code = 412 else: self.status_code = 304 if (self.automatically_set_content_length and ('content-length' not in self.headers)): length = self.calculate_content_length() if (length is not None): self.headers['Content-Length'] = length return self def add_etag(self, overwrite=False, weak=False): if (overwrite or ('etag' not in self.headers)): self.set_etag(generate_etag(self.get_data()), weak) def set_etag(self, etag, weak=False): self.headers['ETag'] = quote_etag(etag, weak) def get_etag(self): return unquote_etag(self.headers.get('ETag')) def freeze(self, no_etag=False): if (not no_etag): self.add_etag() super(ETagResponseMixin, self).freeze() accept_ranges = header_property('Accept-Ranges', doc="The `Accept-Ranges` header. Even though the name would\n indicate that multiple values are supported, it must be one\n string token only.\n\n The values ``'bytes'`` and ``'none'`` are common.\n\n .. versionadded:: 0.7") def content_range(self): def on_update(rng): if (not rng): del self.headers['content-range'] else: self.headers['Content-Range'] = rng.to_header() rv = parse_content_range_header(self.headers.get('content-range'), on_update) if (rv is None): rv = ContentRange(None, None, None, on_update=on_update) return rv _range.setter def content_range(self, value): if (not value): del self.headers['content-range'] elif isinstance(value, string_types): self.headers['Content-Range'] = value else: self.headers['Content-Range'] = value.to_header()
def write_file(file): file.write(TXT) for words in test_words(): record = Record() for word in words: record.add(word) file.write(str(record)) file.write('\n')
class EntropyEstimator(BaseEstimator, ABC, metaclass=EntropyEstimatorType): def __init__(self): self.estimate_ = None self.err_ = None self.input_data_ndim = 1 def __call__(self, nk, k=None, zk=None): return self.fit(nk, k=k, zk=zk).estimate_ def algorithm(self): return self.__class__.__name__ def check_alpha(a): error_msg = ('alpha must be a positive number (got %r).' % a) if (a is None): raise AlphaError(error_msg) try: a = numpy.float64(a) except ValueError: raise AlphaError(error_msg) if (a <= 0): raise AlphaError(error_msg) return a def fit(self, nk, k=None, zk=None):
def test_vector_draw(verbose=False): np.random.seed(0) ppg = pypolyagamma.PyPolyaGamma(np.random.randint((2 ** 16))) n = 5 v2 = np.zeros(n) a = (14 * np.ones(n, dtype=np.float)) b = (0 * np.ones(n, dtype=np.float)) ppg.pgdrawv(a, b, v2) if verbose: print(v2) return True
class Parser(argparse.ArgumentParser, ABC): def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) model_args = self.add_argument_group('model_args') self._add_model_args(model_args) data_args = self.add_argument_group('data_args') self._add_data_args(data_args) partitioning_args = self.add_argument_group('partitioning_args') self._add_partitioning_args(partitioning_args) heuristics_args = self.add_argument_group('heuristics_args') self._add_heurisitcs_args(heuristics_args) presets_args = self.add_argument_group('presets') self._add_presets_args(presets_args) METIS_args = self.add_argument_group('METIS_args') self._add_METIS_args(METIS_args) acyclic_args = self.add_argument_group('acyclic_args') self._add_acyclic_args(acyclic_args) binpack_args = self.add_argument_group('binpack_args') self._add_binpack_args(binpack_args) mpipe_args = self.add_argument_group('mpipe_args') self._add_mpipe_args(mpipe_args) analysis_args = self.add_argument_group('analysis_args') self._add_analysis_args(analysis_args) extra_args = self.add_argument_group('extra_args') extra_args.add_argument('--debug', action='store_true', default=False) self._extra(extra_args) self.set_defaults(**self._default_values()) def _add_model_args(self, group): def _add_data_args(self, group): def _add_analysis_args(self, group): analysis_mode = group.add_mutually_exclusive_group() analysis_mode.add_argument('--no_analysis', action='store_true', default=False, help='disable partition analysis') analysis_mode.add_argument('--analysis_only', action='store_true', default=False, help='run only analysis for partitioned model') group.add_argument('--analysis_batch_size', default=32, type=int, help='batch size to use during the post partition analysis') group.add_argument('--analysis_as_async_pipeline', default=False, action='store_true', help='Force analysis as async pipeline') def _add_heurisitcs_args(self, group): group.add_argument('--bw', type=float, default=12, help='data transfer rate between gpus in GBps (Gigabytes per second)') group.add_argument('--bwd_to_fwd_ratio', type=float, default=1, help='bwd to fwd ratio for heuristics') group.add_argument('--auto_infer_node_bwd_to_fwd_ratio', action='store_true', default=False, help='Automatically infer bwd to fwd ratio for nodes (computation). Expected Ratio for edges should be given `by bwd_to_fwd_ratio`') group.add_argument('--penalize_non_tensors', action='store_true', default=False, help='penalize edges with non tensor outputs by default no penalties are applied') group.add_argument('--weight_mult_factor', type=float, default=10000.0, help='a constant to multiply weights with (useful if weights are really small)') group.add_argument('--edge_penalty', type=float, default=.0, help='multiplicative penalty for edges if `penalize_non_tensors` is set') def _add_partitioning_args(self, group): group.add_argument('-b', '--partitioning_batch_size', type=int, default=128) group.add_argument('-p', '--n_partitions', type=int, default=4) group.add_argument('-o', '--output_file', default='') group.add_argument('--disable_autogenerated_name', action='store_true', default=False) group.add_argument('--n_iter', type=int, default=10, help='number of iteration used in order to profile the network and run analysis') group.add_argument('--no_recomputation', action='store_true', default=False, help='whether to (not) use recomputation for the backward pass') group.add_argument('--depth', default=10000, type=int, help='the depth in which we will partition the model') group.add_argument('--basic_blocks', nargs='*', default=[]) group.add_argument('--use_network_profiler', default=False, action='store_true', help='whether to use the old network_profiler instead of the newer graph based profiler') group.add_argument('--sanity_check', default=False, action='store_true', help='whether to use do sanity check after partitioning') group.add_argument('--disable_op_profiling', default=False, action='store_true', help='weheter to not profile ops when using the GraphProfiler') group.add_argument('--partitioning_method', '-m', choices=['acyclic', 'metis', '2dbin', 'mpipe', 'pipedream'], default='acyclic') group.add_argument('--generate_explicit_del', action='store_true', default=False, help='whether to generate del statements in partitioned code') group.add_argument('--no_activation_propagation', action='store_true', default=False, help='whether to not propgate activations in the pipeline, and having direct sends instead') group.add_argument('-a', '--async_pipeline', default=False, action='store_true', help='Do partitioning and analysis for async pipeline') group.add_argument('--dont_use_async_meta_alg', default=False, action='store_true', help='Explicitly avoid the async meta alg. (e.g when number of stages is big)') group.add_argument('--dot', default=False, action='store_true', help='Save and plot it using graphviz') group.add_argument('--save_memory_mode', default=False, action='store_true', help=('Save memory during profiling by storing everything on cpu,' + ' but sending each layer to GPU before the profiling.')) group.add_argument('--trace_on_gpu', default=False, action='store_true', help='Used together with save_memory_mode: if true, will trace the model on GPU despite swapping during profiling.') group.add_argument('--force_no_recomputation_scopes', nargs='*', default=[]) group.add_argument('--cp', '--profiles_cache_name', default='', type=str, dest='profiles_cache_name', help='Profile cache to use in case of multiple runs') group.add_argument('--overwrite_profiles_cache', action='store_true', default=False, help='overwrite profile cache') group.add_argument('--ct', '--trace_cache_name', default='', type=str, dest='trace_cache_name', help='Trace cache to use in case of multiple runs') group.add_argument('--overwrite_trace_cache', action='store_true', default=False, help='overwrite trace cache') def _add_METIS_args(self, group): group.add_argument('--metis_attempts', type=int, default=10, help='number of attempts for running the METIS partitioning algorithm') group.add_argument('--metis_verbose_on_error', action='store_true', default=False, help='whether to print the cause of the error') group.add_argument('--metis_seed', required=False, type=int, help='Random seed for Metis algorithm') group.add_argument('--metis_compress', default=False, action='store_true', help='Compress') group.add_argument('--metis_niter', type=int, help='Specifies the number of iterations for the refinement algorithms at each stage of the uncoarsening process.Default is 10.') group.add_argument('--metis_nseps', type=int, help='Specifies the number of different separators that it will compute at each level of nested dissection.The final separator that is used is the smallest one. Default is 1.') group.add_argument('--metis_ncuts', type=int, help='Specifies the number of different partitionings that it will compute. The final partitioning is the one that achieves the best edgecut or communication volume.Default is 1.') group.add_argument('--metis_dbglvl', type=int, help='Metis debug level. Refer to the docs for explanation') group.add_argument('--metis_objtype', type=int, help='Extra objective type to miminize (0: edgecut, 1: vol, default: edgecut)') group.add_argument('--metis_contig', default=False, action='store_true', help='A boolean to create contigous partitions.') def _add_binpack_args(self, group): group.add_argument('--n_clusters', default=2, type=int, help='number of clusters in the model') group.add_argument('--analyze_n_clusters', action='store_true', default=False, help='analyze number of clusters') group.add_argument('--reminder_policy', type=str, choices=list(ReminderPolicy._value2member_map_.keys()), default='last', help=f'Policy for reminder items in cluster, {ReminderPolicy._value2member_map_}') group.add_argument('--second_and_on_cluster_policy', type=str, choices=list(SecondAndOnClusterPolicy._value2member_map_.keys()), default='best_fit', help=f'Policy for 2nd and on cluster {SecondAndOnClusterPolicy._value2member_map_}') group.add_argument('--THRESHOLD', type=float, default=0, help='values <= threshold will be contagious with closest stage') def _add_mpipe_args(self, group): group.add_argument('--special_blocks', type=str, nargs='*', default=[]) group.add_argument('--L', nargs='*', type=int, default=[]) def _add_acyclic_args(self, group): group.add_argument('--epsilon', default=0.1, type=float, help='imbalance factor') group.add_argument('--rounds', default=10, type=int, help='number of optimization rounds default is 10') group.add_argument('--allocated_seconds', default=20, type=int, help='run time allocated to the partitioning algorithm default is 20 seconds') group.add_argument('--multilevel', action='store_true', default=False, help='whether to use multilevel partitioning algorithm') group.add_argument('--objective', choices=['edge_cut', 'stage_time'], default='stage_time', help='partitioning optimization objective') group.add_argument('--constraint', choices=['time', 'memory'], default='time', help='partitioning constraint') group.add_argument('--maximum_constraint_value', required=False, type=float, default=None, help='maximum constraint value a single stage can have,for example for memory constraint this is the maximum number of parameters a stage can have') def _add_presets_args(self, group): group.add_argument('--preset', choices=['ftpipe', 'pipedream', 'gpipe'], required=False, help='set preset partitioning and analysis arguments') def parse_presets(self, args): if (args.preset == 'ftpipe'): args.async_pipeline = True args.bwd_to_fwd_ratio = 1 elif (args.preset == 'pipedream'): args.async_pipeline = False args.bwd_to_fwd_ratio = 1 args.analysis_as_async_pipeline = True elif (args.preset == 'gpipe'): args.auto_infer_node_bwd_to_fwd_ratio = True args.async_pipeline = False args.bwd_to_fwd_ratio = (- 1) args.analysis_as_async_pipeline = False elif args.preset: raise NotImplementedError() return args def _extra(self, group): def _default_values(self): return dict() def _post_parse(self, args, argv): if argv: msg = gettext('unrecognized arguments: %s') self.error((msg % ' '.join(argv))) return args def _auto_file_name(self, args) -> str: return '' def parse_args(self, args=None, namespace=None) -> Namespace: (args, extra) = super().parse_known_args(args, namespace) self.parse_presets(args) args.acyclic_opt = self._acyclic_opts_dict_from_parsed_args(args) args.METIS_opt = self._metis_opts_dict_from_parsed_args(args) args.binpack_opt = self._binpack_opts_dict_from_parsed_args(args) args.mpipe_opt = self._mpipe_opts_dict_from_parsed_args(args) if ((not args.disable_autogenerated_name) or (not args.output_file)): args.output_file = self._auto_file_name(args) if args.output_file.endswith('.py'): args.output_file = args.output_file[:(- 3)] device = torch.device(('cuda' if torch.cuda.is_available() else 'cpu')) args.device = device args.force_no_recomputation_scopes_fn = (lambda scope: any(((s in scope) for s in args.force_no_recomputation_scopes))) return self._post_parse(args, extra) def _acyclic_opts_dict_from_parsed_args(args): if (args.objective == 'edge_cut'): objective = Objective.EDGE_CUT else: objective = Objective.STAGE_TIME if args.multilevel: meta_algorithm = META_ALGORITH.MULTI_LEVEL else: meta_algorithm = META_ALGORITH.SINGLE_LEVEL if (args.constraint == 'time'): constraint = Constraint.TIME else: constraint = Constraint.MEMORY return {'epsilon': args.epsilon, 'rounds': args.rounds, 'allocated_seconds': args.allocated_seconds, 'meta_algorithm': meta_algorithm, 'objective': objective, 'constraint': constraint, 'maximum_constraint_value': args.maximum_constraint_value} def _metis_opts_dict_from_parsed_args(args): METIS_opt = {'verbose_on_error': getattr(args, 'metis_verbose_on_error', False), 'attempts': getattr(args, 'metis_attempts', 1000), 'seed': getattr(args, 'metis_seed', None), 'nseps': getattr(args, 'nseps', None), 'niter': getattr(args, 'metis_niter', None), 'compress': getattr(args, 'metis_compress', None), 'ncuts': getattr(args, 'metis_ncuts', None), 'objtype': getattr(args, 'metis_objtype', None), 'contig': getattr(args, 'metis_contig', None), '_dbglvl': 1} return METIS_opt def _binpack_opts_dict_from_parsed_args(self, args): d = dict() d['n_clusters'] = args.n_clusters d['analyze_n_clusters'] = args.analyze_n_clusters if hasattr(args, 'second_and_on_cluster_policy'): d['second_and_on_cluster_policy'] = args.second_and_on_cluster_policy if hasattr(args, 'reminder_policy'): d['reminder_policy'] = args.reminder_policy d['THRESHOLD'] = args.THRESHOLD return d def _mpipe_opts_dict_from_parsed_args(self, args): d = dict() d['depth'] = args.depth d['L_list'] = args.L return d
def add_to_ld_library_path(path): library_path = os.environ.get('LD_LIBRARY_PATH', '') library_paths = (library_path.split(':') if library_path else []) if (path not in library_paths): os.environ['LD_LIBRARY_PATH'] = ':'.join(([path] + library_paths))
def _class_counter(data_dict): counter = Counter() for (data_id, data) in data_dict.items(): counter.update([data['class_name']]) return counter
def inference(data): with tf.variable_scope('inference') as scope: W_1 = utils.weight_variable([((IMAGE_SIZE * IMAGE_SIZE) * 50)], name='W_1') b_1 = utils.bias_variable([50], name='b_1') h_1 = tf.nn.relu((tf.matmul(data, tf.reshape(W_1, [(IMAGE_SIZE * IMAGE_SIZE), 50])) + b_1), name='h_1') utils.add_activation_summary(h_1) W_2 = utils.weight_variable([(50 * 50)], name='W_2') b_2 = utils.bias_variable([50], name='b_2') h_2 = tf.nn.relu((tf.matmul(h_1, tf.reshape(W_2, [50, 50])) + b_2), name='h_2') utils.add_activation_summary(h_2) W_3 = utils.weight_variable([(50 * 50)], name='W_3') b_3 = utils.bias_variable([50], name='b_3') h_3 = tf.nn.relu((tf.matmul(h_2, tf.reshape(W_3, [50, 50])) + b_3), name='h_3') utils.add_activation_summary(h_3) W_4 = utils.weight_variable([(50 * 50)], name='W_4') b_4 = utils.bias_variable([50], name='b_4') h_4 = tf.nn.relu((tf.matmul(h_3, tf.reshape(W_4, [50, 50])) + b_4), name='h_4') utils.add_activation_summary(h_4) W_final = utils.weight_variable([(50 * 10)], name='W_final') b_final = utils.bias_variable([10], name='b_final') pred = tf.nn.softmax((tf.matmul(h_4, tf.reshape(W_final, [50, 10])) + b_final), name='h_final') return pred
class MSELoss(Loss): def __init__(self): super().__init__('MSELoss') def compute_loss(self, y_true, output_model): if (output_model is None): raise TypeError('Argument: output_model must be set.') if (y_true is None): raise TypeError('Argument: y_true must be set.') if (not isinstance(output_model, torch.Tensor)): raise TypeError('Argument: output_model must be a pytorch tensor.') if (not isinstance(y_true, torch.Tensor)): raise TypeError('Argument: y_true must be a pytorch tensor.') self.__check_dim_pred_gt__(output_model, y_true) result = (((output_model - y_true) ** 2).sum() / output_model.data.nelement()) if (not (0 <= result <= 1)): raise ValueError('The output of MSELoss.compute_loss must be in [0,1]') return result
def make_numpy_ndarray_fromstring(s, dtype, shape): return numpy.fromstring(s, dtype=dtype).reshape(shape)
def plot_recall(measures, eval_dir, plot_file): plt.figure(figsize=(10, 8)) plt.xlabel('cut-off', fontsize=15) plt.ylabel('recall', fontsize=15) for (name, measure) in measures.items(): (xs, ys) = zip(*measure.values()) labels = measure.keys() plt.scatter(xs, ys, marker='o') plt.plot(xs, ys, label=name) plt.legend(loc='lower right') plt.savefig(os.path.join(eval_dir, plot_file))
def check_cuda_kernel_launches(): torch_dir = os.path.dirname(os.path.realpath(__file__)) torch_dir = os.path.dirname(torch_dir) torch_dir = os.path.dirname(torch_dir) kernels_without_checks = 0 files_without_checks = [] for (root, dirnames, filenames) in os.walk(torch_dir): if ((root == os.path.join(torch_dir, 'build')) or (root == os.path.join(torch_dir, 'torch/include'))): dirnames[:] = [] continue for x in filenames: filename = os.path.join(root, x) file_result = check_file(filename) if (file_result > 0): kernels_without_checks += file_result files_without_checks.append(filename) if (kernels_without_checks > 0): count_str = f"Found {kernels_without_checks} instances in {len(files_without_checks)} files where kernel launches didn't have checks." print(count_str, file=sys.stderr) print('Files without checks:', file=sys.stderr) for x in files_without_checks: print(f' {x}', file=sys.stderr) print(count_str, file=sys.stderr) return kernels_without_checks
class Aggregator(object): def __init__(self, batch_size, dim, dropout, act, name): if (not name): layer = self.__class__.__name__.lower() name = ((layer + '_') + str(get_layer_id(layer))) self.name = name self.dropout = dropout self.act = act self.batch_size = batch_size self.dim = dim def __call__(self, self_vectors, neighbor_vectors): outputs = self._call(self_vectors, neighbor_vectors) return outputs def _call(self, self_vectors, neighbor_vectors): pass def _mix_neighbor_vectors(self, neighbor_vectors): print(neighbor_vectors) b = tf.reduce_max(neighbor_vectors, (- 1)) c = (b > 0.0) d = tf.reduce_sum(tf.cast(c, tf.float32), (- 1), keepdims=True) d = tf.nn.bias_add(d, [1e-10]) print('agg', b, c, d) e = tf.tile(d, [1, 1, self.dim]) neighbors_aggregated = (tf.reduce_sum(neighbor_vectors, axis=2) / e) return neighbors_aggregated
((not tf), 'no TF') def test_demo_tf_task12ax_eval(): cfg_filename = 'demos/demo-tf-native-lstm.12ax.config' train_dataset_repr = '{"class": "Task12AXDataset", "num_seqs": 10}' dev_dataset_repr = '{"class": "Task12AXDataset", "num_seqs": 10}' fer1 = run_config_get_fer(cfg_filename, '++num_epochs', '2', '++train', train_dataset_repr, '++dev', dev_dataset_repr, print_stdout=True, post_cleanup=False) fer2 = run_config_get_fer(cfg_filename, '++task', 'eval', '++load_epoch', '2', '++train', 'None', '++dev', dev_dataset_repr, print_stdout=True, pre_cleanup=False, post_cleanup=False) assert (fer1 == fer2) model_filename = get_model_filename(cfg_filename) ep2_files = glob((model_filename + '.002.*')) assert ep2_files, f'No model files found for epoch 2, {model_filename}' for fn in ep2_files: shutil.copy(fn, fn.replace('.002.', '.003.')) fer3 = run_config_get_fer(cfg_filename, '++task', 'eval', '++load_epoch', '3', '++train', 'None', '++dev', dev_dataset_repr, print_stdout=True, pre_cleanup=False) assert (fer3 != fer2)
def sample_queries(data_dir, db_name, out_dir): in_json = os.path.join(data_dir, '{}.json'.format(db_name)) sqls = load_sqls(in_json, normalize_variables=True) num_samples = 3 count = 0 for idx in np.random.randint(0, len(sqls), num_samples): out_txt = os.path.join(out_dir, '{}-{}.txt'.format(db_name, count)) with open(out_txt, 'w') as o_f: o_f.write(sqls[idx]) print('{} SQL query saved to {}'.format(db_name, out_txt)) count += 1
class SetupCallback(Callback): def __init__(self, resume, now, logdir, ckptdir, cfgdir, config, lightning_config): super().__init__() self.resume = resume self.now = now self.logdir = logdir self.ckptdir = ckptdir self.cfgdir = cfgdir self.config = config self.lightning_config = lightning_config def on_keyboard_interrupt(self, trainer, pl_module): if (trainer.global_rank == 0): print('Summoning checkpoint.') ckpt_path = os.path.join(self.ckptdir, 'last.ckpt') trainer.save_checkpoint(ckpt_path) def on_fit_start(self, trainer, pl_module): if (trainer.global_rank == 0): os.makedirs(self.logdir, exist_ok=True) os.makedirs(self.ckptdir, exist_ok=True) os.makedirs(self.cfgdir, exist_ok=True) if ('callbacks' in self.lightning_config): if ('metrics_over_trainsteps_checkpoint' in self.lightning_config['callbacks']): os.makedirs(os.path.join(self.ckptdir, 'trainstep_checkpoints'), exist_ok=True) print('Project config') print(OmegaConf.to_yaml(self.config)) OmegaConf.save(self.config, os.path.join(self.cfgdir, '{}-project.yaml'.format(self.now))) print('Lightning config') print(OmegaConf.to_yaml(self.lightning_config)) OmegaConf.save(OmegaConf.create({'lightning': self.lightning_config}), os.path.join(self.cfgdir, '{}-lightning.yaml'.format(self.now))) elif ((not self.resume) and os.path.exists(self.logdir)): (dst, name) = os.path.split(self.logdir) dst = os.path.join(dst, 'child_runs', name) os.makedirs(os.path.split(dst)[0], exist_ok=True) try: os.rename(self.logdir, dst) except FileNotFoundError: pass
class ManinSymbolList_group(ManinSymbolList): def __init__(self, level, weight, syms): self.__level = level self.__syms = syms L = [(i, u, v) for i in range(((weight - 2) + 1)) for (u, v) in syms.list()] ManinSymbolList.__init__(self, weight, L) def level(self): return self.__level def apply_S(self, j): (i, u, v) = self._symbol_list[j] k = self.index((((self._weight - 2) - i), v, (- u))) return (k, ((- 1) if (i % 2) else 1)) def _apply_S_only_0pm1(self): return True def apply_I(self, j): (i, u, v) = self._symbol_list[j] k = self.index((i, (- u), v)) if ((i % 2) == 0): return (k, 1) else: return (k, (- 1)) def apply_T(self, j): k = self._weight (i, u, v) = self._symbol_list[j] (u, v) = self.__syms.normalize(v, ((- u) - v)) if (((k - 2) % 2) == 0): s = 1 else: s = (- 1) z = [] a = Integer(((k - 2) - i)) for j in range((((k - 2) - i) + 1)): m = self.index((j, u, v)) z.append((m, (s * a.binomial(j)))) s *= (- 1) return z def apply_TT(self, j): k = self._weight (i, u, v) = self._symbol_list[j] (u, v) = self.__syms.normalize(((- u) - v), u) if ((((k - 2) - i) % 2) == 0): s = 1 else: s = (- 1) z = [] a = Integer(i) for j in range((i + 1)): m = self.index(((((k - 2) - i) + j), u, v)) z.append((m, (s * a.binomial(j)))) s *= (- 1) return z def apply(self, j, m): (a, b, c, d) = (m[0], m[1], m[2], m[3]) (i, u, v) = self._symbol_list[j] P = apply_to_monomial(i, (self._weight - 2), a, b, c, d) m = self.index((0, ((u * a) + (v * c)), ((u * b) + (v * d)))) if (m == (- 1)): return [] r = len(self.__syms) return [((m + (r * k)), P[k]) for k in range(((self._weight - 2) + 1)) if (P[k] != 0)] def normalize(self, x): (u, v) = self.__syms.normalize(x[1], x[2]) return (x[0], u, v)
def register_Ns3CallbackImplBase_methods(root_module, cls): cls.add_constructor([]) cls.add_constructor([param('ns3::CallbackImplBase const &', 'arg0')]) cls.add_method('GetTypeid', 'std::string', [], is_pure_virtual=True, is_const=True, is_virtual=True) cls.add_method('IsEqual', 'bool', [param('ns3::Ptr< ns3::CallbackImplBase const >', 'other')], is_pure_virtual=True, is_const=True, is_virtual=True) cls.add_method('Demangle', 'std::string', [param('std::string const &', 'mangled')], is_static=True, visibility='protected') cls.add_method('GetCppTypeid', 'std::string', [], is_static=True, visibility='protected', template_parameters=[u'ns3::ObjectBase*']) cls.add_method('GetCppTypeid', 'std::string', [], is_static=True, visibility='protected', template_parameters=[u'void']) cls.add_method('GetCppTypeid', 'std::string', [], is_static=True, visibility='protected', template_parameters=[u'ns3::Ptr<ns3::MobilityModel const> ']) return
.parametrize('env_name', ['cartpole-random', 'pendulum-random']) def test_get_dataset(env_name: str) -> None: (_, env) = get_dataset(env_name) if (env_name == 'cartpole-random'): assert (env.unwrapped.spec.id == 'CartPole-v1') elif (env_name == 'pendulum-random'): assert (env.unwrapped.spec.id == 'Pendulum-v1')
def check_damping_factor(damping_factor: float): if ((damping_factor < 0) or (damping_factor >= 1)): raise ValueError('A damping factor must have a value in [0, 1[.')
