MCPcopy Create free account

hub / github.com/Tencent/TurboTransformers / functions

Functions392 in github.com/Tencent/TurboTransformers

⨍ Functions 392◇ Types & classes 103
↓ 112 callersMethoddevice_type
turbo_transformers/core/tensor.h:184
↓ 87 callersFunctionconvert2tt_tensor
(t)
turbo_transformers/python/turbo_transformers/layers/utils.py:34
↓ 63 callersFunctiontry_convert
(t)
turbo_transformers/python/turbo_transformers/layers/utils.py:38
↓ 53 callersMethoddata
turbo_transformers/core/tensor.h:173
↓ 41 callersMethodn_dim
turbo_transformers/core/tensor.h:131
↓ 41 callersMethodnumel
turbo_transformers/core/tensor.h:147
↓ 33 callersFunctionconvert_returns_as_type
(tensor: cxx.Tensor, rtype: Optional[ReturnType] )
turbo_transformers/python/turbo_transformers/layers/return_type.py:33
↓ 31 callersFunctioncreate_empty_if_none
(output)
turbo_transformers/python/turbo_transformers/layers/utils.py:58
↓ 31 callersMethodfrom_torch
(layer: TorchBertLayer)
turbo_transformers/python/turbo_transformers/layers/modeling_bert.py:245
↓ 26 callersMethoddevice_ctx
turbo_transformers/core/tensor.h:193
↓ 18 callersFunctionMatMul
turbo_transformers/layers/kernels/mat_mul.cpp:30
↓ 18 callersMethodend_profile
turbo_transformers/core/profiler.cpp:130
↓ 18 callersMethodstart_profile
turbo_transformers/core/profiler.cpp:125
↓ 15 callersMethodstream
turbo_transformers/core/cuda_device_context.cpp:36
↓ 14 callersMethoddevice_id
turbo_transformers/core/tensor.h:189
↓ 14 callersFunctionfloat_eq
turbo_transformers/layers/kernels/mat_mul_test.cpp:27
↓ 13 callersFunctionMatmulBenchmarkHelper
benchmark/turbo_transformers/layers/kernels/matmul_benchmark.cpp:27
↓ 13 callersFunctionis_same_device_ctx
turbo_transformers/layers/kernels/common.cpp:21
↓ 10 callersFunctionTestFuncSpeed
benchmark/turbo_transformers/layers/kernels/benchmark_help.h:73
↓ 10 callersMethodclear
turbo_transformers/core/profiler.cpp:140
↓ 9 callersFunctionMemcpy
turbo_transformers/core/memory.cpp:56
↓ 9 callersMethodis_null
turbo_transformers/core/tensor.h:198
↓ 8 callersFunctionCopy
turbo_transformers/core/tensor_copy.h:28
↓ 8 callersMethodfrom_onmt
(model: OnmtTransformerDecoder, device: Optional[torch.device] = None)
turbo_transformers/python/turbo_transformers/layers/modeling_decoder.py:623
↓ 8 callersMethodfrom_pretrained
(model_id_or_path: str, device: Optional[torch.device] = None,
turbo_transformers/python/turbo_transformers/layers/modeling_bert.py:574
↓ 7 callersFunctionAddBiasTransposeForScore
add bias and transpose(2,3)
turbo_transformers/layers/kernels/transpose.cpp:115
↓ 7 callersMethodfrom_torch
(bert_layer)
turbo_transformers/python/turbo_transformers/layers/qmodeling_bert.py:70
↓ 6 callersFunctionApplyMaskAndSoftmax
turbo_transformers/layers/kernels/softmax.cpp:83
↓ 6 callersMethodBertModel
example/cpp/bert_model.cpp:238
↓ 6 callersMethod__call__
(self, hidden_states: AnyTensor, attention_mask: Optional[AnyTensor] = None,
turbo_transformers/python/turbo_transformers/layers/modeling_bert.py:221
↓ 6 callersMethodcublas_handle
turbo_transformers/core/cuda_device_context.cpp:38
↓ 6 callersMethodfrom_npz
(file_name: str, layer_num: int, num_attention_heads: int)
turbo_transformers/python/turbo_transformers/layers/modeling_bert.py:251
↓ 5 callersMethodfrom_torch
( model: TorchBertModel, # from_torch函数实现 device: Optional[torch.device] = None)
example/python/bert_for_sequence_classification_example.py:57
↓ 4 callersFunctionAddInputBias
turbo_transformers/layers/kernels/utils.cpp:134
↓ 4 callersFunctionCheckCppBert
example/cpp/bert_model_test.cpp:27
↓ 4 callersFunctionCheckCppBertWithPooler
example/cpp/bert_model_test.cpp:60
↓ 4 callersFunctionIsCompiledWithCUDA
turbo_transformers/core/config.h:23
↓ 4 callersFunctionLayerNormBenmarkHelper
benchmark/turbo_transformers/layers/kernels/layernorm_benchmark.cpp:28
↓ 4 callersFunctionSplitAddBiasTransposeForScore
turbo_transformers/layers/kernels/transpose.cpp:149
↓ 4 callersFunctionToMemcpyFlag
turbo_transformers/core/memory.cpp:73
↓ 4 callersFunctionallocate_impl
turbo_transformers/core/allocator.cpp:57
↓ 4 callersMethodfrom_pretrained
(model_id_or_path: str, device: Optional[torch.device] = None)
example/python/bert_for_sequence_classification_example.py:70
↓ 3 callersFunctionAddBias
turbo_transformers/layers/kernels/utils.cpp:103
↓ 3 callersFunctionRandomFillHost
turbo_transformers/layers/kernels/common.h:29
↓ 3 callersMethodSub
turbo_transformers/loaders/npz_load.h:36
↓ 3 callersFunctionTransposeForScore
turbo_transformers/layers/kernