code
stringlengths
101
5.91M
class Pool(): def __init__(self, processes=1): self.processes = processes def _worker_loop(self, func, ip, inputQueue): while True: args = inputQueue.get(block=True, timeout=None) if (args == (- 1)): break p = mp.Process(target=func, args=([ip] + args), daemon=False) p.start() p.join() if (p.exitcode != 0): print(('ERROR: process %i terminated unexpectedly' % os.getpid())) break def starmap(self, func, iterable): inputQueue = mp.Queue() for args in iterable: inputQueue.put(list(args)) for _ in range(self.processes): inputQueue.put((- 1)) ps = [mp.Process(target=self._worker_loop, args=(func, ip, inputQueue), daemon=False) for ip in range(self.processes)] for p in ps: p.start() for p in ps: p.join() def __enter__(self): return self def __exit__(self, *args): pass
class SquadDataTrainingArguments(): def __init__(self, *args, **kwargs): requires_pytorch(self)
class TestChoi(ChannelTestCase): def test_init(self): mat4 = (np.eye(4) / 2.0) chan = Choi(mat4) self.assertAllClose(chan.data, mat4) self.assertEqual(chan.dim, (2, 2)) mat8 = (np.eye(8) / 2.0) chan = Choi(mat8, input_dims=4) self.assertAllClose(chan.data, mat8) self.assertEqual(chan.dim, (4, 2)) chan = Choi(mat8, input_dims=2) self.assertAllClose(chan.data, mat8) self.assertEqual(chan.dim, (2, 4)) mat16 = (np.eye(16) / 4) chan = Choi(mat16) self.assertAllClose(chan.data, mat16) self.assertEqual(chan.dim, (4, 4)) self.assertRaises(QiskitError, Choi, mat8, input_dims=[4], output_dims=[4]) def test_circuit_init(self): (circuit, target) = self.simple_circuit_no_measure() op = Choi(circuit) target = Choi(target) self.assertEqual(op, target) def test_circuit_init_except(self): circuit = self.simple_circuit_with_measure() self.assertRaises(QiskitError, Choi, circuit) def test_equal(self): mat = self.rand_matrix(4, 4) self.assertEqual(Choi(mat), Choi(mat)) def test_copy(self): mat = np.eye(4) orig = Choi(mat) cpy = orig.copy() cpy._data[(0, 0)] = 0.0 self.assertFalse((cpy == orig)) def test_evolve(self): input_psi = [0, 1] input_rho = [[0, 0], [0, 1]] chan = Choi(self.choiI) target_rho = np.array([[0, 0], [0, 1]]) self.assertAllClose(chan._evolve(input_psi), target_rho) self.assertAllClose(chan._evolve(np.array(input_psi)), target_rho) self.assertAllClose(chan._evolve(input_rho), target_rho) self.assertAllClose(chan._evolve(np.array(input_rho)), target_rho) chan = Choi(self.choiH) target_rho = (np.array([[1, (- 1)], [(- 1), 1]]) / 2) self.assertAllClose(chan._evolve(input_psi), target_rho) self.assertAllClose(chan._evolve(np.array(input_psi)), target_rho) self.assertAllClose(chan._evolve(input_rho), target_rho) self.assertAllClose(chan._evolve(np.array(input_rho)), target_rho) chan = Choi(self.depol_choi(1)) target_rho = (np.eye(2) / 2) self.assertAllClose(chan._evolve(input_psi), target_rho) self.assertAllClose(chan._evolve(np.array(input_psi)), target_rho) self.assertAllClose(chan._evolve(input_rho), target_rho) self.assertAllClose(chan._evolve(np.array(input_rho)), target_rho) def test_is_cptp(self): self.assertTrue(Choi(self.depol_choi(0.25)).is_cptp()) self.assertFalse(Choi(((1.25 * self.choiI) - (0.25 * self.depol_choi(1)))).is_cptp()) def test_conjugate(self): (Zp, Zm) = (np.diag([1, 0]), np.diag([0, 1])) (Yp, Ym) = ((np.array([[1, (- 1j)], [1j, 1]]) / 2), (np.array([[1, 1j], [(- 1j), 1]]) / 2)) chan = Choi((np.kron(Zp, Yp) + np.kron(Zm, Ym))) targ = Choi((np.kron(Zp, Ym) + np.kron(Zm, Yp))) chan_conj = chan.conjugate() self.assertEqual(chan_conj, targ) def test_transpose(self): (Zp, Zm) = (np.diag([1, 0]), np.diag([0, 1])) (Yp, Ym) = ((np.array([[1, (- 1j)], [1j, 1]]) / 2), (np.array([[1, 1j], [(- 1j), 1]]) / 2)) chan = Choi((np.kron(Zp, Yp) + np.kron(Zm, Ym))) targ = Choi((np.kron(Yp, Zp) + np.kron(Ym, Zm))) chan_t = chan.transpose() self.assertEqual(chan_t, targ) def test_adjoint(self): (Zp, Zm) = (np.diag([1, 0]), np.diag([0, 1])) (Yp, Ym) = ((np.array([[1, (- 1j)], [1j, 1]]) / 2), (np.array([[1, 1j], [(- 1j), 1]]) / 2)) chan = Choi((np.kron(Zp, Yp) + np.kron(Zm, Ym))) targ = Choi((np.kron(Ym, Zp) + np.kron(Yp, Zm))) chan_adj = chan.adjoint() self.assertEqual(chan_adj, targ) def test_compose_except(self): self.assertRaises(QiskitError, Choi(np.eye(4)).compose, Choi(np.eye(8))) self.assertRaises(QiskitError, Choi(np.eye(4)).compose, 2) def test_compose(self): chan1 = Choi(self.choiX) chan2 = Choi(self.choiY) chan = chan1.compose(chan2) targ = Choi(self.choiZ) self.assertEqual(chan, targ) chan1 = Choi(self.depol_choi(0.5)) chan = chan1.compose(chan1) targ = Choi(self.depol_choi(0.75)) self.assertEqual(chan, targ) (Zp, Zm) = (np.diag([1, 0]), np.diag([0, 1])) (Xp, Xm) = ((np.array([[1, 1], [1, 1]]) / 2), (np.array([[1, (- 1)], [(- 1), 1]]) / 2)) chan1 = Choi((np.kron(Zp, Xp) + np.kron(Zm, Xm))) chan2 = Choi(self.choiX) targ = Choi((np.kron(Zp, Xp) + np.kron(Zm, Xm))) self.assertEqual(chan1.compose(chan2), targ) self.assertEqual((chan1 chan2), targ) targ = Choi((np.kron(Zm, Xp) + np.kron(Zp, Xm))) self.assertEqual(chan2.compose(chan1), targ) self.assertEqual((chan2 chan1), targ) chan1 = Choi((np.eye(8) / 4), input_dims=2, output_dims=4) chan2 = Choi((np.eye(8) / 2), input_dims=4, output_dims=2) chan = chan1.compose(chan2) self.assertEqual(chan.dim, (2, 2)) chan = chan2.compose(chan1) self.assertEqual(chan.dim, (4, 4)) def test_compose_front(self): chan1 = Choi(self.choiX) chan2 = Choi(self.choiY) chan = chan1.compose(chan2, front=True) targ = Choi(self.choiZ) self.assertEqual(chan, targ) chan1 = Choi(self.depol_choi(0.5)) chan = chan1.compose(chan1, front=True) targ = Choi(self.depol_choi(0.75)) self.assertEqual(chan, targ) (Zp, Zm) = (np.diag([1, 0]), np.diag([0, 1])) (Xp, Xm) = ((np.array([[1, 1], [1, 1]]) / 2), (np.array([[1, (- 1)], [(- 1), 1]]) / 2)) chan1 = Choi((np.kron(Zp, Xp) + np.kron(Zm, Xm))) chan2 = Choi(self.choiX) chan = chan2.compose(chan1, front=True) targ = Choi((np.kron(Zp, Xp) + np.kron(Zm, Xm))) self.assertEqual(chan, targ) chan = chan1.compose(chan2, front=True) targ = Choi((np.kron(Zm, Xp) + np.kron(Zp, Xm))) self.assertEqual(chan, targ) chan1 = Choi((np.eye(8) / 4), input_dims=2, output_dims=4) chan2 = Choi((np.eye(8) / 2), input_dims=4, output_dims=2) chan = chan1.compose(chan2, front=True) self.assertEqual(chan.dim, (4, 4)) chan = chan2.compose(chan1, front=True) self.assertEqual(chan.dim, (2, 2)) def test_expand(self): (rho0, rho1) = (np.diag([1, 0]), np.diag([0, 1])) rho_init = np.kron(rho0, rho0) chan1 = Choi(self.choiI) chan2 = Choi(self.choiX) chan = chan1.expand(chan2) rho_targ = np.kron(rho1, rho0) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) chan = chan2.expand(chan1) rho_targ = np.kron(rho0, rho1) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) chan_dep = Choi(self.depol_choi(1)) chan = chan_dep.expand(chan_dep) rho_targ = (np.diag([1, 1, 1, 1]) / 4) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) def test_tensor(self): (rho0, rho1) = (np.diag([1, 0]), np.diag([0, 1])) rho_init = np.kron(rho0, rho0) chan1 = Choi(self.choiI) chan2 = Choi(self.choiX) rho_targ = np.kron(rho1, rho0) chan = chan2.tensor(chan1) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) chan = (chan2 ^ chan1) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) rho_targ = np.kron(rho0, rho1) chan = chan1.tensor(chan2) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) chan = (chan1 ^ chan2) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) rho_targ = (np.diag([1, 1, 1, 1]) / 4) chan_dep = Choi(self.depol_choi(1)) chan = chan_dep.tensor(chan_dep) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) chan = (chan_dep ^ chan_dep) self.assertEqual(chan.dim, (4, 4)) self.assertAllClose(chan._evolve(rho_init), rho_targ) def test_power(self): p_id = 0.9 depol = Choi(self.depol_choi((1 - p_id))) p_id3 = (p_id ** 3) chan3 = depol.power(3) targ3 = Choi(self.depol_choi((1 - p_id3))) self.assertEqual(chan3, targ3) def test_power_except(self): chan = Choi(self.depol_choi(1)) self.assertRaises(QiskitError, chan.power, 0.5) def test_add(self): mat1 = (0.5 * self.choiI) mat2 = (0.5 * self.depol_choi(1)) targ = Choi((mat1 + mat2)) chan1 = Choi(mat1) chan2 = Choi(mat2) self.assertEqual(chan1.add(chan2), targ) self.assertEqual((chan1 + chan2), targ) def test_add_except(self): chan1 = Choi(self.choiI) chan2 = Choi(np.eye(8)) self.assertRaises(QiskitError, chan1.add, chan2) self.assertRaises(QiskitError, chan1.add, 5) def test_subtract(self): mat1 = (0.5 * self.choiI) mat2 = (0.5 * self.depol_choi(1)) targ = Choi((mat1 - mat2)) chan1 = Choi(mat1) chan2 = Choi(mat2) self.assertEqual(chan1.subtract(chan2), targ) self.assertEqual((chan1 - chan2), targ) def test_subtract_except(self): chan1 = Choi(self.choiI) chan2 = Choi(np.eye(8)) self.assertRaises(QiskitError, chan1.subtract, chan2) self.assertRaises(QiskitError, chan1.subtract, 5) def test_multiply(self): chan = Choi(self.choiI) val = 0.5 targ = Choi((val * self.choiI)) self.assertEqual(chan.multiply(val), targ) self.assertEqual((val * chan), targ) self.assertEqual((chan * val), targ) def test_multiply_except(self): chan = Choi(self.choiI) self.assertRaises(QiskitError, chan.multiply, 's') self.assertRaises(QiskitError, chan.multiply, chan) def test_negate(self): chan = Choi(self.choiI) targ = Choi(((- 1) * self.choiI)) self.assertEqual((- chan), targ)
class wrong_loss(nn.Module): def __init__(self): super(wrong_loss, self).__init__() def forward(self, pred_score, target_score): tar_sum = torch.sum(target_score, dim=1, keepdim=True) tar_sum_is_0 = torch.eq(tar_sum, 0) tar_sum.masked_fill_(tar_sum_is_0, 1e-06) tar = (target_score / tar_sum) res = F.log_softmax(pred_score, dim=1) loss = F.kl_div(res, tar, size_average=True) loss *= target_score.size(1) return loss
_tokenizer('bpe') class SentencePieceBPETokenizer(SentencePieceUnigramTokenizer): MODEL_TYPE = 'bpe'
def find_model_using_name(model_name): model_filename = (('models.' + model_name) + '_model') modellib = importlib.import_module(model_filename) model = None target_model_name = (model_name.replace('_', '') + 'model') for (name, cls) in modellib.__dict__.items(): if ((name.lower() == target_model_name.lower()) and issubclass(cls, torch.nn.Module)): model = cls if (model is None): print(('In %s.py, there should be a subclass of torch.nn.Module with class name that matches %s in lowercase.' % (model_filename, target_model_name))) exit(0) return model
class BaseTrainer(): def __init__(self, actor, loaders, optimizer, settings, lr_scheduler=None): self.actor = actor self.optimizer = optimizer self.lr_scheduler = lr_scheduler self.loaders = loaders self.update_settings(settings) self.epoch = 0 self.stats = {} self.device = getattr(settings, 'device', None) if (self.device is None): self.device = torch.device(('cuda:0' if (torch.cuda.is_available() and settings.use_gpu) else 'cpu')) self.actor.to(self.device) def update_settings(self, settings=None): if (settings is not None): self.settings = settings if (self.settings.env.workspace_dir is not None): self.settings.env.workspace_dir = os.path.expanduser(self.settings.env.workspace_dir) self._checkpoint_dir = os.path.join(self.settings.env.workspace_dir, 'checkpoints') if (not os.path.exists(self._checkpoint_dir)): os.makedirs(self._checkpoint_dir) else: self._checkpoint_dir = None def train(self, max_epochs, load_latest=False, fail_safe=True): epoch = (- 1) num_tries = 10 for i in range(num_tries): try: if load_latest: self.load_checkpoint() for epoch in range((self.epoch + 1), (max_epochs + 1)): self.epoch = epoch self.train_epoch() if (self.lr_scheduler is not None): self.lr_scheduler.step() if self._checkpoint_dir: self.save_checkpoint() except: print('Training crashed at epoch {}'.format(epoch)) if fail_safe: self.epoch -= 1 load_latest = True print('Traceback for the error!') print(traceback.format_exc()) print('Restarting training from last epoch ...') else: raise print('Finished training!') def train_epoch(self): raise NotImplementedError def save_checkpoint(self): net = (self.actor.net.module if multigpu.is_multi_gpu(self.actor.net) else self.actor.net) actor_type = type(self.actor).__name__ net_type = type(net).__name__ state = {'epoch': self.epoch, 'actor_type': actor_type, 'net_type': net_type, 'net': net.state_dict(), 'net_info': getattr(net, 'info', None), 'constructor': getattr(net, 'constructor', None), 'optimizer': self.optimizer.state_dict(), 'stats': self.stats, 'settings': self.settings} directory = '{}/{}'.format(self._checkpoint_dir, self.settings.project_path) if (not os.path.exists(directory)): os.makedirs(directory) if ((self.epoch % 3) == 0): tmp_file_path = '{}/{}_ep{:04d}.tmp'.format(directory, net_type, self.epoch) torch.save(state, tmp_file_path) file_path = '{}/{}_ep{:04d}.pth.tar'.format(directory, net_type, self.epoch) os.rename(tmp_file_path, file_path) def load_checkpoint(self, checkpoint=None, fields=None, ignore_fields=None, load_constructor=False): net = (self.actor.net.module if multigpu.is_multi_gpu(self.actor.net) else self.actor.net) actor_type = type(self.actor).__name__ net_type = type(net).__name__ if (checkpoint is None): checkpoint_list = sorted(glob.glob('{}/{}/{}_ep*.pth.tar'.format(self._checkpoint_dir, self.settings.project_path, net_type))) if checkpoint_list: checkpoint_path = checkpoint_list[(- 1)] else: print('No matching checkpoint file found') return elif isinstance(checkpoint, int): checkpoint_path = '{}/{}/{}_ep{:04d}.pth.tar'.format(self._checkpoint_dir, self.settings.project_path, net_type, checkpoint) elif isinstance(checkpoint, str): if os.path.isdir(checkpoint): checkpoint_list = sorted(glob.glob('{}/*_ep*.pth.tar'.format(checkpoint))) if checkpoint_list: checkpoint_path = checkpoint_list[(- 1)] else: raise Exception('No checkpoint found') else: checkpoint_path = os.path.expanduser(checkpoint) else: raise TypeError checkpoint_dict = loading.torch_load_legacy(checkpoint_path) assert (net_type == checkpoint_dict['net_type']), 'Network is not of correct type.' if (fields is None): fields = checkpoint_dict.keys() if (ignore_fields is None): ignore_fields = ['settings'] ignore_fields.extend(['lr_scheduler', 'constructor', 'net_type', 'actor_type', 'net_info']) for key in fields: if (key in ignore_fields): continue if (key == 'net'): net.load_state_dict(checkpoint_dict[key]) elif (key == 'optimizer'): self.optimizer.load_state_dict(checkpoint_dict[key]) else: setattr(self, key, checkpoint_dict[key]) if (load_constructor and ('constructor' in checkpoint_dict) and (checkpoint_dict['constructor'] is not None)): net.constructor = checkpoint_dict['constructor'] if (('net_info' in checkpoint_dict) and (checkpoint_dict['net_info'] is not None)): net.info = checkpoint_dict['net_info'] if ('epoch' in fields): self.lr_scheduler.last_epoch = self.epoch return True
def orient_shapes_hwd(data: (list | tuple), slice_axis: int) -> np.ndarray: if (slice_axis == 0): return np.array(data)[[2, 1, 0]] elif (slice_axis == 1): return np.array(data)[[2, 0, 1]] elif (slice_axis == 2): return np.array(data)
def parse_args(): parser = argparse.ArgumentParser(description='Train a music captioning model') parser.add_argument('experiment_id', type=str) parser.add_argument('--metrics', type=bool, default=False) parser.add_argument('--device_num', type=str, default='0') parser.add_argument('--decoding', type=str, help='type of decoding to use in inference', default=None) parser.add_argument('--beam_size', type=int, help='beam size to use in beam search decoding', default=None) args = parser.parse_args() return args
def filter_kwargs(func): (func) def wrapper(*args, **kwargs): new_kwargs = {k: v for (k, v) in kwargs.items() if (k in ['backend', 'layers', 'models', 'utils'])} return func(*args, **new_kwargs) return wrapper
def get_morgan_bit_fps(data, bits=2048, radius=2): X = [[c for c in AllChem.GetMorganFingerprintAsBitVect(m, radius, nBits=bits).ToBitString()] for m in data] X = pd.DataFrame(X) return X
class TestInstructions(QiskitTestCase): def test_instructions_equal(self): hop1 = Instruction('h', 1, 0, []) hop2 = Instruction('s', 1, 0, []) hop3 = Instruction('h', 1, 0, []) uop1 = Instruction('u', 1, 0, [0.4, 0.5, 0.5]) uop2 = Instruction('u', 1, 0, [0.4, 0.6, 0.5]) uop3 = Instruction('v', 1, 0, [0.4, 0.5, 0.5]) uop4 = Instruction('u', 1, 0, [0.4, 0.5, 0.5]) self.assertFalse((hop1 == hop2)) self.assertTrue((hop1 == hop3)) self.assertFalse((uop1 == uop2)) self.assertTrue((uop1 == uop4)) self.assertFalse((uop1 == uop3)) self.assertTrue((HGate() == HGate())) self.assertFalse((HGate() == CnotGate())) self.assertFalse((hop1 == HGate())) def test_instructions_equal_with_parameters(self): theta = Parameter('theta') phi = Parameter('phi') self.assertEqual(Instruction('u', 1, 0, [theta, phi, 0.4]), Instruction('u', 1, 0, [theta, phi, 0.4])) self.assertNotEqual(Instruction('u', 1, 0, [theta, phi, 0]), Instruction('u', 1, 0, [phi, theta, 0])) self.assertNotEqual(Instruction('u', 1, 0, [theta, phi, 0.4]), Instruction('u', 1, 0, [theta, phi, 0.5])) self.assertNotEqual(Instruction('u', 1, 0, [0.3, phi, 0.4]), Instruction('u', 1, 0, [theta, phi, 0.5])) def circuit_instruction_circuit_roundtrip(self): q = QuantumRegister(4) c = ClassicalRegister(4) circ1 = QuantumCircuit(q, c, name='circ1') circ1.h(q[0]) circ1.crz(0.1, q[0], q[1]) circ1.iden(q[1]) circ1.u3(0.1, 0.2, (- 0.2), q[0]) circ1.barrier() circ1.measure(q, c) circ1.rz(0.8, q[0]).c_if(c, 6) inst = circ1.to_instruction() circ2 = QuantumCircuit(q, c, name='circ2') circ2.append(inst, q[:]) self.assertEqual(circ1, circ2) def test_append_opaque_wrong_dimension(self): qr = QuantumRegister(2) circ = QuantumCircuit(qr) opaque_gate = Gate(name='crz_2', num_qubits=2, params=[0.5]) self.assertRaises(QiskitError, circ.append, opaque_gate, [qr[0]]) def test_opaque_gate(self): q = QuantumRegister(4) c = ClassicalRegister(4) circ = QuantumCircuit(q, c, name='circ') opaque_gate = Gate(name='crz_2', num_qubits=2, params=[0.5]) circ.append(opaque_gate, [q[2], q[0]]) self.assertEqual(circ.data[0][0].name, 'crz_2') self.assertEqual(circ.decompose(), circ) def test_opaque_instruction(self): q = QuantumRegister(4) c = ClassicalRegister(2) circ = QuantumCircuit(q, c) opaque_inst = Instruction(name='my_inst', num_qubits=3, num_clbits=1, params=[0.5]) circ.append(opaque_inst, [q[3], q[1], q[0]], [c[1]]) self.assertEqual(circ.data[0][0].name, 'my_inst') self.assertEqual(circ.decompose(), circ) def test_mirror_gate(self): q = QuantumRegister(4) c = ClassicalRegister(4) circ = QuantumCircuit(q, c, name='circ') circ.h(q[0]) circ.crz(0.1, q[0], q[1]) circ.iden(q[1]) circ.u3(0.1, 0.2, (- 0.2), q[0]) gate = circ.to_instruction() circ = QuantumCircuit(q, c, name='circ') circ.u3(0.1, 0.2, (- 0.2), q[0]) circ.iden(q[1]) circ.crz(0.1, q[0], q[1]) circ.h(q[0]) gate_mirror = circ.to_instruction() self.assertEqual(gate.mirror().definition, gate_mirror.definition) def test_mirror_instruction(self): q = QuantumRegister(4) c = ClassicalRegister(4) circ = QuantumCircuit(q, c, name='circ') circ.t(q[1]) circ.u3(0.1, 0.2, (- 0.2), q[0]) circ.barrier() circ.measure(q[0], c[0]) circ.rz(0.8, q[0]).c_if(c, 6) inst = circ.to_instruction() circ = QuantumCircuit(q, c, name='circ') circ.rz(0.8, q[0]).c_if(c, 6) circ.measure(q[0], c[0]) circ.barrier() circ.u3(0.1, 0.2, (- 0.2), q[0]) circ.t(q[1]) inst_mirror = circ.to_instruction() self.assertEqual(inst.mirror().definition, inst_mirror.definition) def test_mirror_opaque(self): opaque_gate = Gate(name='crz_2', num_qubits=2, params=[0.5]) self.assertEqual(opaque_gate.mirror(), opaque_gate) hgate = HGate() self.assertEqual(hgate.mirror(), hgate) def test_inverse_gate(self): q = QuantumRegister(4) circ = QuantumCircuit(q, name='circ') circ.h(q[0]) circ.crz(0.1, q[0], q[1]) circ.iden(q[1]) circ.u3(0.1, 0.2, (- 0.2), q[0]) gate = circ.to_instruction() circ = QuantumCircuit(q, name='circ') circ.u3((- 0.1), 0.2, (- 0.2), q[0]) circ.iden(q[1]) circ.crz((- 0.1), q[0], q[1]) circ.h(q[0]) gate_inverse = circ.to_instruction() self.assertEqual(gate.inverse().definition, gate_inverse.definition) def test_inverse_recursive(self): qr0 = QuantumRegister(2) circ0 = QuantumCircuit(qr0, name='circ0') circ0.t(qr0[0]) circ0.rx(0.4, qr0[1]) circ0.cx(qr0[1], qr0[0]) little_gate = circ0.to_instruction() qr1 = QuantumRegister(4) circ1 = QuantumCircuit(qr1, name='circ1') circ1.cu1((- 0.1), qr1[0], qr1[2]) circ1.iden(qr1[1]) circ1.append(little_gate, [qr1[2], qr1[3]]) circ_inv = QuantumCircuit(qr1, name='circ1_dg') circ_inv.append(little_gate.inverse(), [qr1[2], qr1[3]]) circ_inv.iden(qr1[1]) circ_inv.cu1(0.1, qr1[0], qr1[2]) self.assertEqual(circ1.inverse(), circ_inv) def test_inverse_instruction_with_measure(self): q = QuantumRegister(4) c = ClassicalRegister(4) circ = QuantumCircuit(q, c, name='circ') circ.t(q[1]) circ.u3(0.1, 0.2, (- 0.2), q[0]) circ.barrier() circ.measure(q[0], c[0]) inst = circ.to_instruction() self.assertRaises(QiskitError, inst.inverse) def test_inverse_instruction_with_conditional(self): q = QuantumRegister(4) c = ClassicalRegister(4) circ = QuantumCircuit(q, c, name='circ') circ.t(q[1]) circ.u3(0.1, 0.2, (- 0.2), q[0]) circ.barrier() circ.measure(q[0], c[0]) circ.rz(0.8, q[0]).c_if(c, 6) inst = circ.to_instruction() self.assertRaises(QiskitError, inst.inverse) def test_inverse_opaque(self): opaque_gate = Gate(name='crz_2', num_qubits=2, params=[0.5]) self.assertRaises(QiskitError, opaque_gate.inverse)
_module() class TPSPreprocessor(BasePreprocessor): def __init__(self, num_fiducial=20, img_size=(32, 100), rectified_img_size=(32, 100), num_img_channel=1, init_cfg=None): super().__init__(init_cfg=init_cfg) assert isinstance(num_fiducial, int) assert (num_fiducial > 0) assert isinstance(img_size, tuple) assert isinstance(rectified_img_size, tuple) assert isinstance(num_img_channel, int) self.num_fiducial = num_fiducial self.img_size = img_size self.rectified_img_size = rectified_img_size self.num_img_channel = num_img_channel self.LocalizationNetwork = LocalizationNetwork(self.num_fiducial, self.num_img_channel) self.GridGenerator = GridGenerator(self.num_fiducial, self.rectified_img_size) def forward(self, batch_img): batch_C_prime = self.LocalizationNetwork(batch_img) build_P_prime = self.GridGenerator.build_P_prime(batch_C_prime, batch_img.device) build_P_prime_reshape = build_P_prime.reshape([build_P_prime.size(0), self.rectified_img_size[0], self.rectified_img_size[1], 2]) batch_rectified_img = F.grid_sample(batch_img, build_P_prime_reshape, padding_mode='border', align_corners=True) return batch_rectified_img
def train(args, train_dataset, model, tokenizer): if (args.local_rank in [(- 1), 0]): tb_writer = SummaryWriter() args.train_batch_size = (args.per_gpu_train_batch_size * max(1, args.n_gpu)) train_sampler = (RandomSampler(train_dataset) if (args.local_rank == (- 1)) else DistributedSampler(train_dataset)) train_dataloader = DataLoader(train_dataset, sampler=train_sampler, batch_size=args.train_batch_size) if (args.max_steps > 0): t_total = args.max_steps args.num_train_epochs = ((args.max_steps // (len(train_dataloader) // args.gradient_accumulation_steps)) + 1) else: t_total = ((len(train_dataloader) // args.gradient_accumulation_steps) * args.num_train_epochs) no_decay = ['bias', 'LayerNorm.weight'] optimizer_grouped_parameters = [{'params': [p for (n, p) in model.named_parameters() if (not any(((nd in n) for nd in no_decay)))], 'weight_decay': args.weight_decay}, {'params': [p for (n, p) in model.named_parameters() if any(((nd in n) for nd in no_decay))], 'weight_decay': 0.0}] optimizer = AdamW(optimizer_grouped_parameters, lr=args.learning_rate, eps=args.adam_epsilon) scheduler = get_linear_schedule_with_warmup(optimizer, num_warmup_steps=args.warmup_steps, num_training_steps=t_total) if (os.path.isfile(os.path.join(args.model_name_or_path, 'optimizer.pt')) and os.path.isfile(os.path.join(args.model_name_or_path, 'scheduler.pt'))): optimizer.load_state_dict(torch.load(os.path.join(args.model_name_or_path, 'optimizer.pt'))) scheduler.load_state_dict(torch.load(os.path.join(args.model_name_or_path, 'scheduler.pt'))) if args.fp16: try: from apex import amp except ImportError: raise ImportError('Please install apex from to use fp16 training.') (model, optimizer) = amp.initialize(model, optimizer, opt_level=args.fp16_opt_level) if (args.n_gpu > 1): model = torch.nn.DataParallel(model) if (args.local_rank != (- 1)): model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.local_rank], output_device=args.local_rank, find_unused_parameters=True) logger.info('***** Running training *****') logger.info(' Num examples = %d', len(train_dataset)) logger.info(' Num Epochs = %d', args.num_train_epochs) logger.info(' Instantaneous batch size per GPU = %d', args.per_gpu_train_batch_size) logger.info(' Total train batch size (w. parallel, distributed & accumulation) = %d', ((args.train_batch_size * args.gradient_accumulation_steps) * (torch.distributed.get_world_size() if (args.local_rank != (- 1)) else 1))) logger.info(' Gradient Accumulation steps = %d', args.gradient_accumulation_steps) logger.info(' Total optimization steps = %d', t_total) global_step = 0 epochs_trained = 0 steps_trained_in_current_epoch = 0 if os.path.exists(args.model_name_or_path): global_step = int(args.model_name_or_path.split('-')[(- 1)].split('/')[0]) epochs_trained = (global_step // (len(train_dataloader) // args.gradient_accumulation_steps)) steps_trained_in_current_epoch = (global_step % (len(train_dataloader) // args.gradient_accumulation_steps)) logger.info(' Continuing training from checkpoint, will skip to saved global_step') logger.info(' Continuing training from epoch %d', epochs_trained) logger.info(' Continuing training from global step %d', global_step) logger.info(' Will skip the first %d steps in the first epoch', steps_trained_in_current_epoch) (tr_loss, logging_loss) = (0.0, 0.0) model.zero_grad() train_iterator = trange(epochs_trained, int(args.num_train_epochs), desc='Epoch', disable=(args.local_rank not in [(- 1), 0])) set_seed(args) for _ in train_iterator: epoch_iterator = tqdm(train_dataloader, desc='Iteration', disable=(args.local_rank not in [(- 1), 0])) for (step, batch) in enumerate(epoch_iterator): if (steps_trained_in_current_epoch > 0): steps_trained_in_current_epoch -= 1 continue model.train() batch = tuple((t.to(args.device) for t in batch)) inputs = {'input_ids': batch[0], 'attention_mask': batch[1], 'labels': batch[3]} if (args.model_type != 'distilbert'): inputs['token_type_ids'] = (batch[2] if (args.model_type in ['bert', 'xlnet', 'albert']) else None) outputs = model(**inputs) loss = outputs[0] if (args.n_gpu > 1): loss = loss.mean() if (args.gradient_accumulation_steps > 1): loss = (loss / args.gradient_accumulation_steps) if args.fp16: with amp.scale_loss(loss, optimizer) as scaled_loss: scaled_loss.backward() else: loss.backward() tr_loss += loss.item() if (((step + 1) % args.gradient_accumulation_steps) == 0): if args.fp16: torch.nn.utils.clip_grad_norm_(amp.master_params(optimizer), args.max_grad_norm) else: torch.nn.utils.clip_grad_norm_(model.parameters(), args.max_grad_norm) optimizer.step() scheduler.step() model.zero_grad() global_step += 1 if ((args.local_rank in [(- 1), 0]) and (args.logging_steps > 0) and ((global_step % args.logging_steps) == 0)): logs = {} if ((args.local_rank == (- 1)) and args.evaluate_during_training): results = evaluate(args, model, tokenizer) for (key, value) in results.items(): eval_key = 'eval_{}'.format(key) logs[eval_key] = value loss_scalar = ((tr_loss - logging_loss) / args.logging_steps) learning_rate_scalar = scheduler.get_lr()[0] logs['learning_rate'] = learning_rate_scalar logs['loss'] = loss_scalar logging_loss = tr_loss for (key, value) in logs.items(): tb_writer.add_scalar(key, value, global_step) print(json.dumps({**logs, **{'step': global_step}})) if ((args.local_rank in [(- 1), 0]) and (args.save_steps > 0) and ((global_step % args.save_steps) == 0)): output_dir = os.path.join(args.output_dir, 'checkpoint-{}'.format(global_step)) if (not os.path.exists(output_dir)): os.makedirs(output_dir) model_to_save = (model.module if hasattr(model, 'module') else model) model_to_save.save_pretrained(output_dir) tokenizer.save_pretrained(output_dir) torch.save(args, os.path.join(output_dir, 'training_args.bin')) logger.info('Saving model checkpoint to %s', output_dir) torch.save(optimizer.state_dict(), os.path.join(output_dir, 'optimizer.pt')) torch.save(scheduler.state_dict(), os.path.join(output_dir, 'scheduler.pt')) logger.info('Saving optimizer and scheduler states to %s', output_dir) if ((args.max_steps > 0) and (global_step > args.max_steps)): epoch_iterator.close() break if ((args.max_steps > 0) and (global_step > args.max_steps)): train_iterator.close() break if (args.local_rank in [(- 1), 0]): tb_writer.close() return (global_step, (tr_loss / global_step))
class AdExchangeAgent(ph.Agent): (frozen=True) class AdExchangeView(ph.AgentView): users_info: dict def __init__(self, agent_id: str, publisher_id: str, advertiser_ids: Iterable=tuple(), strategy: str='first'): super().__init__(agent_id) self.publisher_id = publisher_id self.advertiser_ids = advertiser_ids self.strategy = strategy def view(self, neighbour_id=None) -> ph.View: if (neighbour_id and neighbour_id.startswith('ADV')): return self.AdExchangeView(users_info={1: {'age': 18, 'zipcode': 94025}, 2: {'age': 40, 'zipcode': 90250}}) else: return super().view(neighbour_id) .msg_handler(ImpressionRequest) def handle_impression_request(self, _ctx: ph.Context, msg: ph.Message[ImpressionRequest]): logger.debug('AdExchange impression request %s', msg) return [(adv_id, msg.payload) for adv_id in self.advertiser_ids] def handle_batch(self, ctx: ph.Context, batch: Sequence[ph.Message]): bids = [] msgs = [] for message in batch: if isinstance(message.payload, Bid): bids.append(message) else: msgs += self.handle_message(ctx, message) if (len(bids) > 0): msgs += self.auction(bids) return msgs def auction(self, bids: Sequence[ph.Message[Bid]]): if (self.strategy == 'first'): (winner, cost) = self._first_price_auction(bids) elif (self.strategy == 'second'): (winner, cost) = self._second_price_auction(bids) else: raise ValueError(f'Unknown auction strategy: {self.strategy}') logger.debug('AdExchange auction done winner: %s cost: %s', winner, cost) msgs = [] advertiser_ids = [m.sender_id for m in bids] msgs.append((self.publisher_id, Ads(advertiser_id=winner.sender_id, theme=winner.payload.theme, user_id=winner.payload.user_id))) for adv_id in advertiser_ids: adv_cost = (cost if (adv_id == winner.sender_id) else 0.0) msgs.append((adv_id, AuctionResult(cost=adv_cost, winning_bid=winner.payload.bid))) return msgs def _first_price_auction(self, bids: Sequence[ph.Message[Bid]]): sorted_bids = sorted(bids, key=(lambda m: m.payload.bid), reverse=True) winner = sorted_bids[0] cost = sorted_bids[0].payload.bid return (winner, cost) def _second_price_auction(self, bids: Sequence[ph.Message[Bid]]): sorted_bids = sorted(bids, key=(lambda m: m.payload.bid), reverse=True) winner = sorted_bids[0] cost = (sorted_bids[1].payload.bid if (len(bids) > 1) else sorted_bids[0].payload.bid) return (winner, cost)
def get_prefix_allowed_tokens_fn(model, split_token='|', title_trie: Trie=None): return _get_end_to_end_prefix_allowed_tokens_fn((lambda x: model.encode(x).tolist()), (lambda x: model.decode(torch.tensor(x))), model.model.decoder.dictionary.bos(), model.model.decoder.dictionary.pad(), model.model.decoder.dictionary.eos(), len(model.model.decoder.dictionary), split_token, title_trie)
def test_starred_assignment_in_middle(): run_cell('a, b, c, d, e = 1, 2, 3, 4, 5') run_cell('x, *star, y = [a, b, c, d, e]') run_cell('a += 1') run_cell('logging.info(x)') assert_detected() run_cell('logging.info(star)') assert_not_detected() run_cell('logging.info(y)') assert_not_detected() run_cell('e += 1') run_cell('logging.info(y)') assert_detected() run_cell('logging.info(star)') assert_not_detected() run_cell('c += 1') run_cell('logging.info(star)') assert_detected()
class nnUNetTrainerV2(nnUNetTrainer): def __init__(self, plans_file, fold, output_folder=None, dataset_directory=None, batch_dice=True, stage=None, unpack_data=True, deterministic=True, fp16=False): super().__init__(plans_file, fold, output_folder, dataset_directory, batch_dice, stage, unpack_data, deterministic, fp16) self.max_num_epochs = 1000 self.initial_lr = 0.01 self.deep_supervision_scales = None self.ds_loss_weights = None self.pin_memory = True def initialize(self, training=True, force_load_plans=False): if (not self.was_initialized): maybe_mkdir_p(self.output_folder) if (force_load_plans or (self.plans is None)): self.load_plans_file() self.process_plans(self.plans) self.setup_DA_params() net_numpool = len(self.net_num_pool_op_kernel_sizes) weights = np.array([(1 / (2 ** i)) for i in range(net_numpool)]) mask = np.array(([True] + [(True if (i < (net_numpool - 1)) else False) for i in range(1, net_numpool)])) weights[(~ mask)] = 0 weights = (weights / weights.sum()) self.ds_loss_weights = weights self.loss = MultipleOutputLoss2(self.loss, self.ds_loss_weights) self.folder_with_preprocessed_data = join(self.dataset_directory, (self.plans['data_identifier'] + ('_stage%d' % self.stage))) if training: (self.dl_tr, self.dl_val) = self.get_basic_generators() if self.unpack_data: print('unpacking dataset') unpack_dataset(self.folder_with_preprocessed_data) print('done') else: print('INFO: Not unpacking data! Training may be slow due to that. Pray you are not using 2d or you will wait all winter for your model to finish!') (self.tr_gen, self.val_gen) = get_moreDA_augmentation(self.dl_tr, self.dl_val, self.data_aug_params['patch_size_for_spatialtransform'], self.data_aug_params, deep_supervision_scales=self.deep_supervision_scales, pin_memory=self.pin_memory, use_nondetMultiThreadedAugmenter=False) self.print_to_log_file(('TRAINING KEYS:\n %s' % str(self.dataset_tr.keys())), also_print_to_console=False) self.print_to_log_file(('VALIDATION KEYS:\n %s' % str(self.dataset_val.keys())), also_print_to_console=False) else: pass self.initialize_network() self.initialize_optimizer_and_scheduler() assert isinstance(self.network, (SegmentationNetwork, nn.DataParallel)) else: self.print_to_log_file('self.was_initialized is True, not running self.initialize again') self.was_initialized = True def initialize_network(self): if self.threeD: conv_op = nn.Conv3d dropout_op = nn.Dropout3d norm_op = nn.InstanceNorm3d else: conv_op = nn.Conv2d dropout_op = nn.Dropout2d norm_op = nn.InstanceNorm2d norm_op_kwargs = {'eps': 1e-05, 'affine': True} dropout_op_kwargs = {'p': 0, 'inplace': True} net_nonlin = nn.LeakyReLU net_nonlin_kwargs = {'negative_slope': 0.01, 'inplace': True} self.network = Generic_UNet(self.num_input_channels, self.base_num_features, self.num_classes, len(self.net_num_pool_op_kernel_sizes), self.conv_per_stage, 2, conv_op, norm_op, norm_op_kwargs, dropout_op, dropout_op_kwargs, net_nonlin, net_nonlin_kwargs, True, False, (lambda x: x), InitWeights_He(0.01), self.net_num_pool_op_kernel_sizes, self.net_conv_kernel_sizes, False, True, True) if torch.cuda.is_available(): self.network.cuda() self.network.inference_apply_nonlin = softmax_helper def initialize_optimizer_and_scheduler(self): assert (self.network is not None), 'self.initialize_network must be called first' self.optimizer = torch.optim.SGD(self.network.parameters(), self.initial_lr, weight_decay=self.weight_decay, momentum=0.99, nesterov=True) self.lr_scheduler = None def run_online_evaluation(self, output, target): target = target[0] output = output[0] return super().run_online_evaluation(output, target) def validate(self, do_mirroring: bool=True, use_sliding_window: bool=True, step_size: float=0.5, save_softmax: bool=True, use_gaussian: bool=True, overwrite: bool=True, validation_folder_name: str='validation_raw', debug: bool=False, all_in_gpu: bool=False, segmentation_export_kwargs: dict=None, run_postprocessing_on_folds: bool=True): ds = self.network.do_ds self.network.do_ds = False ret = super().validate(do_mirroring=do_mirroring, use_sliding_window=use_sliding_window, step_size=step_size, save_softmax=save_softmax, use_gaussian=use_gaussian, overwrite=overwrite, validation_folder_name=validation_folder_name, debug=debug, all_in_gpu=all_in_gpu, segmentation_export_kwargs=segmentation_export_kwargs, run_postprocessing_on_folds=run_postprocessing_on_folds) self.network.do_ds = ds return ret def predict_preprocessed_data_return_seg_and_softmax(self, data: np.ndarray, do_mirroring: bool=True, mirror_axes: Tuple[int]=None, use_sliding_window: bool=True, step_size: float=0.5, use_gaussian: bool=True, pad_border_mode: str='constant', pad_kwargs: dict=None, all_in_gpu: bool=False, verbose: bool=True, mixed_precision=True) -> Tuple[(np.ndarray, np.ndarray)]: ds = self.network.do_ds self.network.do_ds = False ret = super().predict_preprocessed_data_return_seg_and_softmax(data, do_mirroring=do_mirroring, mirror_axes=mirror_axes, use_sliding_window=use_sliding_window, step_size=step_size, use_gaussian=use_gaussian, pad_border_mode=pad_border_mode, pad_kwargs=pad_kwargs, all_in_gpu=all_in_gpu, verbose=verbose, mixed_precision=mixed_precision) self.network.do_ds = ds return ret def run_iteration(self, data_generator, do_backprop=True, run_online_evaluation=False): data_dict = next(data_generator) data = data_dict['data'] target = data_dict['target'] data = maybe_to_torch(data) target = maybe_to_torch(target) if torch.cuda.is_available(): data = to_cuda(data) target = to_cuda(target) self.optimizer.zero_grad() if self.fp16: with autocast(): output = self.network(data) del data l = self.loss(output, target) if do_backprop: self.amp_grad_scaler.scale(l).backward() self.amp_grad_scaler.unscale_(self.optimizer) torch.nn.utils.clip_grad_norm_(self.network.parameters(), 12) self.amp_grad_scaler.step(self.optimizer) self.amp_grad_scaler.update() else: output = self.network(data) del data l = self.loss(output, target) if do_backprop: l.backward() torch.nn.utils.clip_grad_norm_(self.network.parameters(), 12) self.optimizer.step() if run_online_evaluation: self.run_online_evaluation(output, target) del target return l.detach().cpu().numpy() def do_split(self): if (self.fold == 'all'): tr_keys = val_keys = list(self.dataset.keys()) else: splits_file = join(self.dataset_directory, 'splits_final.pkl') if (not isfile(splits_file)): splits = [] all_keys_sorted = np.sort(list(self.dataset.keys())) kfold = KFold(n_splits=3, shuffle=True, random_state=12345) self.print_to_log_file('Creating new {}-fold cross-validation split...'.format(kfold)) for (i, (train_idx, test_idx)) in enumerate(kfold.split(all_keys_sorted)): train_keys = np.array(all_keys_sorted)[train_idx] test_keys = np.array(all_keys_sorted)[test_idx] splits.append(OrderedDict()) splits[(- 1)]['train'] = train_keys splits[(- 1)]['val'] = test_keys save_pickle(splits, splits_file) else: self.print_to_log_file('Using splits from existing split file:', splits_file) splits = load_pickle(splits_file) self.print_to_log_file(('The split file contains %d splits.' % len(splits))) self.print_to_log_file(('Desired fold for training: %d' % self.fold)) if (self.fold < len(splits)): tr_keys = splits[self.fold]['train'] val_keys = splits[self.fold]['val'] self.print_to_log_file(('This split has %d training and %d validation cases.' % (len(tr_keys), len(val_keys)))) else: self.print_to_log_file(('INFO: You requested fold %d for training but splits contain only %d folds. I am now creating a random (but seeded) 80:20 split!' % (self.fold, len(splits)))) rnd = np.random.RandomState(seed=(12345 + self.fold)) keys = np.sort(list(self.dataset.keys())) idx_tr = rnd.choice(len(keys), int((len(keys) * 0.8)), replace=False) idx_val = [i for i in range(len(keys)) if (i not in idx_tr)] tr_keys = [keys[i] for i in idx_tr] val_keys = [keys[i] for i in idx_val] self.print_to_log_file(('This random 80:20 split has %d training and %d validation cases.' % (len(tr_keys), len(val_keys)))) tr_keys.sort() val_keys.sort() self.dataset_tr = OrderedDict() for i in tr_keys: self.dataset_tr[i] = self.dataset[i] self.dataset_val = OrderedDict() for i in val_keys: self.dataset_val[i] = self.dataset[i] def setup_DA_params(self): self.deep_supervision_scales = ([[1, 1, 1]] + list((list(i) for i in (1 / np.cumprod(np.vstack(self.net_num_pool_op_kernel_sizes), axis=0))))[:(- 1)]) if self.threeD: self.data_aug_params = default_3D_augmentation_params self.data_aug_params['rotation_x'] = (((((- 30.0) / 360) * 2.0) * np.pi), (((30.0 / 360) * 2.0) * np.pi)) self.data_aug_params['rotation_y'] = (((((- 30.0) / 360) * 2.0) * np.pi), (((30.0 / 360) * 2.0) * np.pi)) self.data_aug_params['rotation_z'] = (((((- 30.0) / 360) * 2.0) * np.pi), (((30.0 / 360) * 2.0) * np.pi)) if self.do_dummy_2D_aug: self.data_aug_params['dummy_2D'] = True self.print_to_log_file('Using dummy2d data augmentation') self.data_aug_params['elastic_deform_alpha'] = default_2D_augmentation_params['elastic_deform_alpha'] self.data_aug_params['elastic_deform_sigma'] = default_2D_augmentation_params['elastic_deform_sigma'] self.data_aug_params['rotation_x'] = default_2D_augmentation_params['rotation_x'] else: self.do_dummy_2D_aug = False if ((max(self.patch_size) / min(self.patch_size)) > 1.5): default_2D_augmentation_params['rotation_x'] = (((((- 15.0) / 360) * 2.0) * np.pi), (((15.0 / 360) * 2.0) * np.pi)) self.data_aug_params = default_2D_augmentation_params self.data_aug_params['mask_was_used_for_normalization'] = self.use_mask_for_norm if self.do_dummy_2D_aug: self.basic_generator_patch_size = get_patch_size(self.patch_size[1:], self.data_aug_params['rotation_x'], self.data_aug_params['rotation_y'], self.data_aug_params['rotation_z'], self.data_aug_params['scale_range']) self.basic_generator_patch_size = np.array(([self.patch_size[0]] + list(self.basic_generator_patch_size))) else: self.basic_generator_patch_size = get_patch_size(self.patch_size, self.data_aug_params['rotation_x'], self.data_aug_params['rotation_y'], self.data_aug_params['rotation_z'], self.data_aug_params['scale_range']) self.data_aug_params['scale_range'] = (0.7, 1.4) self.data_aug_params['do_elastic'] = False self.data_aug_params['selected_seg_channels'] = [0] self.data_aug_params['patch_size_for_spatialtransform'] = self.patch_size self.data_aug_params['num_cached_per_thread'] = 2 def maybe_update_lr(self, epoch=None): if (epoch is None): ep = (self.epoch + 1) else: ep = epoch self.optimizer.param_groups[0]['lr'] = poly_lr(ep, self.max_num_epochs, self.initial_lr, 0.9) self.print_to_log_file('lr:', np.round(self.optimizer.param_groups[0]['lr'], decimals=6)) def on_epoch_end(self): super().on_epoch_end() continue_training = (self.epoch < self.max_num_epochs) if (self.epoch == 100): if (self.all_val_eval_metrics[(- 1)] == 0): self.optimizer.param_groups[0]['momentum'] = 0.95 self.network.apply(InitWeights_He(0.01)) self.print_to_log_file('At epoch 100, the mean foreground Dice was 0. This can be caused by a too high momentum. High momentum (0.99) is good for datasets where it works, but sometimes causes issues such as this one. Momentum has now been reduced to 0.95 and network weights have been reinitialized') return continue_training def run_training(self): self.maybe_update_lr(self.epoch) ds = self.network.do_ds self.network.do_ds = True ret = super().run_training() self.network.do_ds = ds return ret
class FlaxLogitsWarper(metaclass=DummyObject): _backends = ['flax'] def __init__(self, *args, **kwargs): requires_backends(self, ['flax'])
def check_md5_hash(path, md5_hash): computed_md5_hash = hashlib.md5() with open(path, 'rb') as f: for chunk in iter((lambda : f.read(4096)), b''): computed_md5_hash.update(chunk) computed_md5_hash = computed_md5_hash.hexdigest() if (md5_hash != computed_md5_hash): print('MD5 mismatch for {}: {} == {}'.format(path, md5_hash, computed_md5_hash)) sys.exit(1)
class Superpixels(meta.Augmenter): def __init__(self, p_replace=0, n_segments=100, max_size=128, interpolation='linear', name=None, deterministic=False, random_state=None): super(Superpixels, self).__init__(name=name, deterministic=deterministic, random_state=random_state) self.p_replace = iap.handle_probability_param(p_replace, 'p_replace', tuple_to_uniform=True, list_to_choice=True) self.n_segments = iap.handle_discrete_param(n_segments, 'n_segments', value_range=(1, None), tuple_to_uniform=True, list_to_choice=True, allow_floats=False) self.max_size = max_size self.interpolation = interpolation def _augment_images(self, images, random_state, parents, hooks): iadt.gate_dtypes(images, allowed=['bool', 'uint8', 'uint16', 'uint32', 'uint64', 'int8', 'int16', 'int32', 'int64'], disallowed=['uint128', 'uint256', 'int128', 'int256', 'float16', 'float32', 'float64', 'float96', 'float128', 'float256'], augmenter=self) nb_images = len(images) rss = ia.derive_random_states(random_state, (1 + nb_images)) n_segments_samples = self.n_segments.draw_samples((nb_images,), random_state=rss[0]) for (i, (image, rs)) in enumerate(zip(images, rss[1:])): replace_samples = self.p_replace.draw_samples((n_segments_samples[i],), random_state=rs) if (np.max(replace_samples) == 0): pass else: image = images[i] (min_value, _center_value, max_value) = iadt.get_value_range_of_dtype(image.dtype) orig_shape = image.shape if (self.max_size is not None): size = max(image.shape[0], image.shape[1]) if (size > self.max_size): resize_factor = (self.max_size / size) (new_height, new_width) = (int((image.shape[0] * resize_factor)), int((image.shape[1] * resize_factor))) image = ia.imresize_single_image(image, (new_height, new_width), interpolation=self.interpolation) image_sp = np.copy(image) segments = segmentation.slic(image, n_segments=n_segments_samples[i], compactness=10) nb_channels = image.shape[2] for c in sm.xrange(nb_channels): regions = measure.regionprops((segments + 1), intensity_image=image[(..., c)]) for (ridx, region) in enumerate(regions): if (replace_samples[(ridx % len(replace_samples))] >= 0.5): mean_intensity = region.mean_intensity image_sp_c = image_sp[(..., c)] if (image_sp_c.dtype.kind in ['i', 'u', 'b']): value = int(np.round(mean_intensity)) value = min(max(value, min_value), max_value) image_sp_c[(segments == ridx)] = value else: image_sp_c[(segments == ridx)] = mean_intensity if (orig_shape != image.shape): image_sp = ia.imresize_single_image(image_sp, orig_shape[0:2], interpolation=self.interpolation) images[i] = image_sp return images def _augment_heatmaps(self, heatmaps, random_state, parents, hooks): return heatmaps def _augment_keypoints(self, keypoints_on_images, random_state, parents, hooks): return keypoints_on_images def get_parameters(self): return [self.p_replace, self.n_segments, self.max_size, self.interpolation]
def load_clip(device): model = CLIPModel.from_pretrained('openai/clip-vit-base-patch32').to(device) processor = CLIPProcessor.from_pretrained('openai/clip-vit-base-patch32') tokenizer = CLIPTokenizer.from_pretrained('openai/clip-vit-base-patch32') return (model, processor, tokenizer)
def ndc_projection(x=0.1, n=1.0, f=50.0): return np.array([[(n / x), 0, 0, 0], [0, (n / (- x)), 0, 0], [0, 0, ((- (f + n)) / (f - n)), ((- ((2 * f) * n)) / (f - n))], [0, 0, (- 1), 0]]).astype(np.float32)
_arg_scope def mobilenet(input_tensor, num_classes=1001, depth_multiplier=1.0, scope='MobilenetV2', conv_defs=None, finegrain_classification_mode=False, min_depth=None, divisible_by=None, activation_fn=None, **kwargs): if (conv_defs is None): conv_defs = V2_DEF if ('multiplier' in kwargs): raise ValueError('mobilenetv2 doesn\'t support generic multiplier parameter use "depth_multiplier" instead.') if finegrain_classification_mode: conv_defs = copy.deepcopy(conv_defs) if (depth_multiplier < 1): conv_defs['spec'][(- 1)].params['num_outputs'] /= depth_multiplier if activation_fn: conv_defs = copy.deepcopy(conv_defs) defaults = conv_defs['defaults'] conv_defaults = defaults[(slim.conv2d, slim.fully_connected, slim.separable_conv2d)] conv_defaults['activation_fn'] = activation_fn depth_args = {} if (min_depth is not None): depth_args['min_depth'] = min_depth if (divisible_by is not None): depth_args['divisible_by'] = divisible_by with slim.arg_scope((lib.depth_multiplier,), **depth_args): return lib.mobilenet(input_tensor, num_classes=num_classes, conv_defs=conv_defs, scope=scope, multiplier=depth_multiplier, **kwargs)
class ptb_rum_single_config(object): cell = 'rum' num_steps = 150 learning_rate = 0.002 T_norm = 1.0 num_layers = 1 init_scale = 0.01 max_grad_norm = 1.0 cell_size = 2000 embed_size = 128 max_epoch = 100 max_max_epoch = max_epoch keep_prob = 0.65 zoneout_h = 0.9 lr_decay = 0.1 batch_size = 128 vocab_size = 50 use_layer_norm = True use_zoneout = True dataset = 'ptb'
class TestTranslation(unittest.TestCase): def setUp(self): logging.disable(logging.CRITICAL) def tearDown(self): logging.disable(logging.NOTSET) def test_fconv(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_fconv') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'fconv_iwslt_de_en') generate_main(data_dir) def test_raw(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_fconv_raw') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir, ['--dataset-impl', 'raw']) train_translation_model(data_dir, 'fconv_iwslt_de_en', ['--dataset-impl', 'raw']) generate_main(data_dir, ['--dataset-impl', 'raw']) ((not torch.cuda.is_available()), 'test requires a GPU') def test_fp16(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_fp16') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'fconv_iwslt_de_en', ['--fp16']) generate_main(data_dir) ((not torch.cuda.is_available()), 'test requires a GPU') def test_memory_efficient_fp16(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_memory_efficient_fp16') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'fconv_iwslt_de_en', ['--memory-efficient-fp16']) generate_main(data_dir) def test_update_freq(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_update_freq') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'fconv_iwslt_de_en', ['--update-freq', '3']) generate_main(data_dir) def test_max_positions(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_max_positions') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) with self.assertRaises(Exception) as context: train_translation_model(data_dir, 'fconv_iwslt_de_en', ['--max-target-positions', '5']) self.assertTrue(('skip this example with --skip-invalid-size-inputs-valid-test' in str(context.exception))) train_translation_model(data_dir, 'fconv_iwslt_de_en', ['--max-target-positions', '5', '--skip-invalid-size-inputs-valid-test']) with self.assertRaises(Exception) as context: generate_main(data_dir) generate_main(data_dir, ['--skip-invalid-size-inputs-valid-test']) def test_generation(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_sampling') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'fconv_iwslt_de_en') generate_main(data_dir, ['--sampling', '--temperature', '2', '--beam', '2', '--nbest', '2']) generate_main(data_dir, ['--sampling', '--sampling-topk', '3', '--beam', '2', '--nbest', '2']) generate_main(data_dir, ['--sampling', '--sampling-topp', '0.2', '--beam', '2', '--nbest', '2']) generate_main(data_dir, ['--diversity-rate', '0.5', '--beam', '6']) with self.assertRaises(ValueError): generate_main(data_dir, ['--diverse-beam-groups', '4', '--match-source-len']) generate_main(data_dir, ['--prefix-size', '2']) def test_eval_bleu(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_eval_bleu') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'fconv_iwslt_de_en', ['--eval-bleu', '--eval-bleu-print-samples', '--eval-bleu-remove-bpe', '--eval-bleu-detok', 'space', '--eval-bleu-args', '{"beam": 4, "min_len": 10}']) def test_lstm(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_lstm') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'lstm_wiseman_iwslt_de_en', ['--encoder-layers', '2', '--decoder-layers', '2', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8', '--decoder-out-embed-dim', '8']) generate_main(data_dir) def test_lstm_bidirectional(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_lstm_bidirectional') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'lstm', ['--encoder-layers', '2', '--encoder-bidirectional', '--encoder-hidden-size', '16', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8', '--decoder-out-embed-dim', '8', '--decoder-layers', '2']) generate_main(data_dir) def test_transformer(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_transformer') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'transformer_iwslt_de_en', ['--encoder-layers', '2', '--decoder-layers', '2', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8'], run_validation=True) generate_main(data_dir) ((not torch.cuda.is_available()), 'test requires a GPU') def test_transformer_fp16(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_transformer') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'transformer_iwslt_de_en', ['--encoder-layers', '2', '--decoder-layers', '2', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8', '--fp16'], run_validation=True) generate_main(data_dir) def test_multilingual_transformer(self): encoder_langtok_flags = [[], ['--encoder-langtok', 'src'], ['--encoder-langtok', 'tgt']] decoder_langtok_flags = [[], ['--decoder-langtok']] with contextlib.redirect_stdout(StringIO()): for i in range(len(encoder_langtok_flags)): for j in range(len(decoder_langtok_flags)): enc_ltok_flag = encoder_langtok_flags[i] dec_ltok_flag = decoder_langtok_flags[j] with tempfile.TemporaryDirectory(f'test_multilingual_transformer_{i}_{j}') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, arch='multilingual_transformer', task='multilingual_translation', extra_flags=((['--encoder-layers', '2', '--decoder-layers', '2', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8'] + enc_ltok_flag) + dec_ltok_flag), lang_flags=['--lang-pairs', 'in-out,out-in'], run_validation=True, extra_valid_flags=(enc_ltok_flag + dec_ltok_flag)) generate_main(data_dir, extra_flags=((['--task', 'multilingual_translation', '--lang-pairs', 'in-out,out-in', '--source-lang', 'in', '--target-lang', 'out'] + enc_ltok_flag) + dec_ltok_flag)) def test_transformer_cross_self_attention(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_transformer_cross_self_attention') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'transformer_iwslt_de_en', ['--encoder-layers', '2', '--decoder-layers', '2', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8', '--decoder-embed-dim', '8', '--no-cross-attention', '--cross-self-attention'], run_validation=True) generate_main(data_dir, extra_flags=[]) def test_lightconv(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_lightconv') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'lightconv_iwslt_de_en', ['--encoder-conv-type', 'lightweight', '--decoder-conv-type', 'lightweight', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8']) generate_main(data_dir) def test_dynamicconv(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_dynamicconv') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'lightconv_iwslt_de_en', ['--encoder-conv-type', 'dynamic', '--decoder-conv-type', 'dynamic', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8']) generate_main(data_dir) def test_cmlm_transformer(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_cmlm_transformer') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir, ['--joined-dictionary']) train_translation_model(data_dir, 'cmlm_transformer', ['--apply-bert-init', '--criterion', 'nat_loss', '--noise', 'full_mask', '--pred-length-offset', '--length-loss-factor', '0.1'], task='translation_lev') generate_main(data_dir, ['--task', 'translation_lev', '--iter-decode-max-iter', '9', '--iter-decode-eos-penalty', '0', '--print-step']) ((not torch.cuda.is_available()), 'test requires a GPU') def test_levenshtein_transformer(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_levenshtein_transformer') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir, ['--joined-dictionary']) train_translation_model(data_dir, 'levenshtein_transformer', ['--apply-bert-init', '--early-exit', '6,6,6', '--criterion', 'nat_loss'], task='translation_lev') generate_main(data_dir, ['--task', 'translation_lev', '--iter-decode-max-iter', '9', '--iter-decode-eos-penalty', '0', '--print-step']) def test_nonautoregressive_transformer(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_nonautoregressive_transformer') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir, ['--joined-dictionary']) train_translation_model(data_dir, 'nonautoregressive_transformer', ['--apply-bert-init', '--src-embedding-copy', '--criterion', 'nat_loss', '--noise', 'full_mask', '--pred-length-offset', '--length-loss-factor', '0.1'], task='translation_lev') generate_main(data_dir, ['--task', 'translation_lev', '--iter-decode-max-iter', '0', '--iter-decode-eos-penalty', '0', '--print-step']) def test_iterative_nonautoregressive_transformer(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_iterative_nonautoregressive_transformer') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir, ['--joined-dictionary']) train_translation_model(data_dir, 'iterative_nonautoregressive_transformer', ['--apply-bert-init', '--src-embedding-copy', '--criterion', 'nat_loss', '--noise', 'full_mask', '--stochastic-approx', '--dae-ratio', '0.5', '--train-step', '3'], task='translation_lev') generate_main(data_dir, ['--task', 'translation_lev', '--iter-decode-max-iter', '9', '--iter-decode-eos-penalty', '0', '--print-step']) def test_insertion_transformer(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_insertion_transformer') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir, ['--joined-dictionary']) train_translation_model(data_dir, 'insertion_transformer', ['--apply-bert-init', '--criterion', 'nat_loss', '--noise', 'random_mask'], task='translation_lev') generate_main(data_dir, ['--task', 'translation_lev', '--iter-decode-max-iter', '9', '--iter-decode-eos-penalty', '0', '--print-step']) def test_mixture_of_experts(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_moe') as data_dir: create_dummy_data(data_dir) preprocess_translation_data(data_dir) train_translation_model(data_dir, 'transformer_iwslt_de_en', ['--task', 'translation_moe', '--user-dir', 'examples/translation_moe/src', '--method', 'hMoElp', '--mean-pool-gating-network', '--num-experts', '3', '--encoder-layers', '2', '--decoder-layers', '2', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8']) generate_main(data_dir, ['--task', 'translation_moe', '--user-dir', 'examples/translation_moe/src', '--method', 'hMoElp', '--mean-pool-gating-network', '--num-experts', '3', '--gen-expert', '0']) def test_alignment(self): with contextlib.redirect_stdout(StringIO()): with tempfile.TemporaryDirectory('test_alignment') as data_dir: create_dummy_data(data_dir, alignment=True) preprocess_translation_data(data_dir, ['--align-suffix', 'align']) train_translation_model(data_dir, 'transformer_align', ['--encoder-layers', '2', '--decoder-layers', '2', '--encoder-embed-dim', '8', '--decoder-embed-dim', '8', '--load-alignments', '--alignment-layer', '1', '--criterion', 'label_smoothed_cross_entropy_with_alignment'], run_validation=True) generate_main(data_dir)
class _Calibrator(object): def __init__(self, dataset: Dataset, results_dir: str, image_shape: Tuple[(int, int)], num_samples: int, theta_dims: int): self.dataset = dataset self.results_dir = results_dir self.image_shape = image_shape self.num_samples = num_samples self.theta_dims = theta_dims def __call__(self, video_name: str, writer: _ScoresWriter): features = self.dataset.features_dataset[video_name] labelled_thetas = skeletons_to_angles(features.skeletons, theta_dims=self.theta_dims) labelled_indexes = features.labelled_indexes centerline_accuracy = CenterlineAccuracyCheck(frame_preprocessing=self.dataset.frame_preprocessing, image_shape=self.image_shape) with self.dataset.frames_dataset.open(video_name) as frames: frames_amount = min(self.num_samples, len(labelled_indexes)) random_label_index = np.random.choice(labelled_indexes, frames_amount, replace=False) thetas = labelled_thetas[random_label_index] for (theta, index) in zip(thetas, random_label_index): cur_frame = frames[index] (score, _) = centerline_accuracy(theta=theta, template_skeleton=features.skeletons[index], template_measurements=features.measurements, template_frame=cur_frame, real_frame_orig=cur_frame) writer.add(locals()) results_file = os.path.join(self.results_dir, (video_name + '_calibration.h5')) if os.path.exists(results_file): os.remove(results_file) writer.write(results_file=results_file) logger.info(f'Evaluated known skeletons reconstruction for {video_name}, average score {np.mean(writer.all_scores):.4f}') return results_file
class TFElectraModel(): def __init__(self, *args, **kwargs): requires_tf(self) def from_pretrained(self, *args, **kwargs): requires_tf(self)
class RetinaPolicy(object): def __init__(self, scale_ranges=[1, 1.1], img_size=[512, 512], translate=[0.1, 0.1], rotation=[(- 20), 20], crop_dims=[480, 480], brightness=None): self.scale_ranges = scale_ranges self.img_size = img_size self.translate = translate self.rotation = rotation self.brightness = brightness self.crop_dims = crop_dims def __call__(self, image, mask=None): tf_mask = None (i, j, h, w) = transforms.RandomCrop.get_params(image, self.crop_dims) tf_image = transforms.functional.crop(image, i, j, h, w) if (mask is not None): tf_mask = transforms.functional.crop(mask, i, j, h, w) angle = transforms.RandomRotation.get_params(self.rotation) tf_image = transforms.functional.rotate(tf_image, angle) if (mask is not None): tf_mask = transforms.functional.rotate(tf_mask, angle) params = transforms.RandomAffine.get_params(degrees=[0, 0], translate=self.translate, scale_ranges=[1, 1], img_size=self.img_size, shears=[0, 0]) tf_image = transforms.functional.affine(tf_image, params[0], params[1], params[2], params[3]) if (mask is not None): tf_mask = transforms.functional.affine(tf_mask, params[0], params[1], params[2], params[3]) params = transforms.RandomAffine.get_params(degrees=[0, 0], translate=[0, 0], scale_ranges=self.scale_ranges, img_size=self.img_size, shears=[0, 0]) tf_image = transforms.functional.affine(tf_image, params[0], params[1], params[2], params[3]) if (mask is not None): tf_mask = transforms.functional.affine(tf_mask, params[0], params[1], params[2], params[3]) if (self.brightness is not None): tf = transforms.ColorJitter(brightness=self.brightness) tf_image = tf(tf_image) if (mask is not None): return (tf_image, tf_mask) else: return tf_image def __repr__(self): return 'Retinal Vessel Segmentation Augmentation Policy'
def reduce_df(dataframe, num_per_class): df_list = [] for i in range(10): df_list.append(dataframe.iloc[(i * 5000):((i * 5000) + num_per_class)]) df = pd.concat(df_list) return df