class EmptyRecord(): def __init__(self, is_tuple, parameters): self.length_ = 0 self.is_tuple_ = is_tuple self.parameters_ = parameters self.set_id(Ref(0)) def append(self): self.length_ += 1 def extend(self, size): self.length_ += size def parameters(self): return self.parameters_ def set_id(self, id: Ref(int)): self.id_ = id.value def clear(self): self.length_ = 0 def length(self): return self.length_ def is_valid(self, error: Ref(str)): return True def buffer_nbytes(self, names_nbytes): pass def to_buffers(self, buffers): pass def form(self): params = ('' if (self.parameters_ == '') else f', parameters: {self.parameters_}') if self.is_tuple_: return f'{{"class": "RecordArray", "contents": [], "form_key": "node{self.id_}"{params}}}' else: return f'{{"class": "RecordArray", "contents": {{}}, "form_key": "node{self.id_}"{params}}}'
def convert_to_list(python_input): if isinstance(python_input, torch.Tensor): return [python_input] else: return list(python_input)
def get_eps_scheduler(args, max_eps, train_data): eps_scheduler = eval(args.scheduler_name)(max_eps, args.scheduler_opts) epoch_length = int((((len(train_data.dataset) + train_data.batch_size) - 1) / train_data.batch_size)) eps_scheduler.set_epoch_length(epoch_length) return eps_scheduler
class MonodepthOptions(): def __init__(self): self.parser = argparse.ArgumentParser(description='Monodepthv2 options') self.parser.add_argument('--data_path', type=str, help='path to the training data', default=os.path.join(file_dir, 'kitti')) self.parser.add_argument('--log_dir', type=str, help='log directory', default='./log') self.parser.add_argument('--model_name', type=str, help='the name of the folder to save the model in', default='mdp') self.parser.add_argument('--split', type=str, help='which training split to use', choices=['eigen_zhou', 'eigen_full', 'eigen_full_left', 'odom', 'benchmark'], default='eigen_full_left') self.parser.add_argument('--num_layers', type=int, help='number of resnet layers', default=50, choices=[18, 34, 50, 101, 152]) self.parser.add_argument('--dataset', type=str, help='dataset to train on', default='kitti', choices=['kitti', 'kitti_odom', 'kitti_depth', 'kitti_test']) self.parser.add_argument('--png', help='if set, trains from raw KITTI png files (instead of jpgs)', action='store_true') self.parser.add_argument('--height', type=int, help='input image height', default=192) self.parser.add_argument('--width', type=int, help='input image width', default=640) self.parser.add_argument('--alpha_smooth', type=float, help='disparity smoothness weight', default=0.04) self.parser.add_argument('--self_distillation', type=float, help='self_distillation weight', default=0.0) self.parser.add_argument('--gamma_smooth', type=float, help='gamma of smooth loss', default=2) self.parser.add_argument('--alpha_pc', type=float, help='perceptual loss weight', default=0.1) self.parser.add_argument('--disp_min', type=float, help='minimum depth', default=2.0) self.parser.add_argument('--disp_max', type=float, help='maximum depth', default=300.0) self.parser.add_argument('--disp_levels', type=int, help='num levels of disp', default=49) self.parser.add_argument('--disp_layers', type=int, help='num layers of disp', default=2) self.parser.add_argument('--novel_frame_ids', nargs='+', type=int, help='frames to load', default=[]) self.parser.add_argument('--net_type', type=str, help='train which network', default='ResNet', choices=['PladeNet', 'ResNet', 'FalNet']) self.parser.add_argument('--num_ep', type=int, help='train which stage', default=8) self.parser.add_argument('--warp_type', type=str, help='the type of warp', default='disp_warp', choices=['depth_warp', 'disp_warp', 'homography_warp']) self.parser.add_argument('--match_aug', action='store_true', help='if set, use color augmented data to compute loss') self.parser.add_argument('--use_denseaspp', action='store_true', help='use DenseAspp block in ResNet') self.parser.add_argument('--use_mom', action='store_true', help='use mirror occlusion mask') self.parser.add_argument('--flip_right', action='store_true', help='use fliped right image to train') self.parser.add_argument('--pc_net', type=str, help='the type of net to compute pc loss', default='vgg19', choices=['vgg19', 'resnet18']) self.parser.add_argument('--xz_levels', type=int, help='num levels of xz plane', default=14) self.parser.add_argument('--yz_levels', type=int, help='num levels of yz plane', default=0) self.parser.add_argument('--use_mixture_loss', action='store_true', help='use mixture loss') self.parser.add_argument('--alpha_self', type=float, help='perceptual loss weight', default=0.0) self.parser.add_argument('--depth_regression_space', type=str, help='how to compute regression depth', default='inv', choices=['inv', 'exp']) self.parser.add_argument('--render_probability', action='store_true', help='If set, render probability as NeRF') self.parser.add_argument('--plane_residual', action='store_true', help='If set, use residual plane based on init plane') self.parser.add_argument('--no_crop', action='store_true', help='if set, do not use resize crop data aug') self.parser.add_argument('--pe_type', type=str, help='the type of positional embedding', default='neural', choices=['neural', 'frequency']) self.parser.add_argument('--use_colmap', action='store_true', help='if set, use colmap instead of predicting pose by posenet') self.parser.add_argument('--colmap_path', type=str, help='path to the colmap data', default='./kitti_colmap') self.parser.add_argument('--no_stereo', action='store_true', help='if set, disable stereo supervised') self.parser.add_argument('--batch_size', type=int, help='batch size', default=8) self.parser.add_argument('--learning_rate', type=float, help='learning rate', default=0.0001) self.parser.add_argument('--beta_1', type=float, help='beta1 of Adam', default=0.5) self.parser.add_argument('--beta_2', type=float, help='beta2 of Adam', default=0.999) self.parser.add_argument('--num_epochs', type=int, help='number of epochs', default=50) self.parser.add_argument('--start_epoch', type=int, help='number of epochs', default=0) self.parser.add_argument('--milestones', default=[30, 40], nargs='*', help='epochs at which learning rate is divided by 2') self.parser.add_argument('--scheduler_step_size', type=int, help='epochs at which learning rate times 0.1', default=15) self.parser.add_argument('--avg_reprojection', help='if set, uses average reprojection loss', action='store_true') self.parser.add_argument('--automask', help='if set, do auto-masking', action='store_true') self.parser.add_argument('--num_workers', type=int, help='number of dataloader workers', default=12) self.parser.add_argument('--load_weights_folder', type=str, help='name of model to load') self.parser.add_argument('--models_to_load', nargs='+', type=str, help='models to load', default=['encoder', 'depth']) self.parser.add_argument('--stage1_weights_folder', type=str, help='path of teacher model to load') self.parser.add_argument('--log_frequency', type=int, help='number of batches between each tensorboard log', default=500) self.parser.add_argument('--log_img_frequency', type=int, help='number of batches between each tensorboard log', default=250) self.parser.add_argument('--use_ssim', help='if set, use ssim in the loss', action='store_true') self.parser.add_argument('--eval_stereo', help='if set evaluates in stereo mode', action='store_true') self.parser.add_argument('--eval_mono', help='if set evaluates in mono mode', action='store_true') self.parser.add_argument('--disable_median_scaling', help='if set disables median scaling in evaluation', action='store_true') self.parser.add_argument('--pred_depth_scale_factor', help='if set multiplies predictions by this number', type=float, default=1) self.parser.add_argument('--ext_disp_to_eval', type=str, help='optional path to a .npy disparities file to evaluate') self.parser.add_argument('--eval_split', type=str, default='eigen_raw', choices=['eigen_raw', 'eigen_improved', 'eigen_benchmark', 'benchmark', 'odom_9', 'odom_10', 'city'], help='which split to run eval on') self.parser.add_argument('--save_pred_disps', help='if set saves predicted disparities', action='store_true') self.parser.add_argument('--no_eval', help='if set disables evaluation', action='store_true') self.parser.add_argument('--eval_eigen_to_benchmark', help='if set assume we are loading eigen results from npy but we want to evaluate using the new benchmark.', action='store_true') self.parser.add_argument('--eval_out_dir', help='if set will output the disparities to this folder', type=str) self.parser.add_argument('--post_process', help='if set will perform the flipping post processing from the original monodepth paper', action='store_true') def parse(self): self.options = self.parser.parse_args() return self.options
def resolve_ssl_version(candidate): if (candidate is None): return PROTOCOL_SSLv23 if isinstance(candidate, str): res = getattr(ssl, candidate, None) if (res is None): res = getattr(ssl, ('PROTOCOL_' + candidate)) return res return candidate
def get_numeracy_adapter_spec(max_train_instances: int, max_eval_instances: int, dim: int, delimiter: str=', ', **kwargs) -> AdapterSpec: return AdapterSpec(**{**{'method': ADAPT_GENERATION, 'instructions': get_dataset_header(dim, delimiter=delimiter, output_prefix=', '), 'max_train_instances': max_train_instances, 'max_eval_instances': max_eval_instances, 'num_outputs': 1, 'num_train_trials': 1, 'model_deployment': 'openai/davinci', 'temperature': 0, 'stop_sequences': ['\n'], 'max_tokens': 20, 'input_prefix': '', 'output_prefix': ', ', 'instance_prefix': '\n'}, **kwargs})
class FGP_Element(ModuleElement): def __init__(self, parent, x, check=DEBUG): if check: assert (x in parent.V()), (('The argument x=' + str(x)) + ' is not in the covering module!') ModuleElement.__init__(self, parent) self._x = x def lift(self): return self._x def __neg__(self): P = self.parent() return P.element_class(P, (- self._x)) def _add_(self, other): P = self.parent() return P.element_class(P, (self._x + other._x)) def _sub_(self, other): P = self.parent() return P.element_class(P, (self._x - other._x)) def _rmul_(self, c): P = self.parent() return P.element_class(P, self._x._rmul_(c)) def _lmul_(self, s): P = self.parent() return P.element_class(P, self._x._lmul_(s)) def _repr_(self): return repr(self.vector()) def __getitem__(self, *args): return self.vector().__getitem__(*args) def vector(self): try: return self.__vector except AttributeError: self.__vector = self.parent().coordinate_vector(self, reduce=True) self.__vector.set_immutable() return self.__vector def __hash__(self): return hash(self.vector()) def _vector_(self, base_ring=None): v = self.vector() if ((base_ring is None) or (v.base_ring() is base_ring)): return v.__copy__() else: return v.change_ring(base_ring) def _richcmp_(self, right, op): return richcmp(self.vector(), right.vector(), op) def additive_order(self): Q = self.parent() I = Q.invariants() v = self.vector() from sage.rings.infinity import infinity from sage.rings.finite_rings.integer_mod import Mod from sage.rings.integer import Integer from sage.arith.functions import lcm n = Integer(1) for (i, a) in enumerate(I): if (a == 0): if (v[i] != 0): return infinity else: n = lcm(n, Mod(v[i], a).additive_order()) return n
def crop_column(crop_colname: str, df: pd.DataFrame, time_start_colname: str='start_secs', time_end_colname: str='end_secs', max_crop_duration: Optional[float]=None) -> pd.DataFrame: for (i, row) in tqdm(df.iterrows(), desc=f'crop {crop_colname}'): (start, end) = (row[time_start_colname], row[time_end_colname]) if max_crop_duration: end = min(end, (start + max_crop_duration)) cropped = [x for x in row[crop_colname] if ((x['time'] >= start) and (x['time'] <= end))] _ = [x.update({'time': (x['time'] - start)}) for x in cropped] assert (all(((x['time'] >= 0) and (x['time'] <= (end - start)))) for x in cropped) df.at[(i, crop_colname)] = cropped return df
def test_num_4(): array = ak.Array(ak.contents.numpyarray.NumpyArray(np.array([[0.0, 1.1], [2.2, 3.3], [4.4, 5.5]]))) cuda_array = ak.to_backend(array, 'cuda') assert (ak.num(cuda_array, 0) == ak.num(array, 0)) assert (ak.num(cuda_array, 1).tolist() == ak.num(array, 1).tolist())
class SDDivGradTerm(Term): name = 'ev_sd_div_grad' arg_types = ('opt_material', 'parameter_u', 'parameter_w', 'parameter_mv') arg_shapes = [{'opt_material': '1, 1', 'parameter_u': 'D', 'parameter_w': 'D', 'parameter_mv': 'D'}, {'opt_material': None}] function = staticmethod(terms.d_sd_div_grad) def get_fargs(self, mat, par_u, par_w, par_mv, mode=None, term_mode=None, diff_var=None, **kwargs): (vg, _) = self.get_mapping(par_u) grad_u = grad_as_vector(self.get(par_u, 'grad')) grad_w = grad_as_vector(self.get(par_w, 'grad')) div_mv = self.get(par_mv, 'div') grad_mv = grad_as_vector(self.get(par_mv, 'grad')) if (mat is None): mat = nm.ones((1, div_mv.shape[1], 1, 1), dtype=nm.float64) return (grad_u, grad_w, div_mv, grad_mv, mat, vg, get_default(term_mode, 1)) def get_eval_shape(self, mat, par_u, par_w, par_mv, mode=None, term_mode=None, diff_var=None, **kwargs): (n_el, n_qp, dim, n_en, n_c) = self.get_data_shape(par_u) return ((n_el, 1, 1, 1), par_u.dtype)
def _seg_45(): return [(64540, 'M', u''), (64541, 'M', u''), (64542, 'M', u''), (64543, 'M', u''), (64544, 'M', u''), (64545, 'M', u''), (64546, 'M', u''), (64547, 'M', u''), (64548, 'M', u''), (64549, 'M', u''), (64550, 'M', u''), (64551, 'M', u''), (64552, 'M', u''), (64553, 'M', u''), (64554, 'M', u''), (64555, 'M', u''), (64556, 'M', u''), (64557, 'M', u''), (64558, 'M', u''), (64559, 'M', u''), (64560, 'M', u''), (64561, 'M', u''), (64562, 'M', u''), (64563, 'M', u''), (64564, 'M', u''), (64565, 'M', u''), (64566, 'M', u''), (64567, 'M', u''), (64568, 'M', u''), (64569, 'M', u''), (64570, 'M', u''), (64571, 'M', u''), (64572, 'M', u''), (64573, 'M', u''), (64574, 'M', u''), (64575, 'M', u''), (64576, 'M', u''), (64577, 'M', u''), (64578, 'M', u''), (64579, 'M', u''), (64580, 'M', u''), (64581, 'M', u''), (64582, 'M', u''), (64583, 'M', u''), (64584, 'M', u''), (64585, 'M', u''), (64586, 'M', u''), (64587, 'M', u''), (64588, 'M', u''), (64589, 'M', u''), (64590, 'M', u''), (64591, 'M', u''), (64592, 'M', u''), (64593, 'M', u''), (64594, 'M', u''), (64595, 'M', u''), (64596, 'M', u''), (64597, 'M', u''), (64598, 'M', u''), (64599, 'M', u''), (64600, 'M', u''), (64601, 'M', u''), (64602, 'M', u''), (64603, 'M', u''), (64604, 'M', u''), (64605, 'M', u''), (64606, '3', u' '), (64607, '3', u' '), (64608, '3', u' '), (64609, '3', u' '), (64610, '3', u' '), (64611, '3', u' '), (64612, 'M', u''), (64613, 'M', u''), (64614, 'M', u''), (64615, 'M', u''), (64616, 'M', u''), (64617, 'M', u''), (64618, 'M', u''), (64619, 'M', u''), (64620, 'M', u''), (64621, 'M', u''), (64622, 'M', u''), (64623, 'M', u''), (64624, 'M', u''), (64625, 'M', u''), (64626, 'M', u''), (64627, 'M', u''), (64628, 'M', u''), (64629, 'M', u''), (64630, 'M', u''), (64631, 'M', u''), (64632, 'M', u''), (64633, 'M', u''), (64634, 'M', u''), (64635, 'M', u''), (64636, 'M', u''), (64637, 'M', u''), (64638, 'M', u''), (64639, 'M', u'')]
