Functions — github.com/DAMO-DI-ML/NeurIPS2022-FiLM

hub / github.com/DAMO-DI-ML/NeurIPS2022-FiLM / functions

Functions571 in github.com/DAMO-DI-ML/NeurIPS2022-FiLM

⨍ Functions 571◇ Types & classes 171

Methodforward

(self, x, keys = None, input_mask = None, input_attn_mask = None, context_mask = None, pos_emb = None, **kwarg

models/reformer_pytorch/reformer_pytorch.py:534

Methodforward

(self, x)

models/reformer_pytorch/reformer_pytorch.py:606

Methodforward

(self, x, **kwargs)

models/reformer_pytorch/reformer_pytorch.py:622

Methodforward

(self, x)

models/reformer_pytorch/reformer_pytorch.py:641

Methodforward

(self, x, seq_dim = 1)

models/reformer_pytorch/reformer_pytorch.py:651

Methodforward

(self, x, **kwargs)

models/reformer_pytorch/reformer_pytorch.py:720

Methodforward

(self, x, **kwargs)

models/reformer_pytorch/reformer_pytorch.py:760

Methodforward

(self, x, return_loss = False, **kwargs)

models/reformer_pytorch/generative_tools.py:77

Methodforward

(self, seq_in, seq_out, return_loss = False, **kwargs)

models/reformer_pytorch/reformer_enc_dec.py:65

Methodforward

(self, x, **kwargs)

models/reformer_pytorch/autopadder.py:29

Methodforward_diff

Computes the 'forward diff' or Euler update rule: (I - d A)^-1 u + d B v d: (...) u: (..., n) v: (...)

utils/op.py:93

Methodforward_mult

Computes (I + d A) u A: (n, n) u: (b1* d, n) d represents memory_size delta: (b2*, d) or scalar Assume len(b2) <=

utils/op.py:183

Functionfunc2

(*args, **kw)

layers/AutoCorrelation.py:15

Methodgenerate

(self, start_tokens, seq_len, eos_token = None, temperature = 1., filter_logits_fn = top_k, filter_thres = 0.9

models/reformer_pytorch/generative_tools.py:38

Methodgenerate

(self, seq_in, seq_out_start, seq_len, **kwargs)

models/reformer_pytorch/reformer_enc_dec.py:60

Functionget_subsequent_mask

Get causal attention mask for decoder.

models/pyraformer/Layers.py:69

Functiongroup_by_key_prefix

(prefix, d)

models/reformer_pytorch/reformer_enc_dec.py:20

Functioninner_fn

(*args, **kwargs)

layers/LSHAttention_reformer.py:29

Functioninner_fn

(*args, **kwargs)

models/reformer_pytorch/reformer_pytorch.py:37

Methodinverse_mult

Computes (I - d A)^-1 u

utils/op.py:223

Methodinverse_transform

(self, data)

utils/tools.py:75

Methodinverse_transform

(self, data)

data_provider/data_loader.py:98

Methodinverse_transform

(self, data)

data_provider/data_loader.py:188

Methodinverse_transform

(self, data)

data_provider/data_loader.py:287

Methodinverse_transform

(self, data)

data_provider/data_loader_mzq.py:99

Methodinverse_transform

(self, data)

data_provider/data_loader_mzq.py:185

Methodinverse_transform

(self, data)

data_provider/data_loader_mzq.py:285

Functionkrylov

Compute the Krylov matrix (b, Ab, A^2b, ...) using the squaring trick. If return_power=True, return A^{L-1} as well

layers/S4.py:206

Methodlook_one_back

(x)

layers/LSHAttention_reformer.py:251

Methodlook_one_back

(x)

models/reformer_pytorch/reformer_pytorch.py:325

(ind_from, ind_to, tensor)

layers/LSHAttention_reformer.py:89

Functionmerge_dims

(ind_from, ind_to, tensor)

models/reformer_pytorch/reformer_pytorch.py:97

Methodmerge_heads

(v)

models/reformer_pytorch/reformer_pytorch.py:552

Functionphi_

(phi_c, x, lb = 0, ub = 1)

layers/utils.py:18

Methodprecompute

deltas: list of step sizes

utils/op.py:136

Methodpredict

(self, setting, load=False)

exp/exp_main.py:317

Functionprocess_inputs_chunk

(fn, chunks=1, dim=0)

layers/LSHAttention_reformer.py:28

(self, attn, buckets)

models/reformer_pytorch/recorder.py:45

Methodsetup_step

Set up dA, dB, dC discretized parameters for stepping

layers/S4.py:790

Functionsoftmax_complex

Complex-valued Neural Networks with Non-parametric Activation Functions (Eq. 36) https://arxiv.org/pdf/1802.08026.pdf

layers/mwt.py:406

Functionsplit_at_index

(dim, index, t)

layers/LSHAttention_reformer.py:95

Functionsplit_at_index

(dim, index, t)

models/reformer_pytorch/reformer_pytorch.py:103

Methodsplit_heads

(v)

models/reformer_pytorch/reformer_pytorch.py:555

Methodstate_to_tensor

(self)

layers/S4.py:1113

Methodstep

Must have called self.setup_step() and created state with self.default_state() before calling this

layers/S4.py:869

Methodstep

(self, u, state, **kwargs)

layers/S4.py:956

Functiontest

(model, test_loader, device, verbose=0, lossFn=None, post_proc = lambda args: args)

(self, setting, test=0)

exp/exp_main.py:232

Methodtest

(self, data, v)

models/pyraformer/Pyraformer_SS.py:81

Functiontest_correctness

()

utils/unroll.py:454

Functiontest_speed

(variable=False, it=1)

utils/unroll.py:591

Functiontest_toeplitz

()

utils/unroll.py:538

Methodtime_delay_agg_full

Standard version of Autocorrelation

layers/AutoCorrelation.py:125

Methodtime_delay_agg_mzq

(self, values, corr)

layers/AutoCorrelation.py:49

Functiontop_k

(logits, thres = 0.9)

models/reformer_pytorch/generative_tools.py:20

Functiontop_p

(logits, thres = 0.9)

models/reformer_pytorch/generative_tools.py:9

Functiontrain

(model, train_loader, optimizer, epoch, device, verbose = 0, lossFn = None, lr_schedule=None, post_pr

models/reformer_pytorch/recorder.py:37

Methodturn_on

(self)

models/reformer_pytorch/recorder.py:34

Methodtvm_callback_cuda_compile

Use nvcc compiler for better perf.

models/pyraformer/hierarchical_mm_tvm.py:37

(self, *args, key_namespace=None, fetch=False, set_cache=True, **kwargs)

layers/LSHAttention_reformer.py:66

Functionwrapper

(self, *args, key_namespace=None, fetch=False, set_cache=True, **kwargs)

models/reformer_pytorch/reformer_pytorch.py:74

← previous501–571 of 571, ranked by callers