def get_model_fn(n_token, cutoffs, train_bin_sizes, eval_bin_sizes): def model_fn(features, labels, mode, params): is_training = (mode == tf.estimator.ModeKeys.TRAIN) batch_size = params['batch_size'] mems = params['cache'] inp = tf.transpose(features['inputs'], [1, 0]) tgt = tf.transpose(features['labels'], [1, 0]) bin_sizes = (train_bin_sizes if is_training else eval_bin_sizes) if bin_sizes: inp_perms = [tf.transpose(features['inp_mask'], [1, 0])] tgt_perms = [tf.transpose(features['tgt_mask'], [1, 0])] head_tgt = tf.transpose(features['head_labels'], [1, 0]) for b in range(len(bin_sizes)): inp_perm = tf.transpose(features['inp_perm_{}'.format(b)], [1, 0, 2]) tgt_perm = tf.transpose(features['tgt_perm_{}'.format(b)], [1, 0, 2]) inp_perms.append(inp_perm) tgt_perms.append(tgt_perm) else: (inp_perms, tgt_perms, head_tgt) = (None, None, None) if (FLAGS.init == 'uniform'): initializer = tf.initializers.random_uniform(minval=(- FLAGS.init_range), maxval=FLAGS.init_range, seed=None) elif (FLAGS.init == 'normal'): initializer = tf.initializers.random_normal(stddev=FLAGS.init_std, seed=None) proj_initializer = tf.initializers.random_normal(stddev=FLAGS.proj_init_std, seed=None) tie_projs = [False for _ in range((len(cutoffs) + 1))] if FLAGS.proj_share_all_but_first: for i in range(1, len(tie_projs)): tie_projs[i] = True tf.logging.info('Vocab size : {}'.format(n_token)) tf.logging.info('Batch size : {}'.format(batch_size)) (loss, new_mems) = model.transformer(dec_inp=inp, target=tgt, mems=mems, n_token=n_token, n_layer=FLAGS.n_layer, d_model=FLAGS.d_model, d_embed=FLAGS.d_embed, n_head=FLAGS.n_head, d_head=FLAGS.d_head, d_inner=FLAGS.d_inner, dropout=FLAGS.dropout, dropatt=FLAGS.dropatt, initializer=initializer, is_training=is_training, mem_len=FLAGS.mem_len, cutoffs=cutoffs, div_val=FLAGS.div_val, tie_projs=tie_projs, input_perms=inp_perms, target_perms=tgt_perms, head_target=head_tgt, same_length=FLAGS.same_length, clamp_len=FLAGS.clamp_len, use_tpu=FLAGS.use_tpu, untie_r=FLAGS.untie_r, proj_same_dim=FLAGS.proj_same_dim) total_loss = tf.reduce_mean(loss) if (mode == tf.estimator.ModeKeys.EVAL): if FLAGS.use_tpu: with tf.colocate_with(total_loss): total_loss = ((tf.contrib.tpu.cross_replica_sum(total_loss) / FLAGS.num_hosts) / FLAGS.num_core_per_host) metric_loss = tf.tile(tf.reshape(total_loss, [1, 1]), [batch_size, 1]) eval_spec = tf.contrib.tpu.TPUEstimatorSpec(mode=mode, loss=total_loss, eval_metrics=(metric_fn, [metric_loss])) eval_spec.cache = new_mems return eval_spec global_step = tf.train.get_global_step() if (FLAGS.warmup_steps > 0): warmup_lr = ((tf.to_float(global_step) / tf.to_float(FLAGS.warmup_steps)) * FLAGS.learning_rate) else: warmup_lr = 0.0 num_params = np.sum([np.prod(v.shape) for v in tf.trainable_variables()]) tf.logging.info('#params: {}'.format(num_params)) decay_lr = tf.train.cosine_decay(FLAGS.learning_rate, global_step=(global_step - FLAGS.warmup_steps), decay_steps=(FLAGS.train_steps - FLAGS.warmup_steps), alpha=FLAGS.min_lr_ratio) learning_rate = tf.where((global_step < FLAGS.warmup_steps), warmup_lr, decay_lr) if FLAGS.use_tpu: optimizer = tf.contrib.tpu.CrossShardOptimizer(tf.train.AdamOptimizer(learning_rate=learning_rate)) else: optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate) grads_and_vars = optimizer.compute_gradients(total_loss) (gradients, variables) = zip(*grads_and_vars) (clipped, _) = tf.clip_by_global_norm(gradients, FLAGS.clip) train_op = optimizer.apply_gradients(zip(clipped, variables), global_step=tf.train.get_global_step()) train_spec = tf.contrib.tpu.TPUEstimatorSpec(mode=mode, loss=total_loss, train_op=train_op) if (FLAGS.mem_len < FLAGS.tgt_len): new_mems = [new_mems[:FLAGS.mem_len] for mem_t in new_mems] train_spec.cache = new_mems return train_spec return model_fn
def fix_parentheses(string): stack = list() output_string = '' for i in range(len(string)): if (string[i] == '('): stack.append(i) output_string += string[i] elif (string[i] == ')'): if (len(stack) == 0): pass else: output_string += string[i] stack.pop() else: output_string += string[i] for i in range(len(stack)): output_string += ')' return output_string
def test_isotropic_eddington_dehnencore_in_nfw_sigmar(): pot = [potential.NFWPotential(amp=2.3, a=1.3)] denspot = potential.DehnenCoreSphericalPotential(amp=2.5, a=1.15) dfp = eddingtondf(pot=pot, denspot=denspot) numpy.random.seed(10) samp = dfp.sample(n=1000000) tol = 0.08 check_sigmar_against_jeans(samp, pot, tol, dens=(lambda r: denspot.dens(r, 0, use_physical=False)), rmin=(pot[0]._scale / 10.0), rmax=(pot[0]._scale * 10.0), bins=31) return None
def print_cfg(blocks): for block in blocks: print(('[%s]' % block['type'])) for (key, value) in block.items(): if (key != 'type'): print(('%s=%s' % (key, value))) print('')
def test_lambda_metric(): env = MockEnv() metric = ph.metrics.LambdaMetric(extract_fn=(lambda env: env.test_property), train_reduce_fn=(lambda values: np.sum(values)), eval_reduce_fn=(lambda values: (np.sum(values) * 2))) values = [] for _ in range(5): env.step() values.append(metric.extract(env)) assert (metric.reduce(values, mode='train') == 15.0) assert (metric.reduce(values, mode='evaluate') == 30.0)
class ResUNetBN2F(ResUNet2): NORM_TYPE = 'BN' CHANNELS = [None, 16, 32, 64, 128] TR_CHANNELS = [None, 16, 32, 64, 128]
class Encoder(nn.Module): def __init__(self, cin, cout, size=64, nf=64, activation=nn.Tanh): super(Encoder, self).__init__() extra = int((np.log2(size) - 6)) network = [nn.Conv2d(cin, nf, kernel_size=4, stride=2, padding=1, bias=False), nn.ReLU(inplace=True), nn.Conv2d(nf, (nf * 2), kernel_size=4, stride=2, padding=1, bias=False), nn.ReLU(inplace=True), nn.Conv2d((nf * 2), (nf * 4), kernel_size=4, stride=2, padding=1, bias=False), nn.ReLU(inplace=True), nn.Conv2d((nf * 4), (nf * 8), kernel_size=4, stride=2, padding=1, bias=False), nn.ReLU(inplace=True)] for i in range(extra): nf *= 2 network += [nn.Conv2d((nf * 4), (nf * 8), kernel_size=4, stride=2, padding=1, bias=False), nn.ReLU(inplace=True)] network += [nn.Conv2d((nf * 8), (nf * 8), kernel_size=4, stride=1, padding=0, bias=False), nn.ReLU(inplace=True), nn.Conv2d((nf * 8), cout, kernel_size=1, stride=1, padding=0, bias=False)] if (activation is not None): network += [activation()] self.network = nn.Sequential(*network) def forward(self, input): return self.network(input).reshape(input.size(0), (- 1))
_lr_scheduler('cosine', dataclass=CosineLRScheduleConfig) class CosineLRSchedule(FairseqLRScheduler): def __init__(self, cfg: CosineLRScheduleConfig, fairseq_optimizer): super().__init__(cfg, fairseq_optimizer) if (isinstance(cfg.lr, Collection) and (len(cfg.lr) > 1)): raise ValueError(f'Cannot use a fixed learning rate schedule with cosine. Consider --lr-scheduler=fixed instead. ({cfg.lr})') self.max_lr = (cfg.lr[0] if isinstance(cfg.lr, Collection) else cfg.lr) assert (self.max_lr > cfg.min_lr), f'max_lr (={cfg.lr}) must be more than min_lr (={cfg.min_lr})' warmup_end_lr = self.max_lr if (cfg.warmup_init_lr < 0): cfg.warmup_init_lr = cfg.min_lr self.t_mult = cfg.t_mult self.period = cfg.lr_period_updates if (self.period <= 0): assert (cfg.max_update > 0), 'Either --max_update or --lr-period-updates must be set' self.period = (cfg.max_update - cfg.warmup_updates) if (cfg.warmup_updates > 0): self.lr_step = ((warmup_end_lr - cfg.warmup_init_lr) / cfg.warmup_updates) else: self.lr_step = 1 self.warmup_updates = cfg.warmup_updates self.lr_shrink = cfg.lr_shrink self.lr = cfg.warmup_init_lr self.optimizer.set_lr(self.lr) def step(self, epoch, val_loss=None): super().step(epoch, val_loss) return self.optimizer.get_lr() def step_update(self, num_updates): if (num_updates < self.cfg.warmup_updates): self.lr = (self.cfg.warmup_init_lr + (num_updates * self.lr_step)) else: curr_updates = (num_updates - self.cfg.warmup_updates) if (self.t_mult != 1): i = math.floor(math.log((1 - ((curr_updates / self.period) * (1 - self.t_mult))), self.t_mult)) t_i = ((self.t_mult ** i) * self.period) t_curr = (curr_updates - (((1 - (self.t_mult ** i)) / (1 - self.t_mult)) * self.period)) else: i = math.floor((curr_updates / self.period)) t_i = self.period t_curr = (curr_updates - (self.period * i)) lr_shrink = (self.lr_shrink ** i) min_lr = (self.cfg.min_lr * lr_shrink) max_lr = (self.max_lr * lr_shrink) self.lr = (min_lr + ((0.5 * (max_lr - min_lr)) * (1 + math.cos(((math.pi * t_curr) / t_i))))) self.optimizer.set_lr(self.lr) return self.lr
def encode_affinity(n_cpu_core=1, n_gpu=0, cpu_reserved=0, contexts_per_gpu=1, gpu_per_run=1, cpu_per_run=1, cpu_per_worker=1, async_sample=False, sample_gpu_per_run=0, optim_sample_share_gpu=False, hyperthread_offset=None, n_socket=None, run_slot=None, alternating=False, set_affinity=True): affinity_code = f'{n_cpu_core}{N_CPU_CORE}_{n_gpu}{N_GPU}' if (hyperthread_offset is None): hyperthread_offset = get_hyperthread_offset() if (n_socket is None): n_socket = get_n_socket() if (contexts_per_gpu > 1): affinity_code += f'_{contexts_per_gpu}{CONTEXTS_PER_GPU}' if (gpu_per_run > 1): affinity_code += f'_{gpu_per_run}{GPU_PER_RUN}' if (n_gpu == 0): affinity_code += f'_{cpu_per_run}{CPU_PER_RUN}' if (cpu_per_worker > 1): affinity_code += f'_{cpu_per_worker}{CPU_PER_WORKER}' if (hyperthread_offset != n_cpu_core): affinity_code += f'_{hyperthread_offset}{HYPERTHREAD_OFFSET}' if (n_socket > 1): affinity_code += f'_{n_socket}{N_SOCKET}' if (cpu_reserved > 0): affinity_code += f'_{cpu_reserved}{CPU_RESERVED}' if async_sample: affinity_code += f'_1{ASYNC_SAMPLE}' if (sample_gpu_per_run > 0): affinity_code += f'_{sample_gpu_per_run}{SAMPLE_GPU_PER_RUN}' if optim_sample_share_gpu: affinity_code += f'_1{OPTIM_SAMPLE_SHARE_GPU}' if alternating: affinity_code += f'_1{ALTERNATING}' if (not set_affinity): affinity_code += f'_0{SET_AFFINITY}' if (run_slot is not None): assert (run_slot <= ((n_gpu * contexts_per_gpu) // gpu_per_run)) affinity_code = (f'{run_slot}{RUN_SLOT}_' + affinity_code) return affinity_code
_task('speech_to_text_wav2vec_triple_dataset') class SpeechToTextTaskWav2VecTripleDataset(LegacyFairseqTask): def add_args(parser): parser.add_argument('data', help='manifest root path') parser.add_argument('--config-yaml', type=str, default='config.yaml', help='Configuration YAML filename (under manifest root)') parser.add_argument('--max-source-positions', default=6000, type=int, metavar='N', help='max number of tokens in the source sequence') parser.add_argument('--max-target-positions', default=1024, type=int, metavar='N', help='max number of tokens in the target sequence') parser.add_argument('--raw-wav', default=True, type=bool, help='whether to input raw wav file') def __init__(self, args, tgt_dict): super().__init__(args) check_import(import_successful) self.tgt_dict = tgt_dict self.data_cfg = S2TDataConfig(op.join(args.data, args.config_yaml)) def setup_task(cls, args, **kwargs): data_cfg = S2TDataConfig(op.join(args.data, args.config_yaml)) dict_path = op.join(args.data, data_cfg.vocab_filename) if (not op.isfile(dict_path)): raise FileNotFoundError(f'Dict not found: {dict_path}') tgt_dict = Dictionary.load(dict_path) logger.info(f'dictionary size ({data_cfg.vocab_filename}): {len(tgt_dict):,}') if (getattr(args, 'train_subset', None) is not None): if (not all((s.startswith('train') for s in args.train_subset.split(',')))): raise ValueError('Train splits should be named like "train*".') return cls(args, tgt_dict) def build_criterion(self, args): from fairseq import criterions if (self.data_cfg.prepend_tgt_lang_tag and (args.ignore_prefix_size != 1)): raise ValueError('Please set "--ignore-prefix-size 1" since target language ID token is prepended as BOS.') return criterions.build_criterion(args, self) def load_dataset(self, split, epoch=1, combine=False, **kwargs): is_train_split = split.startswith('train') pre_tokenizer = self.build_tokenizer(self.args) bpe_tokenizer = self.build_bpe(self.args) self.datasets[split] = SpeechToTextDatasetCreator.from_tsv(self.args.data, self.data_cfg, split, self.tgt_dict, pre_tokenizer, bpe_tokenizer, is_train_split=is_train_split, epoch=epoch, seed=self.args.seed) def target_dictionary(self): return self.tgt_dict def source_dictionary(self): return None def max_positions(self): return (self.args.max_source_positions, self.args.max_target_positions) def build_model(self, args): args.input_feat_per_channel = self.data_cfg.input_feat_per_channel args.input_channels = self.data_cfg.input_channels return super(SpeechToTextTaskWav2VecTripleDataset, self).build_model(args) def build_generator(self, models, args, seq_gen_cls=None, extra_gen_cls_kwargs=None): if (self.data_cfg.prepend_tgt_lang_tag and (args.prefix_size != 1)): raise ValueError('Please set "--prefix-size 1" since target language ID token is prepended as BOS.') lang_token_ids = {i for (s, i) in self.tgt_dict.indices.items() if SpeechToTextDataset.is_lang_tag(s)} extra_gen_cls_kwargs = {'symbols_to_strip_from_output': lang_token_ids} return super().build_generator(models, args, seq_gen_cls=None, extra_gen_cls_kwargs=extra_gen_cls_kwargs) def build_tokenizer(self, args): logger.info(f'pre-tokenizer: {self.data_cfg.pre_tokenizer}') return encoders.build_tokenizer(Namespace(**self.data_cfg.pre_tokenizer)) def build_bpe(self, args): logger.info(f'tokenizer: {self.data_cfg.bpe_tokenizer}') return encoders.build_bpe(Namespace(**self.data_cfg.bpe_tokenizer)) def get_interactive_tokens_and_lengths(self, lines, encode_fn): n_frames = [get_features_or_waveform(p).shape[0] for p in lines] return (lines, n_frames) def build_dataset_for_inference(self, src_tokens, src_lengths, **kwargs): return SpeechToTextDataset('interactive', False, self.data_cfg, src_tokens, src_lengths)
class TrainSetTransform(): def __init__(self, aug_mode): self.aug_mode = aug_mode self.transform = None if (aug_mode == 0): t = None elif (aug_mode == 1): t = [RandomRotation(max_theta=5, max_theta2=0, axis=np.array([0, 0, 1])), RandomFlip([0.25, 0.25, 0.0])] else: raise NotImplementedError('Unknown aug_mode: {}'.format(aug_mode)) if (t is None): self.transform = None else: self.transform = transforms.Compose(t) def __call__(self, e): if (self.transform is not None): e = self.transform(e) return e
def call_output(cmd): print(f'Executing: {cmd}') ret = check_output(cmd, shell=True) print(ret) return ret
class GatherViewer(MjViewer): def __init__(self, env): self.env = env super(GatherViewer, self).__init__() green_ball_model = MjModel(osp.abspath(osp.join(MODEL_DIR, 'green_ball.xml'))) self.green_ball_renderer = EmbeddedViewer() self.green_ball_model = green_ball_model self.green_ball_renderer.set_model(green_ball_model) red_ball_model = MjModel(osp.abspath(osp.join(MODEL_DIR, 'red_ball.xml'))) self.red_ball_renderer = EmbeddedViewer() self.red_ball_model = red_ball_model self.red_ball_renderer.set_model(red_ball_model) green_ball_highlighted_model = MjModel(osp.abspath(osp.join(MODEL_DIR, 'green_ball_highlighted.xml'))) self.green_ball_highlighted_renderer = EmbeddedViewer() self.green_ball_highlighted_model = green_ball_highlighted_model self.green_ball_highlighted_renderer.set_model(green_ball_highlighted_model) red_ball_highlighted_model = MjModel(osp.abspath(osp.join(MODEL_DIR, 'red_ball_highlighted.xml'))) self.red_ball_highlighted_renderer = EmbeddedViewer() self.red_ball_highlighted_model = red_ball_highlighted_model self.red_ball_highlighted_renderer.set_model(red_ball_highlighted_model) def start(self): super(GatherViewer, self).start() self.green_ball_renderer.start(self.window) self.red_ball_renderer.start(self.window) self.green_ball_highlighted_renderer.start(self.window) self.red_ball_highlighted_renderer.start(self.window) def handle_mouse_move(self, window, xpos, ypos): super(GatherViewer, self).handle_mouse_move(window, xpos, ypos) self.green_ball_renderer.handle_mouse_move(window, xpos, ypos) self.red_ball_renderer.handle_mouse_move(window, xpos, ypos) self.green_ball_highlighted_renderer.handle_mouse_move(window, xpos, ypos) self.red_ball_highlighted_renderer.handle_mouse_move(window, xpos, ypos) def handle_scroll(self, window, x_offset, y_offset): super(GatherViewer, self).handle_scroll(window, x_offset, y_offset) self.green_ball_renderer.handle_scroll(window, x_offset, y_offset) self.red_ball_renderer.handle_scroll(window, x_offset, y_offset) self.green_ball_highlighted_renderer.handle_scroll(window, x_offset, y_offset) self.red_ball_highlighted_renderer.handle_scroll(window, x_offset, y_offset) def render(self): super(GatherViewer, self).render() tmpobjects = mjcore.MJVOBJECTS() mjlib.mjlib.mjv_makeObjects(byref(tmpobjects), 1000) for obj in self.env.objects_in_view: (x, y, typ) = obj qpos = np.zeros_like(self.green_ball_highlighted_model.data.qpos) qpos[(0, 0)] = x qpos[(1, 0)] = y if (typ == APPLE): self.green_ball_highlighted_model.data.qpos = qpos self.green_ball_highlighted_model.forward() self.green_ball_highlighted_renderer.render() mjextra.append_objects(tmpobjects, self.green_ball_highlighted_renderer.objects) else: self.red_ball_highlighted_model.data.qpos = qpos self.red_ball_highlighted_model.forward() self.red_ball_highlighted_renderer.render() mjextra.append_objects(tmpobjects, self.red_ball_highlighted_renderer.objects) for obj in self.env.objects: if (not (obj in self.env.objects_in_view)): (x, y, typ) = obj qpos = np.zeros_like(self.green_ball_model.data.qpos) qpos[(0, 0)] = x qpos[(1, 0)] = y if (typ == APPLE): self.green_ball_model.data.qpos = qpos self.green_ball_model.forward() self.green_ball_renderer.render() mjextra.append_objects(tmpobjects, self.green_ball_renderer.objects) else: self.red_ball_model.data.qpos = qpos self.red_ball_model.forward() self.red_ball_renderer.render() mjextra.append_objects(tmpobjects, self.red_ball_renderer.objects) mjextra.append_objects(tmpobjects, self.objects) mjlib.mjlib.mjv_makeLights(self.model.ptr, self.data.ptr, byref(tmpobjects)) mjlib.mjlib.mjr_render(0, self.get_rect(), byref(tmpobjects), byref(self.ropt), byref(self.cam.pose), byref(self.con)) try: import OpenGL.GL as GL except: return def draw_rect(x, y, width, height): GL.glBegin(GL.GL_QUADS) GL.glVertex2f(x, y) GL.glVertex2f((x + width), y) GL.glVertex2f((x + width), (y + height)) GL.glVertex2f(x, (y + height)) GL.glEnd() def refresh2d(width, height): GL.glViewport(0, 0, width, height) GL.glMatrixMode(GL.GL_PROJECTION) GL.glLoadIdentity() GL.glOrtho(0.0, width, 0.0, height, 0.0, 1.0) GL.glMatrixMode(GL.GL_MODELVIEW) GL.glLoadIdentity() GL.glLoadIdentity() (width, height) = glfw.get_framebuffer_size(self.window) refresh2d(width, height) GL.glDisable(GL.GL_LIGHTING) GL.glEnable(GL.GL_BLEND) GL.glColor4f(0.0, 0.0, 0.0, 0.8) draw_rect(10, 10, 300, 100) (apple_readings, bomb_readings) = self.env.get_readings() for (idx, reading) in enumerate(apple_readings): if (reading > 0): GL.glColor4f(0.0, 1.0, 0.0, reading) draw_rect((20 * (idx + 1)), 10, 5, 50) for (idx, reading) in enumerate(bomb_readings): if (reading > 0): GL.glColor4f(1.0, 0.0, 0.0, reading) draw_rect((20 * (idx + 1)), 60, 5, 50)
class FunctionGetFetches(object): def __init__(self, inputs, outputs, updates=None, name=None, **session_kwargs): updates = (updates or []) if (not isinstance(inputs, (list, tuple))): raise TypeError('`inputs` to a TensorFlow backend function should be a list or tuple.') if (not isinstance(outputs, (list, tuple))): raise TypeError('`outputs` of a TensorFlow backend function should be a list or tuple.') if (not isinstance(updates, (list, tuple))): raise TypeError('`updates` in a TensorFlow backend function should be a list or tuple.') self.inputs = list(inputs) self.outputs = list(outputs) with tf.control_dependencies(self.outputs): updates_ops = [] for update in updates: if isinstance(update, tuple): (p, new_p) = update updates_ops.append(tf.assign(p, new_p)) else: updates_ops.append(update) self.updates_op = tf.group(*updates_ops) self.name = name self.feed_dict = session_kwargs.pop('feed_dict', {}) self.fetches = session_kwargs.pop('fetches', []) if (not isinstance(self.fetches, list)): self.fetches = [self.fetches] self.session_kwargs = session_kwargs def __call__(self, inputs): if (not isinstance(inputs, (list, tuple))): raise TypeError('`inputs` should be a list or tuple.') feed_dict = self.feed_dict.copy() for (tensor, value) in zip(self.inputs, inputs): if is_sparse(tensor): sparse_coo = value.tocoo() indices = np.concatenate((np.expand_dims(sparse_coo.row, 1), np.expand_dims(sparse_coo.col, 1)), 1) value = (indices, sparse_coo.data, sparse_coo.shape) feed_dict[tensor] = value fetches = ((self.outputs + [self.updates_op]) + self.fetches) session = get_session() updated = session.run(fetches=fetches, feed_dict=feed_dict, **self.session_kwargs) return updated
def check_balancing_schedule(balancing_schedule): if callable(balancing_schedule): try: return_value = balancing_schedule({}, {}, 0, 0) except Exception as e: e_args = list(e.args) e_args[0] += BALANCING_SCHEDULE_INFO e.args = tuple(e_args) raise e else: if (not isinstance(return_value, dict)): raise TypeError(((f' The self-defined `balancing_schedule` must return a `dict`,' + f' got {type(return_value)}') + BALANCING_SCHEDULE_INFO)) return balancing_schedule if (balancing_schedule in BALANCING_KIND): return BALANCING_KIND_MAPPING[balancing_schedule] else: balancing_schedule_info = (balancing_schedule if isinstance(balancing_schedule, str) else type(balancing_schedule)) raise TypeError(f"'balancing_schedule' should be one of {BALANCING_KIND} or `callable`, got {balancing_schedule_info}.")