class CommutativeRings(CategoryWithAxiom): class ParentMethods(): def _test_divides(self, **options): tester = self._tester(**options) a = self.an_element() try: a.divides except AttributeError: return z = self.zero() o = self.one() tester.assertTrue(z.divides(z)) tester.assertTrue(o.divides(o)) tester.assertTrue(o.divides(z)) tester.assertIs(z.divides(o), self.is_zero()) if (not self.is_exact()): return for (a, b) in tester.some_elements(repeat=2): try: test = a.divides((a * b)) except NotImplementedError: pass else: tester.assertTrue(test) def over(self, base=None, gen=None, gens=None, name=None, names=None): from sage.rings.ring_extension import RingExtension if (name is not None): if (names is not None): raise ValueError("keyword argument 'name' cannot be combined with 'names'") names = (name,) if (gen is not None): if (gens is not None): raise ValueError("keyword argument 'gen' cannot be combined with 'gens'") gens = (gen,) return RingExtension(self, base, gens, names) class ElementMethods(): pass class Finite(CategoryWithAxiom): class ParentMethods(): def cyclotomic_cosets(self, q, cosets=None): q = self(q) try: (~ q) except ZeroDivisionError: raise ValueError(('%s is not invertible in %s' % (q, self))) if (cosets is None): rest = set(self) else: rest = {self(x) for x in cosets} orbits = [] while rest: x0 = rest.pop() o = [x0] x = (q * x0) while (x != x0): o.append(x) rest.discard(x) x *= q o.sort() orbits.append(o) orbits.sort() return orbits class CartesianProducts(CartesianProductsCategory): def extra_super_categories(self): return [CommutativeRings()]
class Meteor(): def __init__(self): self.meteor_cmd = ['java', '-jar', '-Xmx2G', METEOR_JAR, '-', '-', '-stdio', '-l', 'en', '-norm'] self.meteor_p = subprocess.Popen(self.meteor_cmd, cwd=os.path.dirname(os.path.abspath(__file__)), stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE) self.lock = threading.Lock() def compute_score(self, gts, res): assert (gts.keys() == res.keys()) imgIds = gts.keys() scores = [] eval_line = 'EVAL' self.lock.acquire() for i in imgIds: assert (len(res[i]) == 1) stat = self._stat(res[i][0], gts[i]) eval_line += ' ||| {}'.format(stat) self.meteor_p.stdin.write('{}\n'.format(eval_line).encode()) self.meteor_p.stdin.flush() for i in range(0, len(imgIds)): scores.append(float(self.meteor_p.stdout.readline().strip())) score = float(self.meteor_p.stdout.readline().strip()) self.lock.release() return (score, scores) def method(self): return 'METEOR' def _stat(self, hypothesis_str, reference_list): hypothesis_str = hypothesis_str.replace('|||', '').replace(' ', ' ') score_line = ' ||| '.join(('SCORE', ' ||| '.join(reference_list), hypothesis_str)) self.meteor_p.stdin.write('{}\n'.format(score_line).encode()) self.meteor_p.stdin.flush() return self.meteor_p.stdout.readline().decode().strip() def _score(self, hypothesis_str, reference_list): self.lock.acquire() hypothesis_str = hypothesis_str.replace('|||', '').replace(' ', ' ') score_line = ' ||| '.join(('SCORE', ' ||| '.join(reference_list), hypothesis_str)) self.meteor_p.stdin.write('{}\n'.format(score_line)) stats = self.meteor_p.stdout.readline().strip() eval_line = 'EVAL ||| {}'.format(stats) self.meteor_p.stdin.write('{}\n'.format(eval_line)) score = float(self.meteor_p.stdout.readline().strip()) score = float(self.meteor_p.stdout.readline().strip()) self.lock.release() return score def __del__(self): self.lock.acquire() self.meteor_p.stdin.close() self.meteor_p.kill() self.meteor_p.wait() self.lock.release()
.parametrize('set_loss', [dict(set_loss_nan=False, set_loss_inf=False), dict(set_loss_nan=True, set_loss_inf=False), dict(set_loss_nan=False, set_loss_inf=True)]) def test_check_invalid_loss_hook(set_loss): class DemoModel(nn.Module): def __init__(self, set_loss_nan=False, set_loss_inf=False): super().__init__() self.set_loss_nan = set_loss_nan self.set_loss_inf = set_loss_inf self.linear = nn.Linear(2, 1) def forward(self, x): return self.linear(x) def train_step(self, x, optimizer, **kwargs): if self.set_loss_nan: return dict(loss=torch.tensor(float('nan'))) elif self.set_loss_inf: return dict(loss=torch.tensor(float('inf'))) else: return dict(loss=self(x)) loader = DataLoader(torch.ones((5, 2))) runner = _build_demo_runner() demo_model = DemoModel(**set_loss) runner.model = demo_model runner.register_hook_from_cfg(dict(type='CheckInvalidLossHook', interval=1)) if ((not set_loss['set_loss_nan']) and (not set_loss['set_loss_inf'])): runner.run([loader], [('train', 1)]) else: with pytest.raises(AssertionError): runner.run([loader], [('train', 1)]) shutil.rmtree(runner.work_dir)
.parametrize('method', ('as_requests_kwargs', 'as_werkzeug_kwargs')) def test_serialize_yaml(open_api_3_schema_with_yaml_payload, method): schema = schemathesis.from_dict(open_api_3_schema_with_yaml_payload) (case=schema['/yaml']['POST'].as_strategy()) (max_examples=1) def test(case): kwargs = getattr(case, method)() assert (kwargs['headers']['Content-Type'] == 'text/yaml') assert (kwargs['data'] == '- 42\n') test()
def _apply_split(dataset: ImageFolder, split: List[str]): img_paths = [] for (path, label) in dataset.samples: root_with_slash = os.path.join(dataset.root, '') img_paths.append(path.replace(root_with_slash, '')) split_set = set(split) samples = [] for (path, sample) in zip(img_paths, dataset.samples): if ((len(split) > 0) and ('.jpg' not in split[0]) and (dataset.name == 'ISIC')): path = path.replace('.jpg', '') if (path in split_set): samples.append(sample) dataset.samples = samples dataset.imgs = samples dataset.targets = [s[1] for s in samples]
def get_combination_wise_output_matrix(y, order): return np.array([set((tuple(combination) for combination in it.combinations_with_replacement(get_indicator_representation(row), order))) for row in y])
def compute_score_with_logits(logits, labels): if (labels.shape[0] == 0): scores = torch.zeros(*labels.size()).to(logits.device) return scores logits = torch.max(logits, 1)[1].data one_hots = torch.zeros(*labels.size()).to(logits.device) one_hots.scatter_(1, logits.view((- 1), 1), 1) scores = (one_hots * labels) return scores
def run_cases(cases, run_lambda, set_key): job_arg = [(case, run_lambda, set_key, index, len(cases)) for (index, case) in enumerate(cases) if (not result_exists(set_key, case))] print(f"{(len(cases) - len(job_arg))}/{len(cases)} cases won't be calculated because their results already exist.") jobs = [] pool = Pool(num_pools) for case in job_arg: jobs.append(pool.apply_async(run, args=case, callback=write_result)) pool.close() pool.join()
class EisensteinExtensionRingCappedRelative(EisensteinExtensionGeneric, pAdicCappedRelativeRingGeneric): def __init__(self, exact_modulus, poly, prec, print_mode, shift_seed, names, implementation='NTL'): unram_prec = (((prec + poly.degree()) - 1) // poly.degree()) ntl_poly = ntl_ZZ_pX([a.lift() for a in poly.list()], (poly.base_ring().prime() ** unram_prec)) shift_poly = ntl_ZZ_pX([a.lift() for a in shift_seed.list()], (shift_seed.base_ring().prime() ** unram_prec)) if (unram_prec <= 30): self.prime_pow = PowComputer_ext_maker(poly.base_ring().prime(), unram_prec, unram_prec, prec, False, ntl_poly, 'small', 'e', shift_poly) else: self.prime_pow = PowComputer_ext_maker(poly.base_ring().prime(), 30, unram_prec, prec, False, ntl_poly, 'big', 'e', shift_poly) self._shift_seed = shift_seed self._exact_modulus = exact_modulus self._implementation = implementation EisensteinExtensionGeneric.__init__(self, poly, prec, print_mode, names, pAdicZZpXCRElement)
class PreNormResidual(nn.Module): def __init__(self, dim, fn): super().__init__() self.fn = fn self.norm = nn.LayerNorm(dim) def forward(self, x): return (self.fn(self.norm(x)) + x)
class TrackedLayers(): def __init__(self, *layers_modules): self._layers_dict = {} self._layers_modules = list(layers_modules) self._namespaces = defaultdict(list) self._current_namespace = '/' def track_module(self, layers_module): self._layers_modules.append(layers_module) def get_layers_dict(self): return self._layers_dict def get_layer(self, name): return self._layers_dict[name] def namespace(self, name): old_namespace = self._current_namespace try: self._current_namespace = ((self._current_namespace + name) + '/') (yield) finally: self._current_namespace = old_namespace def get_namespaces(self): return self._namespaces def __getattr__(self, attr): layer_class = None for module in self._layers_modules: if hasattr(module, attr): layer_class = getattr(module, attr) break if (layer_class is None): raise KeyError(attr) def get_layer(*args, **kwargs): layer_name = kwargs['name'] if (not (layer_name in self._layers_dict)): new_layer = layer_class(*args, **kwargs) self._layers_dict[layer_name] = new_layer self._namespaces[self._current_namespace].append(new_layer) return self._layers_dict[layer_name] return get_layer
def calculate_inception_stats(image_path, num_expected=None, seed=0, max_batch_size=64, num_workers=3, prefetch_factor=2, device=torch.device('cuda')): if (dist.get_rank() != 0): torch.distributed.barrier() dist.print0('Loading Inception-v3 model...') detector_url = ' detector_kwargs = dict(return_features=True) feature_dim = 2048 with dnnlib.util.open_url(detector_url, verbose=(dist.get_rank() == 0)) as f: detector_net = pickle.load(f).to(device) dist.print0(f'Loading images from "{image_path}"...') dataset_obj = dataset.ImageFolderDataset(path=image_path, max_size=num_expected, random_seed=seed) if ((num_expected is not None) and (len(dataset_obj) < num_expected)): raise click.ClickException(f'Found {len(dataset_obj)} images, but expected at least {num_expected}') if (len(dataset_obj) < 2): raise click.ClickException(f'Found {len(dataset_obj)} images, but need at least 2 to compute statistics') if (dist.get_rank() == 0): torch.distributed.barrier() num_batches = ((((len(dataset_obj) - 1) // (max_batch_size * dist.get_world_size())) + 1) * dist.get_world_size()) all_batches = torch.arange(len(dataset_obj)).tensor_split(num_batches) rank_batches = all_batches[dist.get_rank()::dist.get_world_size()] data_loader = torch.utils.data.DataLoader(dataset_obj, batch_sampler=rank_batches, num_workers=num_workers, prefetch_factor=prefetch_factor) dist.print0(f'Calculating statistics for {len(dataset_obj)} images...') mu = torch.zeros([feature_dim], dtype=torch.float64, device=device) sigma = torch.zeros([feature_dim, feature_dim], dtype=torch.float64, device=device) for (images, _labels) in tqdm.tqdm(data_loader, unit='batch', disable=(dist.get_rank() != 0)): torch.distributed.barrier() if (images.shape[0] == 0): continue if (images.shape[1] == 1): images = images.repeat([1, 3, 1, 1]) features = detector_net(images.to(device), **detector_kwargs).to(torch.float64) mu += features.sum(0) sigma += (features.T features) torch.distributed.all_reduce(mu) torch.distributed.all_reduce(sigma) mu /= len(dataset_obj) sigma -= (mu.ger(mu) * len(dataset_obj)) sigma /= (len(dataset_obj) - 1) return (mu.cpu().numpy(), sigma.cpu().numpy())
def create_splits_logs(split: str, nusc: 'NuScenes') -> List[str]: scene_splits = create_splits_scenes(verbose=False) assert (split in scene_splits.keys()), 'Requested split {} which is not a known nuScenes split.'.format(split) version = nusc.version if (split in {'train', 'val', 'train_detect', 'train_track'}): assert version.endswith('trainval'), 'Requested split {} which is not compatible with NuScenes version {}'.format(split, version) elif (split in {'mini_train', 'mini_val'}): assert version.endswith('mini'), 'Requested split {} which is not compatible with NuScenes version {}'.format(split, version) elif (split == 'test'): assert version.endswith('test'), 'Requested split {} which is not compatible with NuScenes version {}'.format(split, version) else: raise ValueError('Requested split {} which this function cannot map to logs.'.format(split)) scene_to_log = {scene['name']: nusc.get('log', scene['log_token'])['logfile'] for scene in nusc.scene} logs = set() scenes = scene_splits[split] for scene in scenes: logs.add(scene_to_log[scene]) return list(logs)
class CohereTokenCounter(TokenCounter): def count_tokens(self, request: Request, completions: List[Sequence]) -> int: return sum((len(sequence.tokens) for sequence in completions))
def test_validation_sha_without_split(tmp_path): tmp_no_split_output_dir = (tmp_path / 'pretraining_sha256_no_split') logging.info(f'temporary no split output directory is in {tmp_no_split_output_dir}') input_file = os.path.join(Path.cwd(), 'tests', 'examples', 'pretraining', 'example_pretraining_data.jsonl') num_workers = os.cpu_count() if (num_workers is None): num_workers = 1 args_dict = {'cmd': 'pipeline', 'num_training_splits': 4, 'dev_ratio': None, 'num_dev_splits': None, 'test_ratio': None, 'num_test_splits': None, 'shuffle': 'on_RAM', 'num_workers': num_workers, 'keep_split_jsonls': False, 'input_file_path': input_file, 'output_path': tmp_no_split_output_dir, 'overwrite_output_path': False, 'disable_space_separator': False, 'keep_prompt_only_sequences': False, 'silent': False, 'do_not_balance_hdf5': False, 'log_file_path': None, 'tokenizer_class': None, 'pretrained_tokenizer': None, 'vocab_file': None, 'merges_file': None, 'max_seq_length': 1024, 'input_packing_config': PackingConfig.get_default(), 'packing_boundary': BoundaryType.JSONL.value, 'attention_boundary': BoundaryType.JSONL.value, 'special_tokens_dict': None, 'prompt_keyword': 'prompt', 'completion_keyword': 'completion', 'prompt_prefix': None, 'prompt_postfix': None, 'categories_path': None} args = Namespace(**args_dict) main(args) assert validate_sha256(tmp_no_split_output_dir)
class ModelInfo(): def __init__(self, factory: nn.Module, args: dict, batch_size: int, dataset_args: dict, use_sgd: bool=False, img_size: int=IMG_SIZE): self.factory = factory self.args = args self.batch_size = batch_size self.dataset_args = dataset_args self.img_size = img_size self.use_sgd = use_sgd
def test_likelihood_executable_realign(msa_sampler): input_aln = ['AKDKG-LDINSAEKFFEALHSESIKHQINVMEK-', 'N--EGPLDKESVRTIYELLMSSSHDIQAEQRQRE', 'GQEQN-LDSNYISQVYHTIIEQSVLSQQEFNNRF', 'N--PGPLDDSAIISMFNLIMDGSRILEKKQTNQH', 'GKEKQ-LDPQYVSQIFHTIIEDSVLYQRS-----'] query_name = 'xyz' reference_sequence = f'''>{query_name} {input_aln[(- 1)]} ''' reference_sequence_unaligned = f'''>{query_name} {input_aln[(- 1)].replace('-', '')} ''' aligned_input_handle = StringIO(reference_sequence) unaligned_input_handle = StringIO(reference_sequence_unaligned) msa_string = ('\n'.join([f'''>{n} {s}''' for (n, s) in enumerate(input_aln[:(- 1)])]) + '\n') alignment_handle = StringIO(msa_string) aligned_output_handle = StringIO() likelihood_esm_msa.main(input_h=aligned_input_handle, output_h=aligned_output_handle, masking_off=True, sampler=msa_sampler, reference_msa_handle=alignment_handle, delete_insertions=False, batch_size=1, subset_strategy='in_order', alignment_size=4, unaligned_queries=False) aligned_output_handle.seek(0) unaligned_output_handle = StringIO() alignment_handle.seek(0) likelihood_esm_msa.main(input_h=unaligned_input_handle, output_h=unaligned_output_handle, masking_off=True, sampler=msa_sampler, reference_msa_handle=alignment_handle, delete_insertions=False, batch_size=1, subset_strategy='in_order', alignment_size=4, unaligned_queries=True) unaligned_output_handle.seek(0) (aligned_out_n, aligned_out_v) = aligned_output_handle.readline().split() assert (aligned_out_n == 'id') assert (aligned_out_v == 'esm-msa') (aligned_out_n, aligned_out_v) = aligned_output_handle.readline().split() aligned_out_v = float(aligned_out_v) assert (aligned_out_n == query_name) (unaligned_out_n, unaligned_out_v) = unaligned_output_handle.readline().split() assert (unaligned_out_n == 'id') assert (unaligned_out_v == 'esm-msa') (unaligned_out_n, unaligned_out_v) = unaligned_output_handle.readline().split() unaligned_out_v = float(unaligned_out_v) assert (unaligned_out_n == query_name) assert (unaligned_out_v == pytest.approx(aligned_out_v))
def get_decode_dir_name(ckpt_name): if ('train' in FLAGS.data_path): dataset = 'train' elif ('val' in FLAGS.data_path): dataset = 'val' elif ('test' in FLAGS.data_path): dataset = 'test' else: raise ValueError(('FLAGS.data_path %s should contain one of train, val or test' % FLAGS.data_path)) dirname = ('decode_%s_%imaxenc_%ibeam_%imindec_%imaxdec' % (dataset, FLAGS.max_enc_steps, FLAGS.beam_size, FLAGS.min_dec_steps, FLAGS.max_dec_steps)) if (ckpt_name is not None): dirname += ('_%s' % ckpt_name) return dirname
(numba_geometry_spec) class NumbaRadial1DGeometry(object): def __init__(self, r_inner, r_outer, v_inner, v_outer): self.r_inner = r_inner self.r_outer = r_outer self.v_inner = v_inner self.v_outer = v_outer
def get_third_party(): txt_files = list(Path('./requirements').rglob('*.txt')) package_list = [] for file in txt_files: with open(file, 'r') as fp: for line in fp: line = line.strip() if (line == ''): continue package_list.append(line.split(' ')[0]) return package_list
.overload_method(UnionType, 'append_content') def Union_append_content(builder, tag): if (isinstance(builder, UnionType) and isinstance(tag, numba.types.Integer)): def append_content(builder, tag): content = builder._contents[numba.literally(tag)] builder._tags.append(tag) builder._index.append(len(content)) return content return append_content
class TriggerPool(): def __init__(self): self.triggers = [] self.results = [] def add(self, trigger): self.triggers.append(trigger) def test(self, model, data): untested_triggers = range(len(self.results), len(self.triggers)) for i in untested_triggers: self.results.append(model.test(data, 0.1, self.triggers[i])) def expand(self, num=1): scores = [result.accuracy() for result in self.results] best_trigger = self.triggers[scores.index(max(scores))] def spawn(trigger): return trigger.duplicate().add_noise(type_='Gaussian', args={'std': 0.1}) for i in range(num): self.add(spawn(best_trigger)) def success_triggers(self, threshold=90): return [self.triggers[i] for i in range(len(self.results)) if (self.results[i].accuracy() >= threshold)]
def transform_proposals_seg(dataset_dict, image_shape, transforms, *, proposal_topk, min_box_size=0): boxes = dataset_dict['proposals'].proposal_boxes.tensor.cpu().numpy() boxes = transforms.apply_box(boxes) boxes = Boxes(boxes) objectness_logits = dataset_dict['proposals'].objectness_logits oh_labels = dataset_dict['proposals'].oh_labels superpixels = dataset_dict['superpixels'].cpu().numpy() boxes.clip(image_shape) keep = boxes.nonempty(threshold=min_box_size) boxes = boxes[keep] objectness_logits = objectness_logits[keep] oh_labels = oh_labels[keep] proposals = Instances(image_shape) proposals.proposal_boxes = boxes[:proposal_topk] proposals.objectness_logits = objectness_logits[:proposal_topk] proposals.oh_labels = oh_labels[:proposal_topk] dataset_dict['proposals'] = proposals superpixels = transforms.apply_segmentation(superpixels.astype('float32')) dataset_dict['superpixels'] = torch.as_tensor(np.ascontiguousarray(superpixels.astype('int32')))
class Fpr(Critic): def __init__(self, recall_level=0.95): super().__init__() self.recall_level = recall_level def get_name(self): return (('FPR(' + str((self.recall_level * 100))) + ')') def stable_cumsum(self, arr, rtol=1e-05, atol=1e-08): out = np.cumsum(arr, dtype=np.float64) expected = np.sum(arr, dtype=np.float64) if (not np.allclose(out[(- 1)], expected, rtol=rtol, atol=atol)): raise RuntimeError('cumsum was found to be unstable: its last element does not correspond to sum') return out def fpr_and_fdr_at_recall(self, y_true, y_score, recall_level, pos_label=None): classes = np.unique(y_true) if ((pos_label is None) and (not (np.array_equal(classes, [0, 1]) or np.array_equal(classes, [(- 1), 1]) or np.array_equal(classes, [0]) or np.array_equal(classes, [(- 1)]) or np.array_equal(classes, [1])))): raise ValueError('Data is not binary and pos_label is not specified') elif (pos_label is None): pos_label = 1.0 y_true = (y_true == pos_label) desc_score_indices = np.argsort(y_score, kind='mergesort')[::(- 1)] y_score = y_score[desc_score_indices] y_true = y_true[desc_score_indices] distinct_value_indices = np.where(np.diff(y_score))[0] threshold_idxs = np.r_[(distinct_value_indices, (y_true.size - 1))] tps = self.stable_cumsum(y_true)[threshold_idxs] fps = ((1 + threshold_idxs) - tps) thresholds = y_score[threshold_idxs] recall = (tps / tps[(- 1)]) last_ind = tps.searchsorted(tps[(- 1)]) sl = slice(last_ind, None, (- 1)) (recall, fps, tps, thresholds) = (np.r_[(recall[sl], 1)], np.r_[(fps[sl], 0)], np.r_[(tps[sl], 0)], thresholds[sl]) cutoff = np.argmin(np.abs((recall - recall_level))) return (fps[cutoff] / np.sum(np.logical_not(y_true))) def evaluate(self, normal_scores, anomaly_scores): all_scores = (normal_scores + anomaly_scores) all_labels = ([1 for _ in range(len(normal_scores))] + [0 for _ in range(len(anomaly_scores))]) return self.fpr_and_fdr_at_recall(np.array(all_labels), np.array(all_scores), self.recall_level)
class cuFFTPlanCache(object): def __init__(self, device_index): self.device_index = device_index size = cuFFTPlanCacheAttrContextProp(torch._cufft_get_plan_cache_size, '.size is a read-only property showing the number of plans currently in the cache. To change the cache capacity, set cufft_plan_cache.max_size.') max_size = cuFFTPlanCacheAttrContextProp(torch._cufft_get_plan_cache_max_size, torch._cufft_set_plan_cache_max_size) def clear(self): return torch._cufft_clear_plan_cache(self.device_index)
def convert_fsmt_checkpoint_to_pytorch(fsmt_checkpoint_path, pytorch_dump_folder_path): assert os.path.exists(fsmt_checkpoint_path) os.makedirs(pytorch_dump_folder_path, exist_ok=True) print(f'Writing results to {pytorch_dump_folder_path}') checkpoint_file = basename(fsmt_checkpoint_path) fsmt_folder_path = dirname(fsmt_checkpoint_path) cls = fairseq.model_parallel.models.transformer.ModelParallelTransformerModel models = cls.hub_models() kwargs = {'bpe': 'fastbpe', 'tokenizer': 'moses'} data_name_or_path = '' print(f'using checkpoint {checkpoint_file}') chkpt = hub_utils.from_pretrained(fsmt_folder_path, checkpoint_file, data_name_or_path, archive_map=models, **kwargs) args = dict(vars(chkpt['args'])) src_lang = args['source_lang'] tgt_lang = args['target_lang'] data_root = dirname(pytorch_dump_folder_path) model_dir = basename(pytorch_dump_folder_path) src_dict_file = os.path.join(fsmt_folder_path, f'dict.{src_lang}.txt') tgt_dict_file = os.path.join(fsmt_folder_path, f'dict.{tgt_lang}.txt') src_dict = Dictionary.load(src_dict_file) src_vocab = rewrite_dict_keys(src_dict.indices) src_vocab_size = len(src_vocab) src_vocab_file = os.path.join(pytorch_dump_folder_path, 'vocab-src.json') print(f'Generating {src_vocab_file}') with open(src_vocab_file, 'w', encoding='utf-8') as f: f.write(json.dumps(src_vocab, ensure_ascii=False, indent=json_indent)) tgt_dict = Dictionary.load(tgt_dict_file) tgt_vocab = rewrite_dict_keys(tgt_dict.indices) tgt_vocab_size = len(tgt_vocab) tgt_vocab_file = os.path.join(pytorch_dump_folder_path, 'vocab-tgt.json') print(f'Generating {tgt_vocab_file}') with open(tgt_vocab_file, 'w', encoding='utf-8') as f: f.write(json.dumps(tgt_vocab, ensure_ascii=False, indent=json_indent)) merges_file = os.path.join(pytorch_dump_folder_path, VOCAB_FILES_NAMES['merges_file']) fsmt_merges_file = os.path.join(fsmt_folder_path, 'bpecodes') with open(fsmt_merges_file, encoding='utf-8') as fin: merges = fin.read() merges = re.sub(' \\d+$', '', merges, 0, re.M) print(f'Generating {merges_file}') with open(merges_file, 'w', encoding='utf-8') as fout: fout.write(merges) fsmt_model_config_file = os.path.join(pytorch_dump_folder_path, 'config.json') assert (args['bpe'] == 'fastbpe'), f"need to extend tokenizer to support bpe={args['bpe']}" assert (args['tokenizer'] == 'moses'), f"need to extend tokenizer to support bpe={args['tokenizer']}" model_conf = {'architectures': ['FSMTForConditionalGeneration'], 'model_type': 'fsmt', 'activation_dropout': args['activation_dropout'], 'activation_function': 'relu', 'attention_dropout': args['attention_dropout'], 'd_model': args['decoder_embed_dim'], 'dropout': args['dropout'], 'init_std': 0.02, 'max_position_embeddings': args['max_source_positions'], 'num_hidden_layers': args['encoder_layers'], 'src_vocab_size': src_vocab_size, 'tgt_vocab_size': tgt_vocab_size, 'langs': [src_lang, tgt_lang], 'encoder_attention_heads': args['encoder_attention_heads'], 'encoder_ffn_dim': args['encoder_ffn_embed_dim'], 'encoder_layerdrop': args['encoder_layerdrop'], 'encoder_layers': args['encoder_layers'], 'decoder_attention_heads': args['decoder_attention_heads'], 'decoder_ffn_dim': args['decoder_ffn_embed_dim'], 'decoder_layerdrop': args['decoder_layerdrop'], 'decoder_layers': args['decoder_layers'], 'bos_token_id': 0, 'pad_token_id': 1, 'eos_token_id': 2, 'is_encoder_decoder': True, 'scale_embedding': (not args['no_scale_embedding']), 'tie_word_embeddings': args['share_all_embeddings']} model_conf['num_beams'] = 5 model_conf['early_stopping'] = False if ((model_dir in best_score_hparams) and ('length_penalty' in best_score_hparams[model_dir])): model_conf['length_penalty'] = best_score_hparams[model_dir]['length_penalty'] else: model_conf['length_penalty'] = 1.0 print(f'Generating {fsmt_model_config_file}') with open(fsmt_model_config_file, 'w', encoding='utf-8') as f: f.write(json.dumps(model_conf, ensure_ascii=False, indent=json_indent)) fsmt_tokenizer_config_file = os.path.join(pytorch_dump_folder_path, TOKENIZER_CONFIG_FILE) tokenizer_conf = {'langs': [src_lang, tgt_lang], 'model_max_length': 1024} print(f'Generating {fsmt_tokenizer_config_file}') with open(fsmt_tokenizer_config_file, 'w', encoding='utf-8') as f: f.write(json.dumps(tokenizer_conf, ensure_ascii=False, indent=json_indent)) model = chkpt['models'][0] model_state_dict = model.state_dict() model_state_dict = OrderedDict(((('model.' + k), v) for (k, v) in model_state_dict.items())) ignore_keys = ['model.model', 'model.encoder.version', 'model.decoder.version', 'model.encoder_embed_tokens.weight', 'model.decoder_embed_tokens.weight', 'model.encoder.embed_positions._float_tensor', 'model.decoder.embed_positions._float_tensor'] for k in ignore_keys: model_state_dict.pop(k, None) config = FSMTConfig.from_pretrained(pytorch_dump_folder_path) model_new = FSMTForConditionalGeneration(config) model_new.load_state_dict(model_state_dict, strict=False) pytorch_weights_dump_path = os.path.join(pytorch_dump_folder_path, WEIGHTS_NAME) print(f'Generating {pytorch_weights_dump_path}') torch.save(model_state_dict, pytorch_weights_dump_path) print('Conversion is done!') print('\nLast step is to upload the files to s3') print(f'cd {data_root}') print(f'transformers-cli upload {model_dir}') print('Note: CDN caches files for up to 24h, so either use a local model path or use `from_pretrained(mname, use_cdn=False)` to use the non-cached version.')
class MMFDatasetBuilder(BaseDatasetBuilder): ZOO_CONFIG_PATH = None ZOO_VARIATION = None def __init__(self, dataset_name, dataset_class=None, zoo_variation='defaults', *args, **kwargs): super().__init__(dataset_name) self.dataset_class = dataset_class self.zoo_type = 'datasets' self.zoo_variation = zoo_variation def dataset_class(self): return self._dataset_class _class.setter def dataset_class(self, dataset_class): self._dataset_class = dataset_class def zoo_variation(self): return self._zoo_variation _variation.setter def zoo_variation(self, zoo_variation): self._zoo_variation = zoo_variation def zoo_config_path(self): if (self.ZOO_CONFIG_PATH is None): self.ZOO_CONFIG_PATH = get_global_config('env.dataset_zoo') return self.ZOO_CONFIG_PATH _config_path.setter def zoo_config_path(self, zoo_config_path): self.ZOO_CONFIG_PATH = zoo_config_path def set_dataset_class(self, dataset_cls): self.dataset_class = dataset_cls def build(self, config, dataset_type='train', *args, **kwargs): self.config = config requirements = config.get('zoo_requirements', []) if (len(requirements) == 0): self._download_requirement(config, self.dataset_name, self.zoo_variation) else: for requirement in requirements: self._download_requirement(config, requirement) def _download_requirement(self, config, requirement_key, requirement_variation='defaults'): (version, resources) = get_zoo_config(requirement_key, requirement_variation, self.zoo_config_path, self.zoo_type) if (resources is None): return requirement_split = requirement_key.split('.') dataset_name = requirement_split[0] if (len(requirement_split) >= 2): dataset_variation = requirement_split[1] else: dataset_variation = requirement_variation download_path = os.path.join(get_mmf_env('data_dir'), 'datasets', dataset_name, dataset_variation) download_path = get_absolute_path(download_path) if (not isinstance(resources, collections.abc.Mapping)): self._download_resources(resources, download_path, version) else: use_features = config.get('use_features', False) use_images = config.get('use_images', False) if use_features: self._download_based_on_attribute(resources, download_path, version, 'features') if use_images: self._download_based_on_attribute(resources, download_path, version, 'images') self._download_based_on_attribute(resources, download_path, version, 'annotations') self._download_resources(resources.get('extras', []), download_path, version) def load(self, config, dataset_type, *args, **kwargs): self.config = config split_dataset_from_train = self.config.get('split_train', False) if split_dataset_from_train: config = self._modify_dataset_config_for_split(config) annotations = self._read_annotations(config, dataset_type) if (annotations is None): return None datasets = [] for imdb_idx in range(len(annotations)): dataset_class = self.dataset_class dataset = dataset_class(config, dataset_type, imdb_idx) datasets.append(dataset) dataset = MMFConcatDataset(datasets) if split_dataset_from_train: dataset = self._split_dataset_from_train(dataset, dataset_type) self.dataset = dataset return self.dataset def _split_dataset_from_train(self, dataset, dataset_type): if ((dataset_type in self.config.split_train.keys()) or (dataset_type == 'train')): (start, end) = self._calculate_split_for_dataset_type(dataset_type) dataset_length = len(dataset) (start, end) = (round((start * dataset_length)), round((end * dataset_length))) if (start > end): raise ValueError(f'Train split ratio for {dataset_type} must be positive.') indices = self._generate_permuted_indexes(dataset_length)[start:end] dataset = MMFSubset(dataset, indices) print(f'Dataset type: {dataset_type} length: {len(dataset)} total: {dataset_length}') return dataset def _generate_permuted_indexes(self, dataset_length): generator = torch.Generator() generator.manual_seed(self.config.get('split_train.seed', 123456)) return torch.randperm(dataset_length, generator=generator) def _modify_dataset_config_for_split(self, config): with open_dict(config): for data_type in config.split_train: if (data_type == 'seed'): continue if config.use_images: config.images[data_type] = deepcopy(config.images.train) if config.use_features: config.features[data_type] = deepcopy(config.features.train) config.annotations[data_type] = deepcopy(config.annotations.train) return config def _read_annotations(self, config, dataset_type): annotations = config.get('annotations', {}).get(dataset_type, []) if isinstance(annotations, str): annotations = [annotations] if (len(annotations) == 0): warnings.warn(((('Dataset type {} is not present or empty in ' + 'annotations of dataset config or either annotations ') + 'key is not present. Returning None. ') + "This dataset won't be used.".format(dataset_type))) return None return annotations def _calculate_split_for_dataset_type(self, dataset_type): start = 0.0 for data_type in self.config.split_train: if (data_type == 'seed'): continue if (dataset_type == data_type): return (start, (start + self.config.split_train[data_type])) start += self.config.split_train[data_type] if (start > 1.0): raise ValueError(('Ratios of val plus test should not exceed 100%.' + ' Need to leave some percentage for training.')) elif (start == 1.0): warnings.warn('All data in training set is used for val and/or test.') if (dataset_type == 'train'): return (start, 1.0) def _download_based_on_attribute(self, resources, download_path, version, attribute): path = os.path.join(download_path, attribute) self._download_resources(resources.get(attribute, []), path, version) def _download_resources(self, resources, path, version): download.download_resources(resources, path, version)
class MPNetTokenizerFast(PreTrainedTokenizerFast): vocab_files_names = VOCAB_FILES_NAMES pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP pretrained_init_configuration = PRETRAINED_INIT_CONFIGURATION max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES slow_tokenizer_class = MPNetTokenizer model_input_names = ['input_ids', 'attention_mask'] def __init__(self, vocab_file, tokenizer_file=None, do_lower_case=True, bos_token='<s>', eos_token='</s>', sep_token='</s>', cls_token='<s>', unk_token='[UNK]', pad_token='<pad>', mask_token='<mask>', tokenize_chinese_chars=True, strip_accents=None, **kwargs): super().__init__(vocab_file, tokenizer_file=tokenizer_file, do_lower_case=do_lower_case, bos_token=bos_token, eos_token=eos_token, sep_token=sep_token, cls_token=cls_token, unk_token=unk_token, pad_token=pad_token, mask_token=mask_token, tokenize_chinese_chars=tokenize_chinese_chars, strip_accents=strip_accents, **kwargs) pre_tok_state = json.loads(self.backend_tokenizer.normalizer.__getstate__()) if ((pre_tok_state.get('do_lower_case', do_lower_case) != do_lower_case) or (pre_tok_state.get('strip_accents', strip_accents) != strip_accents)): pre_tok_class = getattr(normalizers, pre_tok_state.pop('type')) pre_tok_state['do_lower_case'] = do_lower_case pre_tok_state['strip_accents'] = strip_accents self.backend_tokenizer.normalizer = pre_tok_class(**pre_tok_state) self.do_lower_case = do_lower_case def mask_token(self) -> str: if ((self._mask_token is None) and self.verbose): logger.error('Using mask_token, but it is not set yet.') return None return str(self._mask_token) _token.setter def mask_token(self, value): value = (AddedToken(value, lstrip=True, rstrip=False) if isinstance(value, str) else value) self._mask_token = value def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None): output = (([self.bos_token_id] + token_ids_0) + [self.eos_token_id]) if (token_ids_1 is None): return output return (((output + [self.eos_token_id]) + token_ids_1) + [self.eos_token_id]) def create_token_type_ids_from_sequences(self, token_ids_0: List[int], token_ids_1: Optional[List[int]]=None) -> List[int]: sep = [self.sep_token_id] cls = [self.cls_token_id] if (token_ids_1 is None): return (len(((cls + token_ids_0) + sep)) * [0]) return (len((((((cls + token_ids_0) + sep) + sep) + token_ids_1) + sep)) * [0]) def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str]=None) -> Tuple[str]: files = self._tokenizer.model.save(save_directory, name=filename_prefix) return tuple(files)
def main(): args = get_args() for class_no in os.listdir(args.feature_dir): print('Class index ', class_no) compute_mean_vector(class_no, args.save_path, args.feature_dir)
class AlignTextConfig(PretrainedConfig): model_type = 'align_text_model' def __init__(self, vocab_size=30522, hidden_size=768, num_hidden_layers=12, num_attention_heads=12, intermediate_size=3072, hidden_act='gelu', hidden_dropout_prob=0.1, attention_probs_dropout_prob=0.1, max_position_embeddings=512, type_vocab_size=2, initializer_range=0.02, layer_norm_eps=1e-12, pad_token_id=0, position_embedding_type='absolute', use_cache=True, **kwargs): super().__init__(**kwargs) self.vocab_size = vocab_size self.hidden_size = hidden_size self.num_hidden_layers = num_hidden_layers self.num_attention_heads = num_attention_heads self.hidden_act = hidden_act self.intermediate_size = intermediate_size self.hidden_dropout_prob = hidden_dropout_prob self.attention_probs_dropout_prob = attention_probs_dropout_prob self.max_position_embeddings = max_position_embeddings self.type_vocab_size = type_vocab_size self.initializer_range = initializer_range self.layer_norm_eps = layer_norm_eps self.position_embedding_type = position_embedding_type self.use_cache = use_cache self.pad_token_id = pad_token_id def from_pretrained(cls, pretrained_model_name_or_path: Union[(str, os.PathLike)], **kwargs) -> 'PretrainedConfig': (config_dict, kwargs) = cls.get_config_dict(pretrained_model_name_or_path, **kwargs) if (config_dict.get('model_type') == 'align'): config_dict = config_dict['text_config'] if (('model_type' in config_dict) and hasattr(cls, 'model_type') and (config_dict['model_type'] != cls.model_type)): logger.warning(f"You are using a model of type {config_dict['model_type']} to instantiate a model of type {cls.model_type}. This is not supported for all configurations of models and can yield errors.") return cls.from_dict(config_dict, **kwargs)
def strip_artist(s): s = s.lower() s = s.replace('the ', '') keys = [' - ', '/', ' ft', 'feat', 'featuring', ' and ', ' with ', '_', ' vs', '&', ';', '+'] for key in keys: loc = s.find(key) if (loc != (- 1)): s = s[:loc] return s
class HessianProblem(): def __init__(self, db: database.Database, form_handler: _forms.ControlFormHandler, adjoint_form_handler: _forms.AdjointFormHandler, gradient_problem: control_gradient_problem.ControlGradientProblem, box_constraints: boxc.BoxConstraints) -> None: self.db = db self.form_handler = form_handler self.adjoint_form_handler = adjoint_form_handler self.gradient_problem = gradient_problem self.box_constraints = box_constraints self.config = self.db.config self.inner_newton = self.config.get('AlgoTNM', 'inner_newton') self.max_it_inner_newton = self.config.getint('AlgoTNM', 'max_it_inner_newton') self.inner_newton_rtol = self.config.getfloat('AlgoTNM', 'inner_newton_rtol') self.inner_newton_atol = self.config.getfloat('AlgoTNM', 'inner_newton_atol') self.test_directions = self.form_handler.hessian_form_handler.test_directions self.residual = _utils.create_function_list(self.db.function_db.control_spaces) self.delta_control = _utils.create_function_list(self.db.function_db.control_spaces) self.temp1 = _utils.create_function_list(self.db.function_db.control_spaces) self.temp2 = _utils.create_function_list(self.db.function_db.control_spaces) self.p = _utils.create_function_list(self.db.function_db.control_spaces) self.p_prev = _utils.create_function_list(self.db.function_db.control_spaces) self.p_pprev = _utils.create_function_list(self.db.function_db.control_spaces) self.s = _utils.create_function_list(self.db.function_db.control_spaces) self.s_prev = _utils.create_function_list(self.db.function_db.control_spaces) self.s_pprev = _utils.create_function_list(self.db.function_db.control_spaces) self.q = _utils.create_function_list(self.db.function_db.control_spaces) self.q_prev = _utils.create_function_list(self.db.function_db.control_spaces) self.hessian_actions = _utils.create_function_list(self.db.function_db.control_spaces) self.inactive_part = _utils.create_function_list(self.db.function_db.control_spaces) self.active_part = _utils.create_function_list(self.db.function_db.control_spaces) self.state_A_tensors = [fenics.PETScMatrix() for _ in range(self.db.parameter_db.state_dim)] self.state_b_tensors = [fenics.PETScVector() for _ in range(self.db.parameter_db.state_dim)] self.adjoint_A_tensors = [fenics.PETScMatrix() for _ in range(self.db.parameter_db.state_dim)] self.adjoint_b_tensors = [fenics.PETScVector() for _ in range(self.db.parameter_db.state_dim)] self.state_dim = self.db.parameter_db.state_dim self.picard_rtol = self.config.getfloat('StateSystem', 'picard_rtol') self.picard_atol = self.config.getfloat('StateSystem', 'picard_atol') self.picard_max_iter = self.config.getint('StateSystem', 'picard_iter') self.picard_verbose = self.config.getboolean('StateSystem', 'picard_verbose') self.no_sensitivity_solves = 0 self.bcs_list_ad = self.adjoint_form_handler.bcs_list_ad option: _typing.KspOption = {'ksp_type': 'cg', 'pc_type': 'hypre', 'pc_hypre_type': 'boomeramg', 'pc_hypre_boomeramg_strong_threshold': 0.7, 'ksp_rtol': 1e-16, 'ksp_atol': 1e-50, 'ksp_max_it': 100} self.riesz_ksp_options: List[_typing.KspOption] = [] for _ in range(len(self.db.function_db.controls)): self.riesz_ksp_options.append(option) def hessian_application(self, h: List[fenics.Function], out: List[fenics.Function]) -> None: for i in range(len(self.test_directions)): self.test_directions[i].vector().vec().aypx(0.0, h[i].vector().vec()) self.test_directions[i].vector().apply('') self.bcs_list_ad = self.adjoint_form_handler.bcs_list_ad if ((not self.config.getboolean('StateSystem', 'picard_iteration')) or (self.state_dim == 1)): for i in range(self.state_dim): _utils.assemble_and_solve_linear(self.form_handler.hessian_form_handler.sensitivity_eqs_lhs[i], self.form_handler.hessian_form_handler.sensitivity_eqs_rhs[i], self.bcs_list_ad[i], fun=self.db.function_db.states_prime[i], ksp_options=self.db.parameter_db.state_ksp_options[i], comm=self.db.geometry_db.mpi_comm, preconditioner_form=self.db.form_db.preconditioner_forms[i]) for i in range(self.state_dim): _utils.assemble_and_solve_linear(self.form_handler.hessian_form_handler.adjoint_sensitivity_eqs_lhs[((- 1) - i)], self.form_handler.hessian_form_handler.w_1[((- 1) - i)], self.bcs_list_ad[((- 1) - i)], fun=self.db.function_db.adjoints_prime[((- 1) - i)], ksp_options=self.db.parameter_db.adjoint_ksp_options[((- 1) - i)], comm=self.db.geometry_db.mpi_comm, preconditioner_form=self.db.form_db.preconditioner_forms[((- 1) - i)]) else: nonlinear_solvers.picard_iteration(self.form_handler.hessian_form_handler.sensitivity_eqs_picard, self.db.function_db.states_prime, self.adjoint_form_handler.bcs_list_ad, max_iter=self.picard_max_iter, rtol=self.picard_rtol, atol=self.picard_atol, verbose=self.picard_verbose, inner_damped=False, inner_inexact=False, inner_verbose=False, inner_max_iter=2, ksp_options=self.db.parameter_db.state_ksp_options, A_tensors=self.state_A_tensors, b_tensors=self.state_b_tensors, inner_is_linear=True, preconditioner_forms=self.db.form_db.preconditioner_forms) nonlinear_solvers.picard_iteration(self.form_handler.hessian_form_handler.adjoint_sensitivity_eqs_picard, self.db.function_db.adjoints_prime, self.adjoint_form_handler.bcs_list_ad, max_iter=self.picard_max_iter, rtol=self.picard_rtol, atol=self.picard_atol, verbose=self.picard_verbose, inner_damped=False, inner_inexact=False, inner_verbose=False, inner_max_iter=2, ksp_options=self.db.parameter_db.adjoint_ksp_options, A_tensors=self.adjoint_A_tensors, b_tensors=self.adjoint_b_tensors, inner_is_linear=True, preconditioner_forms=self.db.form_db.preconditioner_forms) for i in range(len(out)): b = fenics.as_backend_type(fenics.assemble(self.form_handler.hessian_form_handler.hessian_rhs[i])).vec() _utils.solve_linear_problem(A=self.form_handler.riesz_projection_matrices[i], b=b, fun=out[i], ksp_options=self.riesz_ksp_options[i], comm=self.db.geometry_db.mpi_comm) self.no_sensitivity_solves += 2 def reduced_hessian_application(self, h: List[fenics.Function], out: List[fenics.Function]) -> None: for j in range(len(out)): out[j].vector().vec().set(0.0) out[j].vector().apply('') self.box_constraints.restrictor.restrict_to_inactive_set(h, self.inactive_part) self.hessian_application(self.inactive_part, self.hessian_actions) self.box_constraints.restrictor.restrict_to_inactive_set(self.hessian_actions, self.inactive_part) self.box_constraints.restrictor.restrict_to_active_set(h, self.active_part) for j in range(len(out)): out[j].vector().vec().aypx(0.0, (self.active_part[j].vector().vec() + self.inactive_part[j].vector().vec())) out[j].vector().apply('') def newton_solve(self) -> List[fenics.Function]: self.gradient_problem.solve() self.box_constraints.restrictor.compute_active_sets() for fun in self.delta_control: fun.vector().vec().set(0.0) fun.vector().apply('') if (self.inner_newton.casefold() == 'cg'): self.cg() elif (self.inner_newton.casefold() == 'cr'): self.cr() return self.delta_control def cg(self) -> None: for j in range(len(self.residual)): self.residual[j].vector().vec().aypx(0.0, (- self.db.function_db.gradient[j].vector().vec())) self.residual[j].vector().apply('') self.p[j].vector().vec().aypx(0.0, self.residual[j].vector().vec()) self.p[j].vector().apply('') rsold = self.form_handler.scalar_product(self.residual, self.residual) eps_0 = np.sqrt(rsold) for _ in range(self.max_it_inner_newton): self.reduced_hessian_application(self.p, self.q) self.box_constraints.restrictor.restrict_to_active_set(self.p, self.temp1) sp_val1 = self.form_handler.scalar_product(self.temp1, self.temp1) self.box_constraints.restrictor.restrict_to_inactive_set(self.p, self.temp1) self.box_constraints.restrictor.restrict_to_inactive_set(self.q, self.temp2) sp_val2 = self.form_handler.scalar_product(self.temp1, self.temp2) sp_val = (sp_val1 + sp_val2) alpha = (rsold / sp_val) for j in range(len(self.delta_control)): self.delta_control[j].vector().vec().axpy(alpha, self.p[j].vector().vec()) self.delta_control[j].vector().apply('') self.residual[j].vector().vec().axpy((- alpha), self.q[j].vector().vec()) self.residual[j].vector().apply('') rsnew = self.form_handler.scalar_product(self.residual, self.residual) eps = np.sqrt(rsnew) _loggers.debug(f'Residual of the CG method: {(eps / eps_0):.3e} (relative)') if (eps < (self.inner_newton_atol + (self.inner_newton_rtol * eps_0))): break beta = (rsnew / rsold) for j in range(len(self.p)): self.p[j].vector().vec().aypx(beta, self.residual[j].vector().vec()) self.p[j].vector().apply('') rsold = rsnew def cr(self) -> None: for j in range(len(self.residual)): self.residual[j].vector().vec().aypx(0.0, (- self.db.function_db.gradient[j].vector().vec())) self.residual[j].vector().apply('') self.p[j].vector().vec().aypx(0.0, self.residual[j].vector().vec()) self.p[j].vector().apply('') eps_0 = np.sqrt(self.form_handler.scalar_product(self.residual, self.residual)) self.reduced_hessian_application(self.residual, self.s) for j in range(len(self.q)): self.q[j].vector().vec().aypx(0.0, self.s[j].vector().vec()) self.q[j].vector().apply('') self.box_constraints.restrictor.restrict_to_active_set(self.residual, self.temp1) self.box_constraints.restrictor.restrict_to_active_set(self.s, self.temp2) sp_val1 = self.form_handler.scalar_product(self.temp1, self.temp2) self.box_constraints.restrictor.restrict_to_inactive_set(self.residual, self.temp1) self.box_constraints.restrictor.restrict_to_inactive_set(self.s, self.temp2) sp_val2 = self.form_handler.scalar_product(self.temp1, self.temp2) rar = (sp_val1 + sp_val2) for i in range(self.max_it_inner_newton): self.box_constraints.restrictor.restrict_to_active_set(self.q, self.temp1) self.box_constraints.restrictor.restrict_to_inactive_set(self.q, self.temp2) denom1 = self.form_handler.scalar_product(self.temp1, self.temp1) denom2 = self.form_handler.scalar_product(self.temp2, self.temp2) denominator = (denom1 + denom2) alpha = (rar / denominator) for j in range(len(self.delta_control)): self.delta_control[j].vector().vec().axpy(alpha, self.p[j].vector().vec()) self.delta_control[j].vector().apply('') self.residual[j].vector().vec().axpy((- alpha), self.q[j].vector().vec()) self.residual[j].vector().apply('') eps = np.sqrt(self.form_handler.scalar_product(self.residual, self.residual)) _loggers.debug(f'Residual of the CR method: {(eps / eps_0):.3e} (relative)') if ((eps < (self.inner_newton_atol + (self.inner_newton_rtol * eps_0))) or (i == (self.max_it_inner_newton - 1))): break self.reduced_hessian_application(self.residual, self.s) self.box_constraints.restrictor.restrict_to_active_set(self.residual, self.temp1) self.box_constraints.restrictor.restrict_to_active_set(self.s, self.temp2) sp_val1 = self.form_handler.scalar_product(self.temp1, self.temp2) self.box_constraints.restrictor.restrict_to_inactive_set(self.residual, self.temp1) self.box_constraints.restrictor.restrict_to_inactive_set(self.s, self.temp2) sp_val2 = self.form_handler.scalar_product(self.temp1, self.temp2) rar_new = (sp_val1 + sp_val2) beta = (rar_new / rar) for j in range(len(self.p)): self.p[j].vector().vec().aypx(beta, self.residual[j].vector().vec()) self.p[j].vector().apply('') self.q[j].vector().vec().aypx(beta, self.s[j].vector().vec()) self.q[j].vector().apply('') rar = rar_new
def count_letter_ngram(sentence, n=3): if (len(sentence) < n): return set(sentence) local_counts = set() for k in range(((len(sentence.strip()) - n) + 1)): local_counts.add(sentence[k:(k + n)]) return local_counts
class network_29layers_Custom(network_29layers): def forward(self, x, nrm=True): x = self.conv1(x) x = self.pool1(x) x = self.block1(x) x = self.group1(x) x = self.pool2(x) x = self.block2(x) x = self.group2(x) x = self.pool3(x) x = self.block3(x) x = self.group3(x) x = self.block4(x) x = self.group4(x) x = self.pool4(x) x = x.view(x.size(0), (- 1)) fc = self.fc(x) fc = F.dropout(fc, training=self.training) if (nrm is False): return fc xnorm = F.normalize(fc, p=2, dim=1) return xnorm
def length(node, indices): return expr.Expr(impl.get_runtime().compiling_callable.ast_builder().expr_snode_length(node._snode.ptr, expr.make_expr_group(indices)), dbg_info=_ti_core.DebugInfo(impl.get_runtime().get_current_src_info()))
def GenerateSM90_TensorOp_1684_complex_gaussian(manifest, cuda_version): if (not CudaToolkitVersionSatisfies(cuda_version, 11, 8)): return layouts = [(LayoutType.ColumnMajor, LayoutType.ColumnMajor, LayoutType.ColumnMajor), (LayoutType.ColumnMajor, LayoutType.RowMajor, LayoutType.ColumnMajor), (LayoutType.RowMajor, LayoutType.ColumnMajor, LayoutType.ColumnMajor), (LayoutType.RowMajor, LayoutType.RowMajor, LayoutType.ColumnMajor)] math_inst = MathInstruction([16, 8, 4], DataType.f64, DataType.f64, DataType.f64, OpcodeClass.TensorOp, MathOperation.multiply_add_complex_gaussian) min_cc = 90 max_cc = 1024 alignment_constraints = [1] tile_descriptions = [TileDescription([64, 64, 8], 3, [4, 2, 1], math_inst, min_cc, max_cc), TileDescription([64, 32, 8], 4, [2, 2, 1], math_inst, min_cc, max_cc), TileDescription([32, 64, 8], 4, [2, 2, 1], math_inst, min_cc, max_cc), TileDescription([32, 32, 8], 4, [2, 2, 1], math_inst, min_cc, max_cc), TileDescription([16, 32, 8], 4, [1, 2, 1], math_inst, min_cc, max_cc), TileDescription([32, 16, 8], 4, [2, 1, 1], math_inst, min_cc, max_cc)] data_type = [DataType.cf64, DataType.cf64, DataType.cf64, DataType.cf64] complex_transforms = [(ComplexTransform.none, ComplexTransform.none), (ComplexTransform.conj, ComplexTransform.none), (ComplexTransform.none, ComplexTransform.conj), (ComplexTransform.conj, ComplexTransform.conj)] CreateGemmOperator(manifest, layouts, tile_descriptions, data_type, alignment_constraints, complex_transforms)
class AssistQVideoPath(Dataset): def __init__(self, root, split, save_dir, num_gpus): super().__init__() (videos, save_dirs) = get_assistq_videos(root, split, save_dir) (self.videos, self.save_dirs) = ([], []) for (video, save_dir) in zip(videos, save_dirs): self.videos.append(video) self.save_dirs.append(save_dir) print(f'total video data size: {len(self.videos)}') n = ((len(self.videos) // num_gpus) + 1) self.videos = [self.videos[(n * i):(n * (i + 1))] for i in range(num_gpus)] self.save_dirs = [self.save_dirs[(n * i):(n * (i + 1))] for i in range(num_gpus)] self.num_gpus = num_gpus def __getitem__(self, index): return (self.videos[index], self.save_dirs[index]) def __len__(self): return self.num_gpus
def save(papers, authors, edges): biadjacency = sparse.coo_matrix((np.ones(len(edges), dtype=np.bool), (edges['paper_node_id'], edges['author_node_id']))) papers.drop('paper_node_id', axis=1, inplace=True) authors.drop('author_node_id', axis=1, inplace=True) edges.drop('paper_node_id', axis=1, inplace=True) edges.drop('author_node_id', axis=1, inplace=True) print('saving:') print(' paper table: {:,} papers, {:,} features'.format(*papers.shape)) papers.to_csv('s2_2_bipartite_graph/papers.csv') print(' edges table: {:,} edges'.format(edges.shape[0])) edges.to_csv('s2_2_bipartite_graph/paper_author_edges.csv', index=False) print(' biadjacency matrix: {:,} papers, {:,} authors, {:,} edges'.format(*biadjacency.shape, biadjacency.nnz)) sparse.save_npz('s2_2_bipartite_graph/paper_author_biadjacency.npz', biadjacency)
class FullTokenizer(object): def __init__(self, vocab_file, do_lower_case=True): self.vocab = load_vocab(vocab_file) self.basic_tokenizer = BasicTokenizer(do_lower_case=do_lower_case) self.wordpiece_tokenizer = WordpieceTokenizer(vocab=self.vocab) def tokenize(self, text): split_tokens = [] for token in self.basic_tokenizer.tokenize(text): for sub_token in self.wordpiece_tokenizer.tokenize(token): split_tokens.append(sub_token) return split_tokens def convert_tokens_to_ids(self, tokens): return convert_tokens_to_ids(self.vocab, tokens)
class Parse_Vuln(): json_data = dict() filename = 'vulnerabilities.json' tools_info_obj = tools_info.Tools_Info() def ParseArgs(self): Args = argparse.ArgumentParser(description='Parser to parse vulnerability result file into JSON') Args.add_argument('--src', required=True, help='result file absolute path to parse') Args.add_argument('--dst', required=True, help='output file absolute path to generate JSON file from result') return Args.parse_args() def read_file_content(self, args): inpFile = args.src output_File = args.dst file_obj = open(inpFile) data = json.load(file_obj) self.tools_info_obj.get_threshold() for filename in data: final_result = {key: {} for key in self.tools_info_obj.vuln_threshold.keys()} majority_result = {} for vuln in data[filename].keys(): if (vuln != 'LE'): for tool_name in data[filename][vuln].keys(): for lineno in data[filename][vuln][tool_name]: final_result[vuln][lineno] = 0 if (vuln == 'LE'): for tool_name in data[filename][vuln]: final_result[vuln][1] = 0 for vuln in data[filename].keys(): if (vuln != 'LE'): for tool_name in data[filename][vuln].keys(): for lineno in data[filename][vuln][tool_name]: final_result[vuln][lineno] += 1 if (vuln == 'LE'): for tool_name in data[filename][vuln]: final_result[vuln][1] += 1 final_result = dict([(k, v) for (k, v) in final_result.items() if (len(v) > 0)]) for vuln in final_result: majority_result[vuln] = [] print_vuln = [] for vuln in final_result: for lineno in final_result[vuln]: if (final_result[vuln][lineno] >= self.tools_info_obj.vuln_threshold[vuln]): if (vuln == 'LE'): majority_result[vuln].append('Yes') elif (vuln != 'LE'): majority_result[vuln].append(lineno) majority_result = dict([(k, v) for (k, v) in majority_result.items() if ((type(v) == int) or (len(v) > 0))]) self.json_data[filename] = majority_result self.json_data = dict([(k, v) for (k, v) in self.json_data.items() if (len(v) > 0)]) file_obj.close() with open(output_File, 'w') as f: json.dump(self.json_data, f, sort_keys=True, indent=4)
class JointPPO(): def __init__(self, actor_critic, clip_param, ppo_epoch, num_mini_batch, value_loss_coef, entropy_coef, lr=None, eps=None, max_grad_norm=None, use_clipped_value_loss=False): self.actor_critic = actor_critic self.clip_param = clip_param self.ppo_epoch = ppo_epoch self.num_mini_batch = num_mini_batch self.value_loss_coef = value_loss_coef self.entropy_coef = entropy_coef self.max_grad_norm = max_grad_norm self.use_clipped_value_loss = use_clipped_value_loss self.optimizer = optim.Adam(actor_critic.parameters(), lr=lr) def update(self, rollouts_list): advantages_list = [] for rollout in rollouts_list: advantages = (rollout.returns[:(- 1)] - rollout.value_preds[:(- 1)]) advantages = ((advantages - advantages.mean()) / (advantages.std() + 1e-05)) advantages_list.append(advantages) value_loss_epoch = 0 action_loss_epoch = 0 dist_entropy_epoch = 0 for e in range(self.ppo_epoch): if self.actor_critic.is_recurrent: raise NotImplementedError('sampler not implemented for recurrent policies') else: data_generator = magent_feed_forward_generator(rollouts_list, advantages_list, self.num_mini_batch) for sample in data_generator: (obs_batch, recurrent_hidden_states_batch, actions_batch, value_preds_batch, return_batch, masks_batch, old_action_log_probs_batch, adv_targ) = sample (values, action_log_probs, dist_entropy, _) = self.actor_critic.evaluate_actions(obs_batch, recurrent_hidden_states_batch, masks_batch, actions_batch) ratio = torch.exp((action_log_probs - old_action_log_probs_batch)) surr1 = (ratio * adv_targ) surr2 = (torch.clamp(ratio, (1.0 - self.clip_param), (1.0 + self.clip_param)) * adv_targ) action_loss = (- torch.min(surr1, surr2).mean()) if self.use_clipped_value_loss: value_pred_clipped = (value_preds_batch + (values - value_preds_batch).clamp((- self.clip_param), self.clip_param)) value_losses = (values - return_batch).pow(2) value_losses_clipped = (value_pred_clipped - return_batch).pow(2) value_loss = (0.5 * torch.max(value_losses, value_losses_clipped).mean()) else: value_loss = (0.5 * F.mse_loss(return_batch, values)) self.optimizer.zero_grad() (((value_loss * self.value_loss_coef) + action_loss) - (dist_entropy * self.entropy_coef)).backward() nn.utils.clip_grad_norm_(self.actor_critic.parameters(), self.max_grad_norm) self.optimizer.step() value_loss_epoch += value_loss.item() action_loss_epoch += action_loss.item() dist_entropy_epoch += dist_entropy.item() num_updates = (self.ppo_epoch * self.num_mini_batch) value_loss_epoch /= num_updates action_loss_epoch /= num_updates dist_entropy_epoch /= num_updates return (value_loss_epoch, action_loss_epoch, dist_entropy_epoch)
def markinnerspaces(line): l = '' f = 0 cc = "'" cb = '' for c in line: if ((cb == '\\') and (c in ['\\', "'", '"'])): l = (l + c) cb = c continue if ((f == 0) and (c in ["'", '"'])): cc = c if (c == cc): f = (f + 1) elif (c == cc): f = (f - 1) elif ((c == ' ') and (f == 1)): l = (l + '_') continue l = (l + c) cb = c return l
class MixSpec(ComplexitySpec): environment = {'asr_acc': 0.7, 'asr_std': 0.15} proposition = {'yn_question': 0.4, 'reject_style': {'reject': 0.5, 'reject+inform': 0.5}, 'multi_slots': {1: 0.7, 2: 0.3}, 'dont_care': 0.1, 'multi_goals': {1: 0.6, 2: 0.4}} interaction = {'hesitation': 0.4, 'self_restart': 0.1, 'self_correct': 0.2} social = {'self_disclosure': None, 'ref_shared': None, 'violation_sn': None}