def gen_data_step(decl_file: str, dest: str, rec_limit: int, depth_limit: int, weight_limit: int): lean_cmd = ['lean'] lean_cmd += ['--run'] lean_cmd += ['./src/lean_step.lean'] lean_cmd += list(map(str, [decl_file, dest, rec_limit, depth_limit, weight_limit])) path = Path(dest) stdout_dest = os.path.join(str(path.parent), (str(path.stem) + '.out')) with open(stdout_dest, 'w') as f: subprocess.run(lean_cmd, stdout=f, stderr=f) return decl_file
def w2v_pad(protein, maxlen_, victor_size): tokenizer = text.Tokenizer(num_words=10000, lower=False, filters='\u3000') tokenizer.fit_on_texts(protein) protein_ = sequence.pad_sequences(tokenizer.texts_to_sequences(protein), maxlen=maxlen_) word_index = tokenizer.word_index nb_words = len(word_index) print(nb_words) protVec_model = {} with open('../dataset/embed/protVec_100d_3grams.csv', encoding='utf8') as f: for line in f: values = eval(line).rsplit('\t') word = values[0] coefs = np.asarray(values[1:], dtype='float32') protVec_model[word] = coefs print('add protVec finished....') count = 0 embedding_matrix = np.zeros(((nb_words + 1), victor_size)) for (word, i) in word_index.items(): embedding_glove_vector = (protVec_model[word] if (word in protVec_model) else None) if (embedding_glove_vector is not None): count += 1 embedding_matrix[i] = embedding_glove_vector else: unk_vec = (np.random.random(victor_size) * 0.5) unk_vec = (unk_vec - unk_vec.mean()) embedding_matrix[i] = unk_vec del protVec_model print(embedding_matrix.shape) return (protein_, word_index, embedding_matrix)
class Transmission(xmlr.Object): def __init__(self, name=None, joint=None, actuator=None): self.name = name self.joint = joint self.actuator = actuator
def collate_train_baseline(batch): if (batch[0][(- 1)] is not None): return collate_eval(batch) indice = [b[0] for b in batch] image = torch.stack([b[1] for b in batch]) return (indice, image)
def precision(tp, fp) -> float: predicted_positives = (tp + fp) if (predicted_positives <= 0): return 0 return (tp / predicted_positives)
def train_classifier(classifier: nn.Module, save_dir: str, train: List[Annotation], val: List[Annotation], documents: Dict[(str, List[List[int]])], model_pars: dict, class_interner: Dict[(str, int)], attention_optimizer=None, classifier_optimizer=None) -> Tuple[(nn.Module, dict)]: logging.info(f'Beginning training classifier with {len(train)} annotations, {len(val)} for validation') classifier_output_dir = os.path.join(save_dir, 'classifier') os.makedirs(save_dir, exist_ok=True) os.makedirs(classifier_output_dir, exist_ok=True) model_save_file = os.path.join(classifier_output_dir, 'classifier.pt') epoch_save_file = os.path.join(classifier_output_dir, 'classifier_epoch_data.pt') (train_ids, train_classes, train_queries, train_docs, train_evidence_spans) = convert_for_training(train, documents, class_interner) (val_ids, val_classes, val_queries, val_docs, val_evidence_spans) = convert_for_training(val, documents, class_interner) if (not bool(model_pars['classifier']['has_query'])): train_queries = None val_queries = None device = next(classifier.parameters()).device if (attention_optimizer is None): attention_optimizer = torch.optim.Adam(classifier.parameters(), lr=model_pars['classifier']['lr']) if (classifier_optimizer is None): classifier_optimizer = torch.optim.Adam(classifier.parameters(), lr=model_pars['classifier']['lr']) attention_criterion = nn.BCELoss(reduction='sum') criterion = nn.CrossEntropyLoss(reduction='sum') batch_size = model_pars['classifier']['batch_size'] epochs = model_pars['classifier']['epochs'] attention_epochs = model_pars['classifier']['attention_epochs'] patience = model_pars['classifier']['patience'] max_grad_norm = model_pars['classifier'].get('max_grad_norm', None) class_labels = [k for (k, v) in sorted(class_interner.items())] results = {'attention_train_losses': [], 'attention_val_losses': [], 'train_loss': [], 'train_f1': [], 'train_acc': [], 'val_loss': [], 'val_f1': [], 'val_acc': []} best_attention_epoch = (- 1) best_classifier_epoch = (- 1) best_attention_loss = float('inf') best_classifier_loss = float('inf') best_model_state_dict = None start_attention_epoch = 0 start_classifier_epoch = 0 epoch_data = {} if os.path.exists(epoch_save_file): logging.info(f'Restoring model from {model_save_file}') classifier.load_state_dict(torch.load(model_save_file)) epoch_data = torch.load(epoch_save_file) start_attention_epoch = (epoch_data.get('attention_epoch', (- 1)) + 1) start_classifier_epoch = (epoch_data.get('classifier_epoch', (- 1)) + 1) best_attention_loss = epoch_data.get('best_attention_loss', float('inf')) best_classifier_loss = epoch_data.get('best_classifier_loss', float('inf')) if bool(epoch_data.get('done_attention', 0)): start_attention_epoch = epochs if bool(epoch_data.get('done_classifier', 0)): start_classifier_epoch = epochs results = epoch_data['results'] best_attention_epoch = start_attention_epoch best_classifier_epoch = start_classifier_epoch best_model_state_dict = OrderedDict({k: v.cpu() for (k, v) in classifier.state_dict().items()}) logging.info(f'Restoring training from attention epoch {start_attention_epoch} / {start_classifier_epoch}') logging.info(f'Training classifier attention from epoch {start_attention_epoch} until epoch {attention_epochs}') for attention_epoch in range(start_attention_epoch, attention_epochs): epoch_train_loss = 0 epoch_train_tokens = 0 epoch_val_loss = 0 epoch_val_tokens = 0 for batch_start in range(0, len(train_ids), batch_size): classifier.train() attention_optimizer.zero_grad() if (train_queries is None): queries = None else: queries = train_queries[batch_start:(batch_start + batch_size)] docs = train_docs[batch_start:(batch_start + batch_size)] train_spans = train_evidence_spans[batch_start:(batch_start + batch_size)] (_, _, _, unnormalized_document_attention, _) = classifier(queries, None, docs, return_attentions=True) partially_normalized_document_attention = torch.sigmoid(unnormalized_document_attention.data.squeeze()) train_spans = PaddedSequence.autopad(train_spans, batch_first=True, device=unnormalized_document_attention.data.device) batch_loss = attention_criterion(partially_normalized_document_attention, train_spans.data.float()) epoch_train_loss += batch_loss.item() train_size = torch.sum(train_spans.batch_sizes).item() epoch_train_tokens += train_size batch_loss = (batch_loss / train_size) batch_loss.backward() attention_optimizer.step() results['attention_train_losses'].append((epoch_train_loss / epoch_train_tokens)) logging.info(f'Epoch {attention_epoch} attention train loss {(epoch_train_loss / epoch_train_tokens)}') with torch.no_grad(): classifier.eval() for batch_start in range(0, len(val_ids), batch_size): if (val_queries is None): queries = None else: queries = val_queries[batch_start:(batch_start + batch_size)] docs = val_docs[batch_start:(batch_start + batch_size)] val_spans = val_evidence_spans[batch_start:(batch_start + batch_size)] (_, _, _, unnormalized_document_attention, _) = classifier(queries, None, docs, return_attentions=True) unnormalized_document_attention = torch.sigmoid(unnormalized_document_attention.data) val_spans = PaddedSequence.autopad(val_spans, batch_first=True, device=device) batch_loss = attention_criterion(unnormalized_document_attention.squeeze(), val_spans.data.float()) epoch_val_loss += batch_loss.item() epoch_val_tokens += torch.sum(val_spans.batch_sizes).item() epoch_val_loss = (epoch_val_loss / epoch_val_tokens) results['attention_val_losses'].append(epoch_val_loss) logging.info(f'Epoch {attention_epoch} attention val loss {epoch_val_loss}') if (epoch_val_loss < best_attention_loss): best_model_state_dict = OrderedDict({k: v.cpu() for (k, v) in classifier.state_dict().items()}) best_attention_epoch = attention_epoch best_attention_loss = epoch_val_loss epoch_data['attention_epoch'] = attention_epoch epoch_data['results'] = results epoch_data['best_attention_loss'] = best_attention_loss epoch_data['best_classifier_loss'] = float('inf') epoch_data['done_attention'] = 0 epoch_data['done_classifier'] = 0 torch.save(classifier.state_dict(), model_save_file) torch.save(epoch_data, epoch_save_file) logging.info(f'Epoch {attention_epoch} new best model with val loss {epoch_val_loss}') if ((attention_epoch - best_attention_epoch) > patience): logging.info(f'Exiting after epoch {attention_epoch} due to no improvement') epoch_data['done_attention'] = 1 torch.save(epoch_data, epoch_save_file) break logging.info(f'Training classifier from epoch {start_classifier_epoch} until epoch {epochs}') for classifier_epoch in range(start_classifier_epoch, epochs): epoch_train_loss = 0 epoch_val_loss = 0 train_preds = [] train_truth = [] classifier.train() for batch_start in range(0, len(train_ids), batch_size): classifier.train() classifier_optimizer.zero_grad() targets = train_classes[batch_start:(batch_start + batch_size)] train_truth.extend(targets) targets = torch.tensor(targets, device=device) if (train_queries is not None): queries = train_queries[batch_start:(batch_start + batch_size)] else: queries = None docs = train_docs[batch_start:(batch_start + batch_size)] classes = classifier(queries, None, docs, return_attentions=False) train_preds.extend((x.item() for x in torch.argmax(classes, dim=1))) batch_loss = criterion(classes.squeeze(), targets) epoch_train_loss += batch_loss.item() batch_loss /= len(docs) batch_loss.backward() classifier_optimizer.step() train_accuracy = accuracy_score(train_truth, train_preds) train_f1 = classification_report(train_truth, train_preds, output_dict=True) results['train_loss'].append((epoch_train_loss / len(train_ids))) results['train_acc'].append(train_accuracy) results['train_f1'].append(train_f1) logging.info(f'Epoch {classifier_epoch} train loss {(epoch_train_loss / len(train_ids))}, accuracy: {train_accuracy}, f1: {train_f1}') with torch.no_grad(): classifier.eval() val_preds = [] val_truth = [] for batch_start in range(0, len(val_ids), batch_size): targets = val_classes[batch_start:(batch_start + batch_size)] val_truth.extend(targets) if (val_queries is not None): queries = val_queries[batch_start:(batch_start + batch_size)] else: queries = None docs = val_docs[batch_start:(batch_start + batch_size)] classes = classifier(queries, None, docs, return_attentions=False) targets = torch.tensor(targets, device=classes.device) val_preds.extend((x.item() for x in torch.argmax(classes, dim=1))) batch_loss = criterion(classes, targets) if (not torch.all((batch_loss == batch_loss))): import pdb pdb.set_trace() epoch_val_loss += batch_loss.item() batch_loss /= len(docs) epoch_val_loss /= len(val_ids) val_accuracy = accuracy_score(val_truth, val_preds) val_f1 = classification_report(val_truth, val_preds, output_dict=True) results['val_loss'].append(epoch_val_loss) results['val_acc'].append(val_accuracy) results['val_f1'].append(val_f1) logging.info(f'Epoch {classifier_epoch} val loss {epoch_val_loss}, accuracy: {val_accuracy}, f1: {val_f1}') if (epoch_val_loss < best_classifier_loss): best_model_state_dict = OrderedDict({k: v.cpu() for (k, v) in classifier.state_dict().items()}) best_classifier_epoch = classifier_epoch best_val_loss = epoch_val_loss epoch_data['classifier_epoch'] = classifier_epoch epoch_data['attention_epoch'] = best_attention_epoch epoch_data['best_attention_loss'] = best_attention_loss epoch_data['results'] = results epoch_data['best_classifier_loss'] = best_val_loss epoch_data['done_classifier'] = 0 epoch_data['done_attention'] = 1 torch.save(classifier.state_dict(), model_save_file) torch.save(epoch_data, epoch_save_file) logging.info(f'Epoch {classifier_epoch} new best model with val loss {epoch_val_loss}') if ((classifier_epoch - best_classifier_epoch) > patience): logging.info(f'Exiting after epoch {classifier_epoch} due to no improvement') epoch_data['done_classifier'] = 1 torch.save(epoch_data, epoch_save_file) break return (classifier, results)
() ('-r', '--results', type=click.Path(exists=True), help='Path of results.') ('-t', '--targets', type=click.Path(exists=True), help='Path of targets.') ('--train-labels', type=click.Path(exists=True), default=None, help='Path of labels for training set.') ('-a', type=click.FLOAT, default=0.55, help='Parameter A for propensity score.') ('-b', type=click.FLOAT, default=1.5, help='Parameter B for propensity score.') def main(results, targets, train_labels, a, b): (res, targets) = (np.load(results, allow_pickle=True), np.load(targets, allow_pickle=True)) mlb = MultiLabelBinarizer(sparse_output=True) targets = mlb.fit_transform(targets) print(',3,5:', get_p_1(res, targets, mlb), get_p_3(res, targets, mlb), get_p_5(res, targets, mlb)) print(',3,5:', get_n_1(res, targets, mlb), get_n_3(res, targets, mlb), get_n_5(res, targets, mlb)) if (train_labels is not None): train_labels = np.load(train_labels, allow_pickle=True) inv_w = get_inv_propensity(mlb.transform(train_labels), a, b) print(',3,5:', get_psp_1(res, targets, inv_w, mlb), get_psp_3(res, targets, inv_w, mlb), get_psp_5(res, targets, inv_w, mlb)) print(',3,5:', get_psndcg_1(res, targets, inv_w, mlb), get_psndcg_3(res, targets, inv_w, mlb), get_psndcg_5(res, targets, inv_w, mlb))
def parse_args(): parser = argparse.ArgumentParser(description='Train a segmentor') parser.add_argument('config', help='train config file path') parser.add_argument('--fvcore', action='store_true', default=False) parser.add_argument('--shape', type=int, nargs='+', default=[1024, 1024], help='input image size') args = parser.parse_args() return args
def script_preset_(model: torch.nn.Module): script_submodules_(model, [nn.Dropout, Attention, GlobalAttention, EvoformerBlock], attempt_trace=False, batch_dims=None)
def build_model(config): module_name = config.pop('name') support_dict = ['DBNet', 'CRNN'] assert (module_name in support_dict) module_class = eval(module_name)(**config) return module_class
def _init_dist_pytorch(backend, **kwargs): rank = int(os.environ['RANK']) num_gpus = torch.cuda.device_count() torch.cuda.set_device((rank % num_gpus)) dist.init_process_group(backend=backend, **kwargs) print(f'init distributed in rank {torch.distributed.get_rank()}')
def test_he_uniform(): from lasagne.init import HeUniform sample = HeUniform().sample((300, 200)) assert ((- 0.1) <= sample.min() < (- 0.09)) assert (0.09 < sample.max() <= 0.1)
def shufflenet(): device = torch.device('cpu') cfg_file = 'tests/configs/shufflenet/shufflenet_v1_3g1x.yaml' cfg.merge_from_file(cfg_file) model = build_recognizer(cfg, device) print(model) cfg_file = 'tests/configs/shufflenet/shufflenet_v2_torchvision.yaml' cfg.merge_from_file(cfg_file) model = build_recognizer(cfg, device) print(model) cfg_file = 'tests/configs/shufflenet/shufflenet_v2_x2_0.yaml' cfg.merge_from_file(cfg_file) model = build_recognizer(cfg, device) print(model)
class INSnipClient(SnipClient): def init_optimizer(self): self.optimizer = SGD(self.model.parameters(), lr=INIT_LR, momentum=MOMENTUM, weight_decay=WEIGHT_DECAY) self.optimizer_scheduler = lr_scheduler.StepLR(self.optimizer, step_size=STEP_SIZE, gamma=(0.5 ** (STEP_SIZE / LR_HALF_LIFE))) self.optimizer_wrapper = OptimizerWrapper(self.model, self.optimizer, self.optimizer_scheduler) def init_train_loader(self, tl): self.train_loader = tl
def build_compressed_embedding_pkl(name): embeddings = [] weights_dir = os.path.join(experiment_path, str(experiment_id), 'weights') with open(os.path.join(weights_dir, name), 'r') as f: lines = f.readlines() for line in lines: tokens = line.strip().split() v = [float(x) for x in tokens[1:]] embeddings.append(v) LM = np.array(embeddings) pickle.dump(LM, open(os.path.join(weights_dir, 'CLM.pkl'), 'wb')) print('LM size {} dumped'.format(LM.shape))
class ResNet(nn.Module): def __init__(self, block, layers, num_classes=1000, stride=8): self.inplanes = 128 super().__init__() self.conv1 = nn.Sequential(nn.Conv2d(3, 64, kernel_size=3, stride=2, padding=1, bias=False), norm_layer(64), nn.ReLU(inplace=True), nn.Conv2d(64, 64, kernel_size=3, stride=1, padding=1, bias=False), norm_layer(64), nn.ReLU(inplace=True), nn.Conv2d(64, 128, kernel_size=3, stride=1, padding=1, bias=False)) self.bn1 = norm_layer(self.inplanes) self.relu = nn.ReLU(inplace=True) self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) self.layer1 = self._make_layer(block, 64, layers[0]) self.layer2 = self._make_layer(block, 128, layers[1], stride=2) if (stride == 16): self.layer3 = self._make_layer(block, 256, layers[2], stride=2) self.layer4 = self._make_layer(block, 512, layers[3], stride=1, dilation=2, grids=[1, 2, 4]) elif (stride == 8): self.layer3 = self._make_layer(block, 256, layers[2], stride=1, dilation=2) self.layer4 = self._make_layer(block, 512, layers[3], stride=1, dilation=4, grids=[1, 2, 4]) self.avgpool = nn.AvgPool2d(7, stride=1) self.fc = nn.Linear((512 * block.expansion), num_classes) for m in self.modules(): if isinstance(m, nn.Conv2d): n = ((m.kernel_size[0] * m.kernel_size[1]) * m.out_channels) m.weight.data.normal_(0, math.sqrt((2.0 / n))) elif isinstance(m, _BatchNorm): m.weight.data.fill_(1) if (m.bias is not None): m.bias.data.zero_() def _make_layer(self, block, planes, blocks, stride=1, dilation=1, grids=None): downsample = None if ((stride != 1) or (self.inplanes != (planes * block.expansion))): downsample = nn.Sequential(nn.Conv2d(self.inplanes, (planes * block.expansion), kernel_size=1, stride=stride, bias=False), norm_layer((planes * block.expansion))) layers = [] if (grids is None): grids = ([1] * blocks) if ((dilation == 1) or (dilation == 2)): layers.append(block(self.inplanes, planes, stride, dilation=1, downsample=downsample, previous_dilation=dilation)) elif (dilation == 4): layers.append(block(self.inplanes, planes, stride, dilation=2, downsample=downsample, previous_dilation=dilation)) else: raise RuntimeError('=> unknown dilation size: {}'.format(dilation)) self.inplanes = (planes * block.expansion) for i in range(1, blocks): layers.append(block(self.inplanes, planes, dilation=(dilation * grids[i]), previous_dilation=dilation)) return nn.Sequential(*layers) def forward(self, x): x = self.conv1(x) x = self.bn1(x) x = self.relu(x) x = self.maxpool(x) x = self.layer1(x) x = self.layer2(x) x = self.layer3(x) x = self.layer4(x) x = self.avgpool(x) x = x.view(x.size(0), (- 1)) x = self.fc(x) return x
def eebls_gpu_custom(t, y, dy, freqs, q_values, phi_values, ignore_negative_delta_sols=False, freq_batch_size=None, nstreams=5, max_memory=None, functions=None, **kwargs): functions = (functions if (functions is not None) else compile_bls(**kwargs)) block_size = kwargs.get('block_size', _default_block_size) ndata = len(t) if (max_memory is None): (free, total) = cuda.mem_get_info() max_memory = int((0.9 * free)) if (freq_batch_size is None): real_type_size = 4 mem0 = ((ndata * 3) * real_type_size) nq = len(q_values) nphi = len(phi_values) mem0 += (nq + nphi) mem0 += ((len(freqs) * 5) * real_type_size) mem_per_f = ((((4 * nstreams) * nq) * nphi) * real_type_size) freq_batch_size = int((float((max_memory - mem0)) / mem_per_f)) if (freq_batch_size == 0): raise Exception('Not enough memory (freq_batch_size = 0)') nbtot = ((len(q_values) * len(phi_values)) * freq_batch_size) grid_size = int(np.ceil((float(nbtot) / block_size))) w = np.power(dy, (- 2)) w /= sum(w) ybar = np.dot(w, y) YY = np.dot(w, np.power((np.array(y) - ybar), 2)) yw = ((np.array(y) - ybar) * np.array(w)) t_g = gpuarray.to_gpu(np.array(t).astype(np.float32)) yw_g = gpuarray.to_gpu(yw.astype(np.float32)) w_g = gpuarray.to_gpu(np.array(w).astype(np.float32)) freqs_g = gpuarray.to_gpu(np.array(freqs).astype(np.float32)) (yw_g_bins, w_g_bins, bls_tmp_gs, bls_tmp_sol_gs, streams) = ([], [], [], [], []) for i in range(nstreams): streams.append(cuda.Stream()) yw_g_bins.append(gpuarray.zeros(nbtot, dtype=np.float32)) w_g_bins.append(gpuarray.zeros(nbtot, dtype=np.float32)) bls_tmp_gs.append(gpuarray.zeros(nbtot, dtype=np.float32)) bls_tmp_sol_gs.append(gpuarray.zeros(nbtot, dtype=np.uint32)) bls_g = gpuarray.zeros(len(freqs), dtype=np.float32) bls_sol_g = gpuarray.zeros(len(freqs), dtype=np.uint32) bls_best_phi = gpuarray.zeros(len(freqs), dtype=np.float32) bls_best_q = gpuarray.zeros(len(freqs), dtype=np.float32) q_values_g = gpuarray.to_gpu(np.asarray(q_values).astype(np.float32)) phi_values_g = gpuarray.to_gpu(np.asarray(phi_values).astype(np.float32)) block = (block_size, 1, 1) grid = (grid_size, 1) nbatches = int(np.ceil((float(len(freqs)) / freq_batch_size))) bls = np.zeros(len(freqs)) bin_func = functions['bin_and_phase_fold_custom'] bls_func = functions['binned_bls_bst'] max_func = functions['reduction_max'] store_func = functions['store_best_sols_custom'] for batch in range(nbatches): imin = (freq_batch_size * batch) imax = min([len(freqs), (freq_batch_size * (batch + 1))]) nf = (imax - imin) j = (batch % nstreams) yw_g_bin = yw_g_bins[j] w_g_bin = w_g_bins[j] bls_tmp_g = bls_tmp_gs[j] bls_tmp_sol_g = bls_tmp_sol_gs[j] stream = streams[j] yw_g_bin.fill(np.float32(0), stream=stream) w_g_bin.fill(np.float32(0), stream=stream) bls_tmp_g.fill(np.float32(0), stream=stream) bls_tmp_sol_g.fill(np.int32(0), stream=stream) bin_grid = (int(np.ceil((float((len(t) * nf)) / block_size))), 1) args = (bin_grid, block, stream) args += (t_g.ptr, yw_g.ptr, w_g.ptr) args += (yw_g_bin.ptr, w_g_bin.ptr, freqs_g.ptr) args += (q_values_g.ptr, phi_values_g.ptr) args += (np.uint32(len(q_values)), np.uint32(len(phi_values))) args += (np.uint32(len(t)), np.uint32(nf)) args += (np.uint32((freq_batch_size * batch)),) bin_func.prepared_async_call(*args) nb = (len(q_values) * len(phi_values)) bls_grid = (int(np.ceil((float((nf * nb)) / block_size))), 1) args = (bls_grid, block, stream) args += (yw_g_bin.ptr, w_g_bin.ptr) args += (bls_tmp_g.ptr, np.uint32((nf * nb))) args += (np.uint32(ignore_negative_delta_sols),) bls_func.prepared_async_call(*args) args = (max_func, bls_tmp_g, bls_tmp_sol_g) args += (nf, nb, stream, bls_g, bls_sol_g) args += ((batch * freq_batch_size), block_size) _reduction_max(*args) store_grid = (int(np.ceil((float(nf) / block_size))), 1) args = (store_grid, block, stream) args += (bls_sol_g.ptr, bls_best_phi.ptr, bls_best_q.ptr) args += (q_values_g.ptr, phi_values_g.ptr) args += (np.uint32(len(q_values)), np.uint32(len(phi_values))) args += (np.uint32(nf), np.uint32((batch * freq_batch_size))) store_func.prepared_async_call(*args) best_q = bls_best_q.get() best_phi = bls_best_phi.get() qphi_sols = list(zip(best_q, best_phi)) return ((bls_g.get() / YY), qphi_sols)
class __FakeLocalTFRunner(): def __init__(*args, **kwargs): raise ImportError('LocalTFRunner requires TensorFlow. To use it, please install TensorFlow.')