class IBMVPCInstance(): def __init__(self, name, ibm_vpc_config, ibm_vpc_client=None, public=False): self.name = name.lower() self.config = ibm_vpc_config self.delete_on_dismantle = self.config['delete_on_dismantle'] self.profile_name = self.config['worker_profile_name'] self.vpc_cli = None self.vpc_cli = (ibm_vpc_client or self._create_vpc_client()) self.public = public self.ssh_client = None self.instance_id = None self.instance_data = None self.private_ip = None self.public_ip = None self.floating_ip_id = None self.home_dir = '/root' self.ssh_credentials = {'username': self.config['ssh_username'], 'password': self.config['ssh_password'], 'key_filename': self.config.get('ssh_key_filename', '~/.ssh/id_rsa')} self.validated = False def __str__(self): ip = (self.public_ip if self.public else self.private_ip) if ((ip is None) or (ip == '0.0.0.0')): return f'VM instance {self.name}' else: return f'VM instance {self.name} ({ip})' ('ibm_vpc', 'ibm_cloud_sdk_core', pip_extra='ibmcloud') def _create_vpc_client(ibm_vpc, ibm_cloud_sdk_core, self): authenticator = ibm_cloud_sdk_core.authenticators.IAMAuthenticator(self.config.get('iam_api_key'), url=self.config.get('iam_endpoint')) ibm_vpc_client = ibm_vpc.VpcV1(VPC_API_VERSION, authenticator=authenticator) ibm_vpc_client.set_service_url((self.config['endpoint'] + '/v1')) decorate_instance(self.vpc_cli, vpc_retry_on_except) return ibm_vpc_client def get_ssh_client(self): if ((not self.validated) and self.public and self.instance_id): key_filename = self.ssh_credentials['key_filename'] key_filename = os.path.abspath(os.path.expanduser(key_filename)) if (not os.path.exists(key_filename)): raise Exception(f"Private key file {key_filename} doesn't exist") initialization_data = self.vpc_cli.get_instance_initialization(self.instance_id).get_result() private_res = paramiko.RSAKey(filename=key_filename).get_base64() names = [] for k in initialization_data['keys']: public_res = self.vpc_cli.get_key(k['id']).get_result()['public_key'].split(' ')[1] if (public_res == private_res): self.validated = True break else: names.append(k['name']) if (not self.validated): raise Exception(f'No public key from keys: {names} on master {self} not a pair for private ssh key {key_filename}') if (self.private_ip or self.public_ip): if (not self.ssh_client): self.ssh_client = SSHClient((self.public_ip or self.private_ip), self.ssh_credentials) return self.ssh_client def del_ssh_client(self): if self.ssh_client: try: self.ssh_client.close() except Exception: pass self.ssh_client = None def is_ready(self): login_type = ('password' if (('password' in self.ssh_credentials) and (not self.public)) else 'publickey') try: self.get_ssh_client().run_remote_command('id') except Exception as err: logger.debug(f'SSH to {(self.public_ip if self.public else self.private_ip)} failed ({login_type}): {err}') self.del_ssh_client() return False return True def wait_ready(self, timeout=INSTANCE_START_TIMEOUT): logger.debug(f'Waiting {self} to become ready') start = time.time() if self.public: self.get_public_ip() else: self.get_private_ip() while ((time.time() - start) < timeout): if self.is_ready(): start_time = round((time.time() - start), 2) logger.debug(f'{self} ready in {start_time} seconds') return True time.sleep(5) raise TimeoutError(f'Readiness probe expired on {self}') ('ibm_cloud_sdk_core', pip_extra='ibmcloud') def _create_instance(ibm_cloud_sdk_core, self, user_data): logger.debug('Creating new VM instance {}'.format(self.name)) security_group_identity_model = {'id': self.config['security_group_id']} subnet_identity_model = {'id': self.config['subnet_id']} primary_network_interface = {'name': 'eth0', 'subnet': subnet_identity_model, 'security_groups': [security_group_identity_model]} boot_volume_data = {'capacity': self.config['boot_volume_capacity'], 'name': f'{self.name}-{str(uuid.uuid4())[:4]}-boot', 'profile': {'name': self.config['boot_volume_profile']}} boot_volume_attachment = {'delete_volume_on_instance_delete': True, 'volume': boot_volume_data} key_identity_model = {'id': self.config['ssh_key_id']} instance_prototype = {} instance_prototype['name'] = self.name instance_prototype['keys'] = [key_identity_model] instance_prototype['profile'] = {'name': self.profile_name} instance_prototype['resource_group'] = {'id': self.config['resource_group_id']} instance_prototype['vpc'] = {'id': self.config['vpc_id']} instance_prototype['image'] = {'id': self.config['image_id']} instance_prototype['zone'] = {'name': self.config['zone_name']} instance_prototype['boot_volume_attachment'] = boot_volume_attachment instance_prototype['primary_network_interface'] = primary_network_interface if user_data: instance_prototype['user_data'] = user_data try: resp = self.vpc_cli.create_instance(instance_prototype) except ibm_cloud_sdk_core.ApiException as err: if ((err.code == 400) and ('already exists' in err.message)): return self.get_instance_data() elif ((err.code == 400) and ('over quota' in err.message)): logger.debug(f'Create VM instance {self.name} failed due to quota limit') else: logger.debug('Create VM instance {} failed with status code {}: {}'.format(self.name, str(err.code), err.message)) raise err logger.debug('VM instance {} created successfully '.format(self.name)) return resp.result def _attach_floating_ip(self, fip, fip_id, instance): instance_primary_ni = instance['primary_network_interface'] if (instance_primary_ni['primary_ipv4_address'] and (instance_primary_ni['id'] == fip_id)): logger.debug('Floating IP {} already attached to eth0'.format(fip)) else: self.vpc_cli.add_instance_network_interface_floating_ip(instance['id'], instance['network_interfaces'][0]['id'], fip_id) def _delete_floating_ip(self, fip_id): response = self.vpc_cli.delete_floating_ip(id=fip_id) def get_instance_data(self): instances_data = self.vpc_cli.list_instances(name=self.name).get_result() if (len(instances_data['instances']) > 0): self.instance_data = instances_data['instances'][0] return self.instance_data def get_instance_id(self): if (not self.instance_id): instance_data = self.get_instance_data() if instance_data: self.instance_id = instance_data['id'] else: logger.debug(f'VM instance {self.name} does not exists') return self.instance_id def get_private_ip(self): while ((not self.private_ip) or (self.private_ip == '0.0.0.0')): time.sleep(1) instance_data = self.get_instance_data() self.private_ip = instance_data['primary_network_interface']['primary_ipv4_address'] return self.private_ip def get_public_ip(self): return self.public_ip def create(self, check_if_exists=False, user_data=None): instance = None vsi_exists = (True if self.instance_id else False) if (check_if_exists and (not vsi_exists)): logger.debug('Checking if VM instance {} already exists'.format(self.name)) instances_data = self.get_instance_data() if instances_data: logger.debug('VM instance {} already exists'.format(self.name)) vsi_exists = True self.instance_id = instances_data['id'] if (not vsi_exists): instance = self._create_instance(user_data=user_data) self.instance_id = instance['id'] else: self.start() if (self.public and instance): self._attach_floating_ip(self.public_ip, self.floating_ip_id, instance) return self.instance_id ('ibm_cloud_sdk_core', pip_extra='ibmcloud') def start(ibm_cloud_sdk_core, self): logger.debug('Starting VM instance {}'.format(self.name)) try: self.vpc_cli.create_instance_action(self.instance_id, 'start') except ibm_cloud_sdk_core.ApiException as err: if (err.code == 404): pass else: raise err logger.debug('VM instance {} started successfully'.format(self.name)) ('ibm_cloud_sdk_core', pip_extra='ibmcloud') def _delete_instance(ibm_cloud_sdk_core, self): logger.debug('Deleting VM instance {}'.format(self.name)) try: self.vpc_cli.delete_instance(self.instance_id) except ibm_cloud_sdk_core.ApiException as err: if (err.code == 404): pass else: raise err self.instance_id = None self.private_ip = None self.del_ssh_client() ('ibm_cloud_sdk_core', pip_extra='ibmcloud') def _stop_instance(ibm_cloud_sdk_core, self): logger.debug('Stopping VM instance {}'.format(self.name)) try: self.vpc_cli.create_instance_action(self.instance_id, 'stop') except ibm_cloud_sdk_core.ApiException as err: if (err.code == 404): pass else: raise err def stop(self): if self.delete_on_dismantle: self._delete_instance() else: self._stop_instance() def delete(self): self._delete_instance() self._delete_floating_ip(self.floating_ip_id) def validate_capabilities(self): if self.config.get('singlesocket'): cmd = "lscpu -p=socket|grep -v '#'" res = self.get_ssh_client().run_remote_command(cmd) sockets = set() for char in res: if (char != '\n'): sockets.add(char) if (len(sockets) != 1): raise Exception(f'Not using single CPU socket as specified, using {len(sockets)} sockets instead')
class PinyinTestCase(unittest.TestCase): def test_single_pinyin(self): sents = ['zhuan', 'zuo'] res = [] for name in sents: (s, r) = m.correct(name) print(s, r) res.append(r) self.assertEqual(res[0], []) self.assertEqual(res[1], []) def test_full_pinyin(self): sents = ['xingfu', 'pingguo'] res = [] for name in sents: (s, r) = m.correct(name) print(s, r) res.append(r) self.assertEqual(res[0], []) self.assertEqual(res[1], [])
class IntFormat(object): def from_number(cls, n, min=None): width = (number_digits(n) + 1) if (n < 0): width += 1 repeat = (80 // width) return cls(width, min, repeat=repeat) def __init__(self, width, min=None, repeat=None): self.width = width self.repeat = repeat self.min = min def __repr__(self): r = 'IntFormat(' if self.repeat: r += ('%d' % self.repeat) r += ('I%d' % self.width) if self.min: r += ('.%d' % self.min) return (r + ')') def fortran_format(self): r = '(' if self.repeat: r += ('%d' % self.repeat) r += ('I%d' % self.width) if self.min: r += ('.%d' % self.min) return (r + ')') def python_format(self): return (('%' + str(self.width)) + 'd')
class Vocab(object): __slots__ = ('_word_dict', '_entity_dict') def __init__(self, word_dict, entity_dict): self._word_dict = word_dict self._entity_dict = entity_dict def word_size(self): return len(self._word_dict) def entity_size(self): return len(self._entity_dict) def words(self): return iter(self._word_dict) def entities(self): return iter(self._entity_dict) def get_word_index(self, word, default=None): try: return self._word_dict.key_id(word) except KeyError: return default def get_entity_index(self, entity, default=None): try: return self._entity_dict.key_id(entity) except KeyError: return default def get_word_by_index(self, index): return self._word_dict.restore_key(index) def get_entity_by_index(self, index): return self._entity_dict.restore_key(index) def build(db, entity_db, min_word_count, min_entity_count): word_counter = Counter() entity_counter = Counter() tokenizer = RegexpTokenizer() with click.progressbar(db.keys()) as bar: for title in bar: obj = db[title] text = obj['text'] tokens = tokenizer.tokenize(text) word_counter.update((t.text.lower() for t in tokens)) for (_, title, _) in obj['links']: title = entity_db.resolve_redirect(title) entity_counter[title] += 1 word_dict = Trie([w for (w, c) in word_counter.iteritems() if (c >= min_word_count)]) entity_dict = Trie([e for (e, c) in entity_counter.iteritems() if (c >= min_entity_count)]) return Vocab(word_dict, entity_dict) def save(self, out_file): self._word_dict.save((out_file + '_word.trie')) self._entity_dict.save((out_file + '_entity.trie')) def load(in_file, mmap=True): word_dict = Trie() entity_dict = Trie() word_dict.mmap((in_file + '_word.trie')) entity_dict.mmap((in_file + '_entity.trie')) return Vocab(word_dict, entity_dict) def __reduce__(self): return (self.__class__, (self._word_dict, self._entity_dict))
def click_play_button_off(): pyautogui.doubleClick('screenshots/play_button_off.PNG') pyautogui.moveTo(1000, 1000, duration=0)
def compute_metrics_for_regression(y_test, y_test_pred): metrics = {} for task in DIMENSIONS: targets_task = [t[DIMENSIONS.index(task)] for t in y_test] pred_task = [l[DIMENSIONS.index(task)] for l in y_test_pred] rmse = mean_squared_error(targets_task, pred_task, squared=False) metrics[f'rmse_{task}'] = rmse return metrics
class DatasetSampler(Dataset): def __init__(self, x): self.x = x def __len__(self): return len(self.x) def __getitem__(self, idx): return self.x[idx].astype('float32')
def check_build_sdist(hooks, build_sys_requires): with BuildEnvironment() as env: try: env.pip_install(build_sys_requires) log.info('Installed static build dependencies') except CalledProcessError: log.error('Failed to install static build dependencies') return False try: reqs = hooks.get_requires_for_build_sdist({}) log.info('Got build requires: %s', reqs) except Exception: log.error('Failure in get_requires_for_build_sdist', exc_info=True) return False try: env.pip_install(reqs) log.info('Installed dynamic build dependencies') except CalledProcessError: log.error('Failed to install dynamic build dependencies') return False td = mkdtemp() log.info('Trying to build sdist in %s', td) try: try: filename = hooks.build_sdist(td, {}) log.info('build_sdist returned %r', filename) except Exception: log.info('Failure in build_sdist', exc_info=True) return False if (not filename.endswith('.tar.gz')): log.error("Filename %s doesn't have .tar.gz extension", filename) return False path = pjoin(td, filename) if isfile(path): log.info('Output file %s exists', path) else: log.error('Output file %s does not exist', path) return False if tarfile.is_tarfile(path): log.info('Output file is a tar file') else: log.error('Output file is not a tar file') return False finally: shutil.rmtree(td) return True
.parametrize('task_name', [tn for tn in (all_tasks - julia_tasks)]) def test_obtain_prior_from_task(task_name): task = get_task(task_name) prior = task.get_prior() assert (prior is not None)
def standard_confusion_matrix(y_test, y_test_pred): [[tn, fp], [fn, tp]] = confusion_matrix(y_test, y_test_pred) return np.array([[tp, fp], [fn, tn]])
class ASPP_Efficientnetv2(nn.Module): def __init__(self, num_classes, concat=True, output_kernel_size=1): super(ASPP_Efficientnetv2, self).__init__() self.concat = concat self.conv_1x1_1 = nn.Conv2d(512, 256, kernel_size=1) self.bn_conv_1x1_1 = nn.BatchNorm2d(256) self.conv_3x3_1 = nn.Conv2d(512, 256, kernel_size=3, stride=1, padding=6, dilation=6) self.bn_conv_3x3_1 = nn.BatchNorm2d(256) self.conv_3x3_2 = nn.Conv2d(512, 256, kernel_size=3, stride=1, padding=12, dilation=12) self.bn_conv_3x3_2 = nn.BatchNorm2d(256) self.conv_3x3_3 = nn.Conv2d(512, 256, kernel_size=3, stride=1, padding=18, dilation=18) self.bn_conv_3x3_3 = nn.BatchNorm2d(256) self.avg_pool = nn.AdaptiveAvgPool2d(1) self.conv_1x1_2 = nn.Conv2d(512, 256, kernel_size=1) self.bn_conv_1x1_2 = nn.BatchNorm2d(256) self.conv_1x1_3 = nn.Conv2d(1280, 256, kernel_size=1) self.bn_conv_1x1_3 = nn.BatchNorm2d(256) kernel_size = 3 padding = 1 output_padding = 0 if (kernel_size == 3): output_padding = 1 elif (kernel_size == 2): padding = 0 if self.concat: self.upsample_1 = self.upsample(256, 256, 3, padding, output_padding) self.upsample_2 = self.upsample((256 + 48), 256, 3, padding, output_padding) else: self.upsample_1 = self.upsample(256, 256, 3, padding, output_padding) self.upsample_2 = self.upsample(256, 256, 3, padding, output_padding) if (output_kernel_size == 3): padding = 1 elif (output_kernel_size == 2): padding = 0 elif (output_kernel_size == 1): padding = 0 self.conv_1x1_4 = nn.Conv2d((256 + 24), num_classes, kernel_size=output_kernel_size, padding=padding) def upsample(self, in_channels, num_filters, kernel_size, padding, output_padding): upsample_layer = nn.Sequential(nn.ConvTranspose2d(in_channels, num_filters, kernel_size=kernel_size, stride=2, padding=padding, output_padding=output_padding, bias=False), nn.BatchNorm2d(num_filters), nn.ReLU(inplace=True), nn.Conv2d(num_filters, num_filters, kernel_size=3, stride=1, padding=1, bias=False), nn.BatchNorm2d(num_filters), nn.ReLU(inplace=True), nn.Conv2d(num_filters, num_filters, kernel_size=3, stride=1, padding=1, bias=False), nn.BatchNorm2d(num_filters), nn.ReLU(inplace=True)) return upsample_layer def forward(self, x_high_feature, l3=None, l2=None): feature_map_h = x_high_feature.size()[2] feature_map_w = x_high_feature.size()[3] out_1x1 = F.relu(self.bn_conv_1x1_1(self.conv_1x1_1(x_high_feature))) out_3x3_1 = F.relu(self.bn_conv_3x3_1(self.conv_3x3_1(x_high_feature))) out_3x3_2 = F.relu(self.bn_conv_3x3_2(self.conv_3x3_2(x_high_feature))) out_3x3_3 = F.relu(self.bn_conv_3x3_3(self.conv_3x3_3(x_high_feature))) out_img = self.avg_pool(x_high_feature) out_img = F.relu(self.bn_conv_1x1_2(self.conv_1x1_2(out_img))) out_img = F.interpolate(out_img, size=(feature_map_h, feature_map_w), mode='bilinear') out = torch.cat([out_1x1, out_3x3_1, out_3x3_2, out_3x3_3, out_img], 1) out = F.relu(self.bn_conv_1x1_3(self.conv_1x1_3(out))) if self.concat: x = self.upsample_1(out) x = torch.cat([x, l3], 1) x = self.upsample_2(x) else: x = self.upsample_1(out) x = self.upsample_2(x) x = self.conv_1x1_4(torch.cat([x, l2], 1)) return x