_materialize('core') class GELU(ElementWiseUnaryOp): in_dtypes = [(i,) for i in DTYPE_GEN_FLOATS] out_dtypes = [(i,) for i in DTYPE_GEN_FLOATS]
def averagees_average_info(model): stop_epoch_sum = 0.0 stop_acc_sum = 0.0 for suffix in [0, 1, 2, 3, 4, 75, 76, 77, 78, 79]: (stop_epoch, stop_acc) = analysis.get_averagees_stopping_point(model=model, file_suffix=suffix) stop_epoch_sum += stop_epoch stop_acc_sum += stop_acc stop_epoch_avg = float((stop_epoch_sum / 10.0)) stop_acc_sum = float((stop_acc_sum / 10.0)) return (stop_epoch_avg, stop_acc_sum)
def prototype_ubuntu_GaussPiecewise_NormOp_VHRED_Baseline_Exp1(): state = prototype_state() state['end_sym_utterance'] = '__eot__' state['unk_sym'] = 0 state['eos_sym'] = 1 state['eod_sym'] = (- 1) state['first_speaker_sym'] = (- 1) state['second_speaker_sym'] = (- 1) state['third_speaker_sym'] = (- 1) state['minor_speaker_sym'] = (- 1) state['voice_over_sym'] = (- 1) state['off_screen_sym'] = (- 1) state['pause_sym'] = (- 1) state['train_dialogues'] = '../UbuntuData/Training.dialogues.pkl' state['test_dialogues'] = '../UbuntuData/Test.dialogues.pkl' state['valid_dialogues'] = '../UbuntuData/Validation.dialogues.pkl' state['dictionary'] = '../UbuntuData/Dataset.dict.pkl' state['save_dir'] = 'Output' state['max_grad_steps'] = 80 state['valid_freq'] = 5000 state['prefix'] = 'UbuntuModel_' state['updater'] = 'adam' state['bidirectional_utterance_encoder'] = True state['deep_dialogue_encoder_input'] = False state['deep_utterance_decoder_out'] = True state['bs'] = 80 state['utterance_decoder_gating'] = 'LSTM' state['direct_connection_between_encoders_and_decoder'] = True state['qdim_encoder'] = 1000 state['qdim_decoder'] = 2000 state['sdim'] = 1000 state['rankdim'] = 400 state['add_latent_gaussian_per_utterance'] = False state['latent_gaussian_per_utterance_dim'] = 100 state['scale_latent_gaussian_variable_variances'] = 0.1 state['add_latent_piecewise_per_utterance'] = False state['latent_piecewise_per_utterance_dim'] = 100 state['latent_piecewise_alpha_variables'] = 3 state['scale_latent_piecewise_variable_alpha_use_softplus'] = False state['scale_latent_piecewise_variable_prior_alpha'] = 1.0 state['scale_latent_piecewise_variable_posterior_alpha'] = 1.0 state['condition_latent_variable_on_dialogue_encoder'] = True state['train_latent_variables_with_kl_divergence_annealing'] = True state['kl_divergence_annealing_rate'] = (1.0 / 75000.0) state['decoder_drop_previous_input_tokens'] = True state['decoder_drop_previous_input_tokens_rate'] = 0.75 state['deep_utterance_decoder_input'] = False state['patience'] = 20 return state
def ReadFileGS(x_axis, tthread, batchInterval, NUM_ITEMS, NUM_ACCESS, key_skewness, overlap_ratio, abort_ratio, txn_length, isCyclic, complexity): (w, h) = (2, len(x_axis)) y = [[] for _ in range(w)] for batchInterval in x_axis: inputEvents = (tthread * batchInterval) op_gs_path = getPathGS('OP_NS_A', inputEvents, tthread, NUM_ITEMS, NUM_ACCESS, key_skewness, overlap_ratio, abort_ratio, txn_length, isCyclic, complexity) lines = open(op_gs_path).readlines() throughput = lines[0].split(': ')[1] y[0].append(float(throughput)) for batchInterval in x_axis: inputEvents = (tthread * batchInterval) op_gs_path = getPathGS('OG_NS_A', inputEvents, tthread, NUM_ITEMS, NUM_ACCESS, key_skewness, overlap_ratio, abort_ratio, txn_length, isCyclic, complexity) lines = open(op_gs_path).readlines() throughput = lines[0].split(': ')[1] y[1].append(float(throughput)) print(y) return y
class SequentialPostProcessor(object): operations: Sequence[Callable] def __post_init__(self): special_tokens = [] for operation in self.operations: if hasattr(operation, 'special_tokens'): special_tokens.extend(operation.special_tokens) self.special_tokens = special_tokens def __call__(self, df: Union[(pd.DataFrame, dict)]) -> Union[(pd.DataFrame, dict)]: for operation in self.operations: df = operation(df) return df
class Experiment(): def __init__(self, store_or_uri: Union[(str, Store)], name: str=None, description: str=None, _id: int=None): if (name is None): name = f'#{random.randint(0, 9999)}' if (description is None): description = f'Experiment: {name}' if isinstance(store_or_uri, str): self._store = Store(store_or_uri) else: self._store = store_or_uri self._name = name self._description = description self._id = _id self._trials = [] def store(self) -> Store: return self._store def id(self) -> int: return self._id def name(self) -> str: return self._name def description(self) -> str: return self._description def trials(self) -> List: return self._trials def run(self, trial_run_fn: FunctionType, trial_gen_fn: FunctionType, timeout: int=None, n_replicates: int=1, executor: Executor=None) -> Executor: if (self._id is None): (self._id, _) = self._store.get_or_create_experiment(self._name, self._description) trial_params = [] for task in self._store.iter_tasks(): generated_trials = trial_gen_fn(task) for generated_trial in generated_trials: if isinstance(generated_trial, tuple): method = generated_trial[0] meta = generated_trial[1] else: method = generated_trial meta = {} if isinstance(method, str): method = Method.from_name(method) elif isinstance(method, Method): pass else: raise TypeError(f'Cannot handle method type: {type(method)}') (method_id, _) = self._store.get_or_create_method(method.name, method.description, method.version, method.params, method.env) method = Method(method_id, method.name, method.description, method.version, method.params, method.env) for replicate_num in range(n_replicates): trial_param = {'experiment_id': self.id, 'task_id': task.id, 'method_id': method.id, 'replicate_num': replicate_num, 'meta': meta} trial_params.append(trial_param) self._trials.append(Trial(None, self, task, method, replicate_num, meta, [])) trial_ids = self._store.create_trials(trial_params) for (_id, trial) in zip(trial_ids, self._trials): trial._id = _id if (executor is None): executor = LocalMachine(self._store) executor.submit(trial_run_fn, self._trials, timeout=timeout) return executor
def test_call_deps(): run_cell('def f(): return 0, 1, 2, 3') run_cell('a, b, c, d = f()') for sym in ('a', 'b', 'c', 'd'): run_cell(f'assert deps({sym}) == [lift(f)]') run_cell(f'assert users({sym}) == []') run_cell('g = lambda: (0, 1, 2, 3)') run_cell('w, x, y, z = g()') for sym in ('w', 'x', 'y', 'z'): run_cell(f'assert deps({sym}) == [lift(g)]') run_cell(f'assert users({sym}) == []')
class ResnetTester(nn.Module): def __init__(self, model): super(ResnetTester, self).__init__() self.layers = model self.classifier = nn.Linear(((2048 * 7) * 7), 200) def forward(self, input): features_in = self.layers(input) features_in = features_in.view(features_in.shape[0], (- 1)) return self.classifier(features_in)
def model2state_dict(file_path): model = torch.load(file_path) if (model['model'] is not None): model_state_dict = model['model'].state_dict() torch.save(model_state_dict, file_path.replace('.pth', 'state_dict.pth')) else: print(type(model)) print(model) print('skip')
def generate_df_pop_by_ags(df): log.info('aggregate (sum) population by AGS') df_pop = df.drop(columns=list((set(df.columns) - set(['ags', 'population_total'])))) df_pop = df_pop.rename(columns={'population_total': 'population'}) df_pop_by_ags = df_pop.groupby('ags').sum() df_pop_by_ags.index = df_pop_by_ags.index.astype(int) df_pop_by_ags.sort_index(inplace=True) dropthis = df_pop_by_ags[(df_pop_by_ags.population == 0)] log.info('delete these rows: %s', dropthis) df_pop_by_ags = df_pop_by_ags[(df_pop_by_ags.population != 0)] log.info('total population checksum: %s', df_pop_by_ags['population'].sum()) return df_pop_by_ags
class ApplySameTransformInputKeyOnList(ApplySameTransformToKeyOnList): def __init__(self, transform: Module, dim: int=1): super().__init__('input', transform=transform, dim=dim) def __repr__(self): return f'{self.__class__.__name__}(transform={self._transform}, dim={self._dim})'
def get_version(): major_value = ctypes.c_int(0) major = ctypes.pointer(major_value) minor_value = ctypes.c_int(0) minor = ctypes.pointer(minor_value) rev_value = ctypes.c_int(0) rev = ctypes.pointer(rev_value) _glfw.glfwGetVersion(major, minor, rev) return (major_value.value, minor_value.value, rev_value.value)
class Generator(object): def __init__(self): self.z_dim = 100 self.x_dim = 784 self.name = 'mnist/mlp/g_net' def __call__(self, z): with tf.variable_scope(self.name) as vs: fc = z fc = tcl.fully_connected(fc, 512, weights_initializer=tf.random_normal_initializer(stddev=0.02), weights_regularizer=tc.layers.l2_regularizer(2.5e-05), activation_fn=tcl.batch_norm) fc = leaky_relu(fc) fc = tcl.fully_connected(fc, 512, weights_initializer=tf.random_normal_initializer(stddev=0.02), weights_regularizer=tc.layers.l2_regularizer(2.5e-05), activation_fn=tcl.batch_norm) fc = leaky_relu(fc) fc = tcl.fully_connected(fc, 512, weights_initializer=tf.random_normal_initializer(stddev=0.02), weights_regularizer=tc.layers.l2_regularizer(2.5e-05), activation_fn=tcl.batch_norm) fc = leaky_relu(fc) fc = tc.layers.fully_connected(fc, 784, weights_initializer=tf.random_normal_initializer(stddev=0.02), weights_regularizer=tc.layers.l2_regularizer(2.5e-05), activation_fn=tf.sigmoid) return fc def vars(self): return [var for var in tf.all_variables() if (self.name in var.name)]
def sentence_tokenizer(sentences): tokenized_sents = nltk.word_tokenize(sentences) return tokenized_sents
.parametrize('dataset_class,model_class,create_submission_f,apply_model', [(T4c22Dataset, DummyArangeNN_eta, create_submission_eta_plain_torch, apply_model_plain), (T4c22GeometricDataset, DummyArangeNN_eta, create_submission_eta_torch_geometric, apply_model_geometric)]) def test_create_submission_eta_city_plain_torch(caplog, dataset_class, model_class, create_submission_f, apply_model): cities = ['london', 'melbourne', 'madrid'] date = '1970-01-01' num_test_slots = 37 expected_score_files = ['gogogo.score'] loss_before_pandas = {} with tempfile.TemporaryDirectory() as basedir: basedir = Path(basedir) submission_names = ['gogogo'] config = _create_dummy_competition_setup_with_arange_submissions(basedir, cities, date, num_test_slots, submission_names, seed=666, dataset_class=dataset_class, model_class=model_class, create_submission_f=create_submission_f, apply_model=apply_model) torch.manual_seed(666) for (city, (test_dataset, _)) in config.items(): (_, y_hat) = _inference_torch(ds=test_dataset, apply_model=apply_model, model=DummyArangeNN_eta(num_supersegments=len(test_dataset.torch_road_graph_mapping.supersegments))) df_y = load_eta_labels(basedir, city=city, split='test', df_filter=None) y = torch.cat([test_dataset.torch_road_graph_mapping._df_eta_to_torch(df_y[(df_y['test_idx'] == test_idx)]) for test_idx in range(num_test_slots)]) loss_f = torch.nn.L1Loss() loss_before_pandas[city] = loss_f(y_hat, y) golden_zip = (((basedir / 'withheld') / 'golden') / 'eta_golden.zip') prediction_zip = ((basedir / 'submission') / 'gogogo.zip') caplog.set_level(logging.INFO, logger='participants-prediction') caplog.set_level(logging.INFO, logger='full-prediction') for city in cities: EXPECTED_NUM_ITEMS[T4c22Competitions.EXTENDED.value][city] = (num_test_slots * NUM_SUPERSEGMENTS) main(['-g', str(golden_zip), '-i', str(prediction_zip), '-c', 'eta']) print(list(basedir.rglob('**/*'))) log_file = (prediction_zip.parent / 'gogogo-full.log') with open(log_file, 'r') as f: content = f.read() logging.info(content) print(content) log_file = (prediction_zip.parent / 'gogogo.log') with open(log_file, 'r') as f: content = f.read() logging.info(content) assert ('contact us for details' not in content), content with (prediction_zip.parent / 'gogogo.score.json').open() as f: content = json.load(f) print(content) for city in cities: assert np.isclose(loss_before_pandas[city], content[city]['all']['loss']) for file_name in expected_score_files: score_file = (prediction_zip.parent / file_name) assert os.path.exists(score_file), str(score_file) with open(score_file, 'r') as f: content = f.read() logging.info(content) assert np.isclose(float(content), np.mean([loss_before_pandas[city] for city in cities])), content
class MetricGraphPrinter(AbstractBaseLogger): def __init__(self, writer, key='train_loss', graph_name='Train Loss', group_name='metric'): self.key = key self.graph_label = graph_name self.group_name = group_name self.writer = writer def log(self, *args, **kwargs): if (self.key in kwargs): self.writer.add_scalar(((self.group_name + '/') + self.graph_label), kwargs[self.key], kwargs['accum_iter']) else: self.writer.add_scalar(((self.group_name + '/') + self.graph_label), 0, kwargs['accum_iter']) def complete(self, *args, **kwargs): self.writer.close()
def iou_boxes_polygons(boxes, polygons, w=0, h=0, xywh=True, ioubp=False): if ((w * h) == 0): p_boxes = [polygon_to_box(p) for p in polygons] region = boxes_region((p_boxes + list(boxes))) w = int((region[2] + 1)) h = int((region[3] + 1)) p_mask = polygons_to_mask(polygons, w, h) b_mask = boxes_to_mask(boxes, w, h, xywh) i = np.sum((np.logical_and(p_mask, b_mask) > 0), axis=None) u = np.sum(((p_mask + b_mask) > 0), axis=None) if (not ioubp): if (i == 0): return 0 else: return ((i * 1.0) / u) elif (i == 0): return (0, 0, 0) else: b = np.sum((b_mask > 0), axis=None) p = np.sum((p_mask > 0), axis=None) return (((i * 1.0) / u), ((i * 1.0) / b), ((i * 1.0) / p))
def _find_rocsolver_config(rocm_install_path): def rocsolver_version_numbers(path): possible_version_files = ['include/rocsolver/rocsolver-version.h', 'rocsolver/include/rocsolver-version.h'] version_file = None for f in possible_version_files: version_file_path = os.path.join(path, f) if os.path.exists(version_file_path): version_file = version_file_path break if (not version_file): raise ConfigError('rocsolver version file not found in {}'.format(possible_version_files)) major = _get_header_version(version_file, 'ROCSOLVER_VERSION_MAJOR') minor = _get_header_version(version_file, 'ROCSOLVER_VERSION_MINOR') patch = _get_header_version(version_file, 'ROCSOLVER_VERSION_PATCH') return (major, minor, patch) (major, minor, patch) = rocsolver_version_numbers(rocm_install_path) rocsolver_config = {'rocsolver_version_number': _get_composite_version_number(major, minor, patch)} return rocsolver_config
class P1203Pv(object): _COEFFS = {'u1': 72.61, 'u2': 0.32, 't1': 30.98, 't2': 1.29, 't3': 64.65, 'q1': 4.66, 'q2': (- 0.07), 'q3': 4.06, 'mode0': {'a1': 11.9983519, 'a2': (- 2.), 'a3': 41., 'a4': 0.}, 'mode1': {'a1': 5., 'a2': (- 1.), 'a3': 41.3585049, 'a4': 0, 'c0': (- 0.), 'c1': 0, 'c2': (- 3.), 'c3': 20.4098663}, 'htv_1': (- 0.60293), 'htv_2': 2.12382, 'htv_3': (- 0.36936), 'htv_4': 0.03409} def degradation_due_to_upscaling(self, coding_res, display_res): scale_factor = (display_res / coding_res) scale_factor = max(scale_factor, 1) u1 = self.coeffs['u1'] u2 = self.coeffs['u2'] deg_scal_v = (u1 * np.log10(((u2 * (scale_factor - 1.0)) + 1.0))) deg_scal_v = utils.constrain(deg_scal_v, 0.0, 100.0) return deg_scal_v def degradation_due_to_frame_rate_reduction(self, deg_cod_v, deg_scal_v, framerate): t1 = self.coeffs['t1'] t2 = self.coeffs['t2'] t3 = self.coeffs['t3'] deg_frame_rate_v = 0 if (framerate < 24): deg_frame_rate_v = ((((100 - deg_cod_v) - deg_scal_v) * (t1 - (t2 * framerate))) / (t3 + framerate)) deg_frame_rate_v = utils.constrain(deg_frame_rate_v, 0.0, 100.0) return deg_frame_rate_v def degradation_integration(self, mos_cod_v, deg_cod_v, deg_scal_v, deg_frame_rate_v): deg_all = utils.constrain(((deg_cod_v + deg_scal_v) + deg_frame_rate_v), 0.0, 100.0) qv = (100 - deg_all) return utils.mos_from_r(qv) _cache() def video_model_function_mode0(self, coding_res, display_res, bitrate_kbps_segment_size, framerate): a1 = self.coeffs['mode0']['a1'] a2 = self.coeffs['mode0']['a2'] a3 = self.coeffs['mode0']['a3'] a4 = self.coeffs['mode0']['a4'] q1 = self.coeffs['q1'] q2 = self.coeffs['q2'] q3 = self.coeffs['q3'] quant = (a1 + (a2 * np.log(((a3 + np.log(bitrate_kbps_segment_size)) + np.log((((bitrate_kbps_segment_size * bitrate_kbps_segment_size) / (coding_res * framerate)) + a4)))))) mos_cod_v = (q1 + (q2 * np.exp((q3 * quant)))) mos_cod_v = utils.constrain(mos_cod_v, 1.0, 5.0) deg_cod_v = (100.0 - utils.r_from_mos(mos_cod_v)) deg_cod_v = utils.constrain(deg_cod_v, 0.0, 100.0) deg_scal_v = self.degradation_due_to_upscaling(coding_res, display_res) deg_frame_rate_v = self.degradation_due_to_frame_rate_reduction(deg_cod_v, deg_scal_v, framerate) score = self.degradation_integration(mos_cod_v, deg_cod_v, deg_scal_v, deg_frame_rate_v) logger.debug(json.dumps({'coding_res': round(coding_res, 2), 'display_res': round(display_res, 2), 'bitrate_kbps_segment_size': round(bitrate_kbps_segment_size, 2), 'framerate': round(framerate, 2), 'mos_cod_v': round(mos_cod_v, 2), 'deg_cod_v': round(deg_cod_v, 2), 'deg_scal_v': round(deg_scal_v, 2), 'deg_frame_rate_v': round(deg_frame_rate_v, 2), 'score': round(score, 2)}, indent=True)) return score def video_model_function_mode1(self, coding_res, display_res, bitrate_kbps_segment_size, framerate, frames, iframe_ratio=None): a1 = self.coeffs['mode1']['a1'] a2 = self.coeffs['mode1']['a2'] a3 = self.coeffs['mode1']['a3'] a4 = self.coeffs['mode1']['a4'] q1 = self.coeffs['q1'] q2 = self.coeffs['q2'] q3 = self.coeffs['q3'] quant = (a1 + (a2 * np.log(((a3 + np.log(bitrate_kbps_segment_size)) + np.log((((bitrate_kbps_segment_size * bitrate_kbps_segment_size) / (coding_res * framerate)) + a4)))))) mos_cod_v = (q1 + (q2 * np.exp((q3 * quant)))) mos_cod_v = utils.constrain(mos_cod_v, 1.0, 5.0) c0 = self.coeffs['mode1']['c0'] c1 = self.coeffs['mode1']['c1'] c2 = self.coeffs['mode1']['c2'] c3 = self.coeffs['mode1']['c3'] if (not iframe_ratio): i_sizes = [] noni_sizes = [] for frame in frames: frame_size = utils.calculate_compensated_size(frame['type'], frame['size'], frame['dts']) if (frame['type'] == 'I'): i_sizes.append(int(frame_size)) else: noni_sizes.append(int(frame_size)) if (i_sizes and noni_sizes): iframe_ratio = (np.mean(i_sizes) / np.mean(noni_sizes)) else: iframe_ratio = 0 complexity = utils.sigmoid(c0, c1, c2, c3, iframe_ratio) mos_cod_v += complexity deg_cod_v = (100.0 - utils.r_from_mos(mos_cod_v)) deg_cod_v = utils.constrain(deg_cod_v, 0.0, 100.0) deg_scal_v = self.degradation_due_to_upscaling(coding_res, display_res) deg_frame_rate_v = self.degradation_due_to_frame_rate_reduction(deg_cod_v, deg_scal_v, framerate) score = self.degradation_integration(mos_cod_v, deg_cod_v, deg_scal_v, deg_frame_rate_v) logger.debug(json.dumps({'coding_res': round(coding_res, 2), 'display_res': round(display_res, 2), 'bitrate_kbps_segment_size': round(bitrate_kbps_segment_size, 2), 'framerate': round(framerate, 2), 'mos_cod_v': round(mos_cod_v, 2), 'deg_cod_v': round(deg_cod_v, 2), 'iframe_ratio': round(iframe_ratio, 2), 'complexity': round(complexity, 2), 'deg_scal_v': round(deg_scal_v, 2), 'deg_frame_rate_v': round(deg_frame_rate_v, 2), 'score': round(score, 2)}, indent=True)) return score def video_model_function_mode2(self, coding_res, display_res, framerate, frames, quant=None, avg_qp_per_noni_frame=[]): if (not quant): if (not avg_qp_per_noni_frame): types = [] qp_values = [] for frame in frames: qp_values.append(frame['qpValues']) frame_type = frame['type'] if (frame_type not in ['I', 'P', 'B', 'Non-I']): raise P1203StandaloneError((('frame type ' + str(frame_type)) + ' not valid; must be I/P/B or I/Non-I')) types.append(frame_type) qppb = [] for (index, frame_type) in enumerate(types): if (frame_type in ['P', 'B', 'Non-I']): qppb.extend(qp_values[index]) avg_qp = np.mean(qppb) else: avg_qp = np.mean(avg_qp_per_noni_frame) quant = (avg_qp / 51.0) q1 = self.coeffs['q1'] q2 = self.coeffs['q2'] q3 = self.coeffs['q3'] mos_cod_v = (q1 + (q2 * math.exp((q3 * quant)))) mos_cod_v = max(min(mos_cod_v, 5), 1) deg_cod_v = (100 - utils.r_from_mos(mos_cod_v)) deg_cod_v = max(min(deg_cod_v, 100), 0) deg_scal_v = self.degradation_due_to_upscaling(coding_res, display_res) deg_frame_rate_v = self.degradation_due_to_frame_rate_reduction(deg_cod_v, deg_scal_v, framerate) score = self.degradation_integration(mos_cod_v, deg_cod_v, deg_scal_v, deg_frame_rate_v) logger.debug(json.dumps({'coding_res': round(coding_res, 2), 'display_res': round(display_res, 2), 'framerate': round(framerate, 2), 'quant': round(quant, 2), 'mos_cod_v': round(mos_cod_v, 2), 'deg_cod_v': round(deg_cod_v, 2), 'deg_scal_v': round(deg_scal_v, 2), 'deg_frame_rate_v': round(deg_frame_rate_v, 2), 'score': round(score, 2)}, indent=True)) return score def video_model_function_mode3(self, coding_res, display_res, framerate, frames, quant=None, avg_qp_per_noni_frame=[]): if (not quant): if (not avg_qp_per_noni_frame): types = [] qp_values = [] for frame in frames: qp_values.append(frame['qpValues']) frame_type = frame['type'] if (frame_type not in ['I', 'P', 'B', 'Non-I']): raise P1203StandaloneError((('frame type ' + str(frame_type)) + ' not valid; must be I/P/B or I/Non-I')) types.append(frame_type) qppb = [] for (index, frame_type) in enumerate(types): if (frame_type in ['P', 'B', 'Non-I']): qppb.extend(qp_values[index]) elif ((frame_type == 'I') and (len(qppb) > 0)): if (len(qppb) > 1): qppb[(- 1)] = qppb[(- 2)] else: qppb = [] avg_qp = np.mean(qppb) else: avg_qp = np.mean(avg_qp_per_noni_frame) quant = (avg_qp / 51.0) q1 = self.coeffs['q1'] q2 = self.coeffs['q2'] q3 = self.coeffs['q3'] mos_cod_v = (q1 + (q2 * math.exp((q3 * quant)))) mos_cod_v = max(min(mos_cod_v, 5), 1) deg_cod_v = (100 - utils.r_from_mos(mos_cod_v)) deg_cod_v = max(min(deg_cod_v, 100), 0) deg_scal_v = self.degradation_due_to_upscaling(coding_res, display_res) deg_frame_rate_v = self.degradation_due_to_frame_rate_reduction(deg_cod_v, deg_scal_v, framerate) score = self.degradation_integration(mos_cod_v, deg_cod_v, deg_scal_v, deg_frame_rate_v) logger.debug(json.dumps({'coding_res': round(coding_res, 2), 'display_res': round(display_res, 2), 'framerate': round(framerate, 2), 'quant': round(quant, 2), 'mos_cod_v': round(mos_cod_v, 2), 'deg_cod_v': round(deg_cod_v, 2), 'deg_scal_v': round(deg_scal_v, 2), 'deg_frame_rate_v': round(deg_frame_rate_v, 2), 'score': round(score, 2)}, indent=True)) return score def handheld_adjustment(self, score): htv_1 = self.coeffs['htv_1'] htv_2 = self.coeffs['htv_2'] htv_3 = self.coeffs['htv_3'] htv_4 = self.coeffs['htv_4'] return max(min((((htv_1 + (htv_2 * score)) + (htv_3 * (score ** 2))) + (htv_4 * (score ** 3))), 5), 1) def model_callback(self, output_sample_timestamp, frames): logger.debug(('Output score at timestamp ' + str(output_sample_timestamp))) output_sample_index = [i for (i, f) in enumerate(frames) if (f['dts'] < output_sample_timestamp)][(- 1)] if (self.mode == 0): if any((('representation' in f) for f in frames)): frames = utils.get_chunk(frames, output_sample_index, type='video') first_frame = frames[0] bitrate = np.mean([f['bitrate'] for f in frames]) display_res = (first_frame.get('displaySize') or self.display_res) score = self.video_model_function_mode0(utils.resolution_to_number(first_frame['resolution']), utils.resolution_to_number(display_res), bitrate, first_frame['fps']) else: score = self.video_model_function_mode0(utils.resolution_to_number(frames[output_sample_index]['resolution']), utils.resolution_to_number((frames[output_sample_index].get('displaySize') or self.display_res)), frames[output_sample_index]['bitrate'], frames[output_sample_index]['fps']) else: frames = utils.get_chunk(frames, output_sample_index, type='video') first_frame = frames[0] display_res = (first_frame.get('displaySize') or self.display_res) if (self.mode == 1): compensated_sizes = [utils.calculate_compensated_size(f['type'], f['size'], f['dts']) for f in frames] duration = np.sum([f['duration'] for f in frames]) bitrate = (((np.sum(compensated_sizes) * 8) / duration) / 1000) score = self.video_model_function_mode1(utils.resolution_to_number(first_frame['resolution']), utils.resolution_to_number(display_res), bitrate, first_frame['fps'], frames) elif (self.mode == 2): score = self.video_model_function_mode2(utils.resolution_to_number(first_frame['resolution']), utils.resolution_to_number(display_res), first_frame['fps'], frames) elif (self.mode == 3): score = self.video_model_function_mode3(utils.resolution_to_number(first_frame['resolution']), utils.resolution_to_number(display_res), first_frame['fps'], frames) else: raise P1203StandaloneError('Unsupported mode: {}'.format(self.mode)) if (self.device in ['mobile', 'handheld']): score = self.handheld_adjustment(score) self.o22.append(score) def check_codec(self): codecs = list(set([s['codec'] for s in self.segments])) for c in codecs: if (c != 'h264'): raise P1203StandaloneError('Unsupported codec: {}'.format(c)) def _calculate_with_measurementwindow(self): measurementwindow = MeasurementWindow() measurementwindow.set_score_callback(self.model_callback) self.mode = 0 for segment in self.segments: if ('frames' not in segment.keys()): self.mode = 0 break if ('frames' in segment): for frame in segment['frames']: if (('frameType' not in frame.keys()) or ('frameSize' not in frame.keys())): raise P1203StandaloneError("Frame definition must have at least 'frameType' and 'frameSize'") if ('qpValues' in frame.keys()): self.mode = 3 else: self.mode = 1 break logger.debug(('Evaluating stream in mode ' + str(self.mode))) self.check_codec() if (self.mode == 0): dts = 0 for segment in self.segments: segment_fps = min(segment['fps'], 120) if (segment_fps != segment['fps']): logger.warning('FPS of segment is higher than 120, capping to prevent incorrect results') num_frames = int((segment['duration'] * segment_fps)) frame_duration = (1.0 / segment_fps) for i in range(int(num_frames)): frame = {'duration': frame_duration, 'dts': dts, 'bitrate': segment['bitrate'], 'codec': segment['codec'], 'fps': segment_fps, 'resolution': segment['resolution']} if ('displaySize' in segment.keys()): frame['displaySize'] = segment['displaySize'] if ('representation' in segment.keys()): frame.update({'representation': segment['representation']}) measurementwindow.add_frame(frame) dts += frame_duration measurementwindow.stream_finished() else: dts = 0 for (segment_index, segment) in enumerate(self.segments): num_frames_assumed = int((segment['duration'] * segment['fps'])) num_frames = len(segment['frames']) if (num_frames != num_frames_assumed): logger.warning(((('Segment specifies ' + str(num_frames)) + ' frames but based on calculations, there should be ') + str(num_frames_assumed))) frame_duration = (1.0 / segment['fps']) for i in range(int(num_frames)): frame = {'duration': frame_duration, 'dts': dts, 'bitrate': segment['bitrate'], 'codec': segment['codec'], 'fps': segment['fps'], 'resolution': segment['resolution'], 'size': segment['frames'][i]['frameSize'], 'type': segment['frames'][i]['frameType']} if ('displaySize' in segment.keys()): frame['displaySize'] = segment['displaySize'] if ('representation' in segment.keys()): frame.update({'representation': segment['representation']}) if (self.mode == 3): qp_values = segment['frames'][i]['qpValues'] if (not qp_values): raise P1203StandaloneError('No QP values for frame {i} of segment {segment_index}'.format(**locals())) frame['qpValues'] = qp_values measurementwindow.add_frame(frame) dts += frame_duration measurementwindow.stream_finished() def _calculate_fast_mode(self): if ((self.mode is not None) and (self.mode != 0)): raise P1203StandaloneError(f'Fast mode only works with mode 0, but it is set to {self.mode}') self.mode = 0 for segment in self.segments: score = self.video_model_function_mode0(utils.resolution_to_number(segment['resolution']), utils.resolution_to_number(segment.get('displaySize', self.display_res)), segment['bitrate'], segment['fps']) self.o22.extend(([score] * math.floor(segment['duration']))) def calculate(self, fast_mode=False): utils.check_segment_continuity(self.segments, 'video') if fast_mode: logger.warning('Using fast mode of the model, results may not be accurate to the second') self._calculate_fast_mode() else: self._calculate_with_measurementwindow() return {'video': {'streamId': self.stream_id, 'mode': self.mode, 'O22': self.o22}} def __init__(self, segments, display_res='1920x1080', device='pc', stream_id=None, coeffs={}): self.segments = segments self.display_res = display_res self.device = device self.stream_id = stream_id self.o22 = [] self.mode = None self.coeffs = {**self._COEFFS, **coeffs}
def available_classes(cls: type[Registry]) -> list[str]: return list(cls.available_classes().keys())
def postprocess_args(args): ROOTDIR = args.root_dir if (args.dataset == 'touchdown'): ft_file_map = {'resnet18': 'resnet18_view_fts.hdf5', 'vit_clip': 'vit_clip_view_fts.hdf5'} args.img_ft_file = os.path.join(ROOTDIR, 'Touchdown', 'features', ft_file_map[args.features]) args.anno_dir = os.path.join(ROOTDIR, 'Touchdown', 'annotations', 'touchdown', 'data') args.graph_dir = os.path.join(ROOTDIR, 'Touchdown', 'annotations', 'touchdown', 'graph') else: ft_file_map = {'imagenet': 'pth_resnet152_imagenet.hdf5', 'imagenet_caffe': 'caffe_resnet152_imagenet.hdf5', 'places365': 'caffe_resnet152_places365.hdf5', 'vitbase': 'pth_vit_base_patch16_224_imagenet.hdf5', 'vitbase_r2rfte2e': 'pth_vit_base_patch16_224_imagenet_r2r.e2e.ft.22k.hdf5', 'vitbase_r2rfte2e.aug': 'pth_vit_base_patch16_224_imagenet_r2r.e2e.dataaug.20k.hdf5', 'vitbase_in21k': 'pth_vit_base_patch16_224_in21k_imagenet.hdf5', 'vitbase_clip': 'pth_vit_base_patch32_224_clip.hdf5', 'vitbase_r2r.e2e.singlestage': 'pth_vit_base_patch16_224_imagenet_r2r.e2e.pt.25k.hdf5', 'vitbase_r2r.e2e.noobs': 'pth_vit_base_patch16_224_imagenet_r2r.e2e.noobs.20k.hdf5', 'vitbase_r2r.e2e.nosprel': 'pth_vit_base_patch16_224_imagenet_r2r.e2e.nosprel.19k.hdf5'} args.img_ft_file = os.path.join(ROOTDIR, 'R2R', 'features', ft_file_map[args.features]) args.connectivity_dir = os.path.join(ROOTDIR, 'R2R', 'connectivity') args.scan_data_dir = os.path.join(ROOTDIR, 'Matterport3D', 'v1_unzip_scans') if (args.dataset == 'rxr'): args.anno_dir = os.path.join(ROOTDIR, 'RxR', 'annotations') else: args.anno_dir = os.path.join(ROOTDIR, 'R2R', 'annotations') args.ckpt_dir = os.path.join(args.output_dir, 'ckpts') args.log_dir = os.path.join(args.output_dir, 'logs') args.pred_dir = os.path.join(args.output_dir, 'preds') os.makedirs(args.output_dir, exist_ok=True) os.makedirs(args.ckpt_dir, exist_ok=True) os.makedirs(args.log_dir, exist_ok=True) os.makedirs(args.pred_dir, exist_ok=True) if (args.vlnbert == 'cmt'): del args.prefix_causal_attn elif (args.vlnbert == 'mmt'): del args.no_lang_ca del args.act_pred_token elif (args.vlnbert == 'causal.cmt'): del args.prefix_causal_attn del args.no_lang_ca del args.act_pred_token return args
def get_current_user_path(path_in): if (path_in == ''): return '' from os.path import expanduser path = path_in.split('/') new_path = ((expanduser('~') + '/') + '/'.join(path[3:])) return str(new_path)
def nano(num_processes): def decorator(func): return _Nano_Customized_Training(func, num_processes) return decorator
class NezhaForQuestionAnswering(metaclass=DummyObject): _backends = ['torch'] def __init__(self, *args, **kwargs): requires_backends(self, ['torch'])
def dbscan_with_masked_image(image, eps=0.5, min_samples=3): if (len(image.shape) != 2): raise ValueError('Input image must be grayscale!') masked_points = np.where((image > 0)) if (len(masked_points) > 0): X = np.column_stack(tuple((masked_points[1], masked_points[0]))) if (X.shape[0] > 0): class_dict = dbscan_with_points(X, eps, min_samples) return class_dict else: return dict() else: return dict()
def main(opt): model = torch.load(opt['model.model_path']) model.eval() model_opt_file = os.path.join(os.path.dirname(opt['model.model_path']), 'opt.json') with open(model_opt_file, 'r') as f: model_opt = json.load(f) model_opt['model.x_dim'] = map(int, model_opt['model.x_dim'].split(',')) model_opt['log.fields'] = model_opt['log.fields'].split(',') data_opt = {('data.' + k): v for (k, v) in filter_opt(model_opt, 'data').items()} episode_fields = {'data.test_way': 'data.way', 'data.test_shot': 'data.shot', 'data.test_query': 'data.query', 'data.test_episodes': 'data.train_episodes'} for (k, v) in episode_fields.items(): if (opt[k] != 0): data_opt[k] = opt[k] elif (model_opt[k] != 0): data_opt[k] = model_opt[k] else: data_opt[k] = model_opt[v] print('Evaluating {:d}-way, {:d}-shot with {:d} query examples/class over {:d} episodes'.format(data_opt['data.test_way'], data_opt['data.test_shot'], data_opt['data.test_query'], data_opt['data.test_episodes'])) torch.manual_seed(1234) if data_opt['data.cuda']: torch.cuda.manual_seed(1234) data = data_utils.load(data_opt, ['test']) if data_opt['data.cuda']: model.cuda() meters = {field: tnt.meter.AverageValueMeter() for field in model_opt['log.fields']} model_utils.evaluate(model, data['test'], meters, desc='test') for (field, meter) in meters.items(): (mean, std) = meter.value() print('test {:s}: {:0.6f} +/- {:0.6f}'.format(field, mean, ((1.96 * std) / math.sqrt(data_opt['data.test_episodes']))))
def test_run_with_ignore_embedded_text(): example = EXAMPLES[2] document = load_document(example.path, use_embedded_text=False) pipe = pipeline('document-question-answering', model=CHECKPOINTS['LayoutLMv1']) for qa in example.qa_pairs: resp = pipe(question=qa.question, **document.context, top_k=1) assert (nested_simplify(resp, decimals=4) == qa.answers['LayoutLMv1__use_embedded_text=False'])
def test_env_instantiation(): env = ArgumentEnv('arg') assert (env.arg == 'arg') assert (env.calls == 1)
((PT_VERSION.release < Version('2.1.0').release), 'Please use PyTroch 2.1.0 or higher version for executor backend') class TestLLMQuantization(unittest.TestCase): def test_qwen(self): tokenizer = AutoTokenizer.from_pretrained('Qwen/Qwen-7B-Chat', trust_remote_code=True) sq_config = SmoothQuantConfig(calib_iters=3, calib_len=5, tokenizer=tokenizer, excluded_precisions=['bf16']) model = AutoModelForCausalLM.from_pretrained('Qwen/Qwen-7B-Chat', quantization_config=sq_config, trust_remote_code=True, use_llm_runtime=False) self.assertTrue(isinstance(model.model, torch.jit.ScriptModule)) def test_chatglm2(self): tokenizer = AutoTokenizer.from_pretrained('THUDM/chatglm2-6b', trust_remote_code=True) sq_config = SmoothQuantConfig(calib_iters=3, calib_len=5, tokenizer=tokenizer) model = AutoModelForCausalLM.from_pretrained('THUDM/chatglm2-6b', quantization_config=sq_config, trust_remote_code=True, use_llm_runtime=False) self.assertTrue(isinstance(model.model, torch.jit.ScriptModule)) def test_chatglm3(self): tokenizer = AutoTokenizer.from_pretrained('THUDM/chatglm3-6b', trust_remote_code=True) sq_config = SmoothQuantConfig(calib_iters=3, calib_len=5, tokenizer=tokenizer) model = AutoModelForCausalLM.from_pretrained('THUDM/chatglm3-6b', quantization_config=sq_config, trust_remote_code=True, use_llm_runtime=False) self.assertTrue(isinstance(model.model, torch.jit.ScriptModule))
def linear_rampup(current, rampup_length=16): if (rampup_length == 0): return 1.0 else: current = np.clip((current / rampup_length), 0.0, 1.0) return float(current)
def create_meter_display(group_dict, ignore_start_with='_'): def prune_dict(dictionary: dict, ignore='_'): for (k, v) in dictionary.copy().items(): if isinstance(v, dict): prune_dict(v, ignore) elif k.startswith(ignore): del dictionary[k] def prune_nan(dictionary: dict, father_dictionary: dict=None): for (k, v) in dictionary.copy().items(): if isinstance(v, dict): prune_nan(v, dictionary) elif math.isnan(v): del dictionary[k] if (father_dictionary is not None): if (len(father_dictionary) == 1): del father_dictionary prune_dict(group_dict, ignore_start_with) display = str(item2str(group_dict)) return display
class TestTensorParallelOptimization(unittest.TestCase): def tearDown(self): destroy_parallel_group() return super().tearDown() def test_tensor_parallel_optimization(self): _run_tensor_parallel_optimization()
class Trainer(): def __init__(self) -> None: self.task = '' self.note = '' self.ckpt = '' self.dataset = 'ImageNet-LT' self.nb_classes = 1000 self.epochs = 800 self.batch = 256 self.accum_iter = 4 self.device = '0,1,2,3' self.model = 'mae_vit_base_patch16' self.resume = '' self.input_size = 224 self.drop_path = 0.1 self.clip_grad = None self.weight_decay = 0.05 self.adamW2 = 0.95 self.lr = None self.blr = 0.00015 self.layer_decay = 0.75 self.min_lr = 1e-06 self.warmup_epochs = 40 self.color_jitter = None self.aa = 'rand-m9-mstd0.5-inc1' self.smoothing = 0.1 self.reprob = 0.25 self.remode = 'pixel' self.recount = 1 self.resplit = False self.mixup = 0.8 self.cutmix = 1.0 self.cutmix_minmax = None self.mixup_prob = 1.0 self.mixup_switch_prob = 0.5 self.mixup_mode = 'batch' self.loss = 'ce' self.bal_tau = 1.0 self.mask_ratio = 0.75 self.global_pool = True self.attn_only = False self.seed = 0 self.prit = 20 self.imbf = 100 self.num_workers = 16 self.master_port = 29500 def get_data_path(self): if (self.dataset == 'ImageNet-LT'): self.nb_classes = 1000 return IMAGENET_LT_PATH if (self.dataset == 'iNat18'): self.nb_classes = 8142 return INAT18_PATH if (self.dataset == 'ImageNet-BAL'): self.nb_classes = 1000 return IMAGENET_BAL_PATH if (self.dataset == 'cifar10-LT'): self.nb_classes = 10 return CIFAR10_PATH if (self.dataset == 'cifar100-LT'): self.nb_classes = 100 return CIFAR100_PATH if (self.dataset == 'Place'): self.nb_classes = 365 return PLCAE_PATH return None def pretrain(self): assert (not ((self.task == '') or (self.note == ''))), 'Need basic setting ...' os.environ['CUDA_VISIBLE_DEVICES'] = self.device task = self.task note = self.note nodes = len(self.device.split(',')) data_path = self.get_data_path() log_dir = os.path.join(WORK_PATH, f'exp/{task}/{self.dataset}/{note}') ckpt_dir = os.path.join(WORK_PATH, f'ckpt/{task}/{self.dataset}/{note}') exe_file = os.path.join(WORK_PATH, 'main_pretrain.py') os.system(f'''python -m torch.distributed.launch --nproc_per_node={nodes} --master_port {self.master_port} {exe_file} --ckpt_dir '{ckpt_dir}' --log_dir '{log_dir}' --batch_size {self.batch} --input_size {self.input_size} --world_size {nodes} --accum_iter {self.accum_iter} --model {self.model} --resume '{self.resume}' --norm_pix_loss --mask_ratio {self.mask_ratio} --epochs {self.epochs} --warmup_epochs {self.warmup_epochs} --blr {self.blr} --weight_decay {self.weight_decay} --data_path '{data_path}' --dataset '{self.dataset}' --num_workers {self.num_workers} ''') def finetune(self): assert (not ((self.task == '') or (self.note == ''))), 'Need basic setting ...' os.environ['CUDA_VISIBLE_DEVICES'] = self.device nodes = len(self.device.split(',')) data_path = self.get_data_path() log_dir = os.path.join(WORK_PATH, f'exp/{self.task}/{self.dataset}/{self.model}/{self.note}') ckpt_dir = os.path.join(WORK_PATH, f'ckpt/{self.task}/{self.dataset}/{self.model}/{self.note}') exe_file = os.path.join(WORK_PATH, 'main_finetune.py') attn_only = ('--attn_only' if self.attn_only else '') cls_type = ('--global_pool' if self.global_pool else '--cls_token') os.system(f'''python -m torch.distributed.launch --nproc_per_node={nodes} --master_port {self.master_port} {exe_file} --ckpt_dir '{ckpt_dir}' --log_dir '{log_dir}' --finetune '{self.ckpt}' --resume '{self.resume}' --batch_size {self.batch} --input_size {self.input_size} --world_size {nodes} --model {self.model} --loss {self.loss} --bal_tau {self.bal_tau} --accum_iter {self.accum_iter} --epochs {self.epochs} --warmup_epochs {self.warmup_epochs} --blr {self.blr} --layer_decay {self.layer_decay} --weight_decay {self.weight_decay} --adamW2 {self.adamW2} --drop_path {self.drop_path} --reprob {self.reprob} --mixup {self.mixup} --cutmix {self.cutmix} --data_path {data_path} --dataset {self.dataset} --imbf {self.imbf} --nb_classes {self.nb_classes} --num_workers {self.num_workers} --prit {self.prit} {attn_only} {cls_type} --dist_eval ''') def evaluate(self): if (self.device != 'cpu'): os.environ['CUDA_VISIBLE_DEVICES'] = self.device task = self.task note = self.note nodes = len(self.device.split(',')) data_path = self.get_data_path() log_dir = os.path.join(WORK_PATH, f'exp/{task}/{self.dataset}/{note}') exe_file = os.path.join(WORK_PATH, 'main_finetune.py') attn_only = ('--attn_only' if self.attn_only else '') cls_type = ('--global_pool' if self.global_pool else '--cls_token') os.system(f'''python -m torch.distributed.launch --nproc_per_node={nodes} --master_port {self.master_port} {exe_file} --log_dir '{log_dir}' --resume '{self.resume}' --finetune '{self.finetune}' --batch_size {self.batch} --input_size {self.input_size} --world_size {nodes} --model {self.model} --drop_path {self.drop_path} --data_path {data_path} --dataset {self.dataset} --nb_classes {self.nb_classes} --num_workers {self.num_workers} --prit {self.prit} {attn_only} {cls_type} --eval ''')
def create_effects_augmentation_chain(effects, ir_dir_path=None, sample_rate=44100, shuffle=False, parallel=False, parallel_weight_factor=None): fx_list = [] apply_prob = [] for cur_fx in effects: if isinstance(cur_fx, tuple): apply_prob.append(cur_fx[1]) cur_fx = cur_fx[0] else: apply_prob.append(1) if (isinstance(cur_fx, AugmentationChain) or isinstance(cur_fx, Processor)): fx_list.append(cur_fx) elif (cur_fx.lower() == 'gain'): fx_list.append(Gain()) elif ('eq' in cur_fx.lower()): fx_list.append(Equaliser(n_channels=2, sample_rate=sample_rate)) elif ('comp' in cur_fx.lower()): fx_list.append(Compressor(sample_rate=sample_rate)) elif ('expand' in cur_fx.lower()): fx_list.append(Expander(sample_rate=sample_rate)) elif ('pan' in cur_fx.lower()): fx_list.append(Panner()) elif ('image' in cur_fx.lower()): fx_list.append(MidSideImager()) elif ('algorithmic' in cur_fx.lower()): fx_list.append(AlgorithmicReverb(sample_rate=sample_rate)) elif ('reverb' in cur_fx.lower()): if (ir_dir_path == None): fx_list.append(AlgorithmicReverb(sample_rate=sample_rate)) else: IR_paths = glob(f'{ir_dir_path}*/RT60_avg/[!0-]*') IR_list = [] IR_dict = {} for IR_path in IR_paths: cur_rt = IR_path.split('/')[(- 1)] if (cur_rt not in IR_dict): IR_dict[cur_rt] = [] IR_dict[cur_rt].extend(create_dataset(path=IR_path, accepted_sampling_rates=[sample_rate], sources=['impulse_response'], mapped_sources={}, load_to_memory=True, debug=False)[0]) long_ir_list = [] for cur_rt in IR_dict: cur_rt_len = int(cur_rt.split('-')[0]) if (cur_rt_len < 3000): IR_list.append(IR_dict[cur_rt]) else: long_ir_list.extend(IR_dict[cur_rt]) IR_list.append(long_ir_list) fx_list.append(ConvolutionalReverb(IR_list, sample_rate)) else: raise ValueError(f'make sure the target effects are in the Augment FX chain : received fx called {cur_fx}') aug_chain_in = [] for (cur_i, cur_fx) in enumerate(fx_list): normalize = (False if (isinstance(cur_fx, AugmentationChain) or (cur_fx.name == 'Gain')) else True) aug_chain_in.append((cur_fx, apply_prob[cur_i], normalize)) return AugmentationChain(fxs=aug_chain_in, shuffle=shuffle, parallel=parallel, parallel_weight_factor=parallel_weight_factor)
class RecurrentTransformerEncoderLayer(Module): def __init__(self, attention, d_model, d_ff=None, dropout=0.1, activation='relu', event_dispatcher=''): super(RecurrentTransformerEncoderLayer, self).__init__() d_ff = (d_ff or (4 * d_model)) self.attention = attention self.linear1 = Linear(d_model, d_ff) self.linear2 = Linear(d_ff, d_model) self.norm1 = LayerNorm(d_model) self.norm2 = LayerNorm(d_model) self.dropout = Dropout(dropout) self.activation = (F.relu if (activation == 'relu') else F.gelu) self.event_dispatcher = EventDispatcher.get(event_dispatcher) def forward(self, x, state=None, memory=None): state = check_state(state, memory) (x2, state) = self.attention(x, x, x, state) x = (x + self.dropout(x2)) y = x = self.norm1(x) y = self.dropout(self.activation(self.linear1(y))) y = self.dropout(self.linear2(y)) return (self.norm2((x + y)), state)
def execute(code, stack, pos, storage, mmemory, data, trace, calldepth, debug, read_from_blockchain): op = code[pos]['o'] halt = False executed = True step = code[pos]['id'] if (op not in allops): print(('Unknown operation %s at pos %x' % (op, pos))) return (pos, True) if (allops[op][1] > len(stack)): if debug: print(('Not enough entries in the stack to execute the operation %8s at step %x: required %d, provided %d' % (op, code[pos]['id'], allops[op][1], len(stack)))) return (pos, True) start_stack_size = len(stack) final_stack_size = ((len(stack) - allops[op][1]) + allops[op][2]) args = [] if ((op.find('SWAP') < 0) and (op.find('DUP') < 0) and (op not in ['JUMPI'])): for i in range(allops[op][1]): args.append(stack.pop()) if (op in ['ISZERO', 'NOT']): stack.append(unary(args[0], step, op)) elif (op in ['ADD', 'MUL', 'SUB', 'DIV', 'SDIV', 'MOD', 'SMOD', 'EXP', 'AND', 'OR', 'XOR', 'LT', 'GT', 'SLT', 'SGT', 'EQ']): stack.append(binary(args[0], args[1], step, op)) elif (op in ['ADDMOD', 'MULMOD']): stack.append(ternary(args[0], args[1], args[2], step, op)) elif (op == 'SIGNEXTEND'): if ((not is_fixed(args[0])) or (not is_fixed(args[1]))): stack.append({'type': 'undefined', 'step': step}) else: o = get_value(args[1]) t = (256 - (8 * (get_value(args[0]) + 1))) tbit = ((o >> t) & 1) n = 0 for i in range(256): n ^= ((tbit if (i <= t) else ((o >> i) & 1)) << i) stack.append({'type': 'undefined', 'step': step, 'z3': BitVecVal(n, 256)}) elif (op == 'SHA3'): addr = simplify(args[0]['z3']) offset = simplify(args[1]['z3']) exact_address = (addr.as_long() if is_bv_value(addr) else (- 1)) exact_offset = (offset.as_long() if is_bv_value(offset) else (- 1)) res = {'type': 'undefined', 'step': step} if ((exact_address >= 0) and (exact_offset >= 0)): if ((exact_offset % 32) == 0): val = '' all_good = True for i in range((exact_offset / 32)): if (((exact_address + (i * 32)) not in mmemory) or (not is_fixed(mmemory[(exact_address + (i * 32))]))): all_good = False break val += ('%064x' % get_value(mmemory[(exact_address + (i * 32))])) if all_good: k = keccak_256() k.update(val.encode('utf-8')) digest = k.hexdigest() res = {'type': 'constant', 'step': step, 'z3': BitVecVal(int(digest, 16), 256)} if (MyGlobals.symbolic_sha and is_undefined(res)): res = {'type': 'constant', 'step': step, 'z3': BitVec(((('sha-' + str(step)) + '-') + str(calldepth)), 256)} stack.append(res) elif (op.find('PUSH') >= 0): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(code[pos]['input'], 16), 256)}) elif (op.find('DUP') >= 0): stack.append(copy.deepcopy(stack[(- int(op[3:]))])) elif (op.find('SWAP') >= 0): tmp1 = stack[(- 1)] tmp2 = stack[((- int(op[4:])) - 1)] stack[(- 1)] = tmp2 stack[((- int(op[4:])) - 1)] = tmp1 elif (op in MyGlobals.symbolic_vars): stack.append({'type': 'constant', 'step': step, 'z3': BitVec(((op + '-') + str(calldepth)), 256)}) elif (op == 'NUMBER'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('block_number', ''), 16), 256)}) elif (op == 'GASLIMIT'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('gas_limit', ''), 16), 256)}) elif (op == 'TIMESTAMP'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('time_stamp', ''), 16), 256)}) elif (op == 'CALLVALUE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('call_value', ''), 16), 256)}) elif (op == 'ADDRESS'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('contract_address', ''), 16), 256)}) elif (op == 'ORIGIN'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('contract_address', ''), 16), 256)}) elif (op == 'GASPRICE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('gas_price', ''), 16), 256)}) elif (op == 'COINBASE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(0, 256)}) elif (op == 'DIFFICULTY'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(0, 256)}) elif (op == 'CALLER'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('my_address', ''), 16), 256)}) elif (op == 'GAS'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('gas', ''), 16), 256)}) elif (op == 'MSIZE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(len(mmemory), 256)}) elif (op == 'BLOCKHASH'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(291, 256)}) elif (op == 'BALANCE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(int(get_params('contract_balance', ''), 10), 256)}) elif (op == 'POP'): pass elif (op.find('LOG') >= 0): pass elif (op == 'CODECOPY'): pass elif (op == 'JUMPDEST'): if (not is_good_jump(code, pos, debug)): return (pos, True) elif (op in ['STOP', 'RETURN', 'REVERT', 'INVALID', 'SUICIDE']): halt = True elif (op in ['CALLDATALOAD']): addr = args[0] if is_fixed(addr): addr = get_value(addr) if (((('data-' + str(calldepth)) + '-') + str(addr)) not in data): data[((('data-' + str(calldepth)) + '-') + str(addr))] = BitVec((((('input' + str(calldepth)) + '[') + str(addr)) + ']'), 256) stack.append({'type': 'constant', 'step': step, 'z3': data[((('data-' + str(calldepth)) + '-') + str(addr))]}) elif is_undefined(addr): if debug: print(('\x1b[95m[-] In CALLDATALOAD the input address cannot be determined at step %x: \x1b[0m' % code[pos]['id'])) print(addr) return (pos, True) else: stack.append(args[0]) return (pos, False) elif (op in ['CALLDATASIZE']): return (pos, False) elif (op == 'CALL'): if (is_fixed(args[5]) and is_fixed(args[6])): addr = get_value(args[5]) value = get_value(args[6]) if (value < 10000): for i in range((value / 32)): mmemory[(addr + (32 * i))] = {'type': 'undefined', 'step': step} stack.append({'type': 'constant', 'step': step, 'z3': (BitVec(('call_at_step_' + str(step)), 256) & 1)}) elif (op == 'CALLDATACOPY'): memaddr = args[0] datapos = args[1] length = args[2] if ((not is_fixed(memaddr)) or (not is_fixed(datapos)) or (not is_fixed(length))): if debug: print('\x1b[95m[-] In CALLDATACOPY the memory address or datapos or length cannot be determined \x1b[0m') print(memaddr) print(datapos) print(length) return (pos, True) memaddr = get_value(memaddr) datapos = get_value(datapos) length = get_value(length) if ((length % 32) != 0): if debug: print(('\x1b[95m[-] In CALLDATACOPY the length of array (%d) is not multiple of 32 \x1b[0m' % length)) return (pos, True) for i in range((length / 32)): data[(datapos + (32 * i))] = BitVec((((('input' + str(calldepth)) + '[') + str((datapos + (32 * i)))) + ']'), 256) store_in_memory(mmemory, (memaddr + (32 * i)), {'type': 'constant', 'step': step, 'z3': data[(datapos + (32 * i))]}) elif (op == 'CALLCODE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(0, 256)}) elif (op == 'DELEGATECALL'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(0, 256)}) elif (op == 'EXTCODESIZE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(0, 256)}) elif (op == 'CREATE'): stack.append({'type': 'constant', 'step': step, 'z3': BitVecVal(0, 256)}) elif (op == 'MLOAD'): addr = args[0] if is_undefined(addr): if debug: print(('\x1b[95m[-] The MLOAD address on %x cannot be determined\x1b[0m' % code[pos]['id'])) return (pos, True) addr = simplify(addr['z3']) if is_bv_value(addr): exact_address = addr.as_long() if (exact_address in mmemory): res = copy.deepcopy(mmemory[exact_address]) else: res = {'type': 'constant', 'step': step, 'z3': BitVecVal(0, 256)} stack.append(res) else: if debug: print(('\x1b[95m[-] The MLOAD address on %x cannot be determined\x1b[0m' % code[pos]['id'])) return (pos, True) elif (op == 'MSTORE'): addr = args[0] if (is_undefined(addr) or (not is_bv_value(simplify(addr['z3'])))): if debug: print(('\x1b[95m[-] The MSTORE the write address on %x cannot be determined\x1b[0m' % code[pos]['id'])) return (pos, True) t = copy.deepcopy(args[1]) addr = get_value(addr) store_in_memory(mmemory, addr, t) elif (op in ['MSTORE8']): addr = args[0] value = args[1] if (not is_fixed(addr)): if debug: print(('\x1b[95m[-] The MSTORE8 the write address on %x cannot be determined\x1b[0m' % code[pos]['id'])) return (pos, True) if (not is_fixed(value)): if debug: print(('\x1b[95m[-] The MSTORE8 value is undefined \x1b[0m' % code[pos]['id'])) return (pos, True) ea = get_value(addr) ev = (get_value(value) % 256) if (((ea / 32) * 32) not in mmemory): mmemory[((ea / 32) * 32)] = {'type': 'constant', 'step': step, 'z3': BitVecVal((ev << (31 - (ea % 32))), 256)} elif is_fixed(mmemory[((ea / 32) * 32)]['z3']): v = get_value(mmemory[((ea / 32) * 32)]['z3']) v = ((v & ((~ BitVecVal(255, 256)) << (31 - (ea % 32)))) ^ (ev << (31 - (ea % 32)))) mmemory[((ea / 32) * 32)]['z3'] = v elif (op == 'SLOAD'): addr = args[0] if is_undefined(addr): if debug: print(('\x1b[95m[-] The SLOAD address on %x cannot be determined\x1b[0m' % code[pos]['id'])) return (pos, True) addr = simplify(addr['z3']) if is_bv_value(addr): exact_address = addr.as_long() if (exact_address in storage): total_values = len(storage[exact_address]) if (total_values == 0): print(('In SLOAD the list at address %x has no elements ' % exact_address)) exit(0) return (pos, True) else: res = copy.deepcopy(storage[exact_address][0]) else: if ((MyGlobals.web3 is not None) and read_from_blockchain): value = MyGlobals.web3.eth.getStorageAt(get_params('contract_address', ''), exact_address) else: value = '0' t = {'type': 'constant', 'step': step, 'z3': BitVecVal(int(value, 16), 256)} storage[exact_address] = [t] res = copy.deepcopy(t) stack.append(res) elif MyGlobals.symbolic_load: stack.append({'type': 'constant', 'step': step, 'z3': BitVec(((('sload-' + str(step)) + '-') + str(calldepth)), 256)}) else: if debug: print(('\x1b[95m[-] The SLOAD address on %x cannot be determined\x1b[0m' % code[pos]['id'])) return (pos, True) elif (op == 'SSTORE'): addr = args[0] if is_undefined(addr): if debug: print(('\x1b[95m[-] The SSTORE address on %x cannot be determined\x1b[0m' % code[pos]['id'])) return (pos, True) t = copy.deepcopy(args[1]) if is_bv_value(simplify(addr['z3'])): va = get_value(addr) storage[va] = [t] elif MyGlobals.symbolic_load: pass else: if debug: print(('\x1b[95m[-] In SSTORE the write address cannot be determined at step %x: \x1b[0m' % code[pos]['id'])) print(addr) return (pos, True) elif (op == 'JUMP'): addr = args[0] if (not is_fixed(addr)): if debug: print('\x1b[95m[-] In JUMP the address cannot be determined \x1b[0m') return (pos, True) jump_dest = get_value(addr) if (jump_dest <= 0): if debug: print(('\x1b[95m[-] The JUMP destination is not a valid address : %x\x1b[0m' % jump_dest)) return (pos, True) new_position = find_pos(code, jump_dest) if (new_position < 0): if debug: print(('\x1b[95m[-] The code has no such JUMP destination: %s at line %x\x1b[0m' % (hex(jump_dest), code[pos]['id']))) return (pos, True) if (not is_good_jump(code, new_position, debug)): return (pos, True) return (new_position, False) elif (op == 'JUMPI'): return (pos, False) elif (op == 'BYTE'): byte_no = args[0] word = args[1] if (is_undefined(word) or is_undefined(byte_no)): res = {'type': 'undefined', 'step': step} else: res = {'type': 'constant', 'step': step, 'z3': ((word['z3'] >> (8 * (31 - byte_no['z3']))) & 255)} stack.append(res) else: executed = False if (executed and (final_stack_size != len(stack))): print(('Incorrect final stack size after executing %s at step %x' % (op, step))) print(len(stack)) print(final_stack_size) exit(2) return ((pos + 1), halt)
def put_in_middle(str1, str2): n = len(str1) m = len(str2) if (n <= m): return str2 else: start = ((n - m) // 2) return ((str1[:start] + str2) + str1[(start + m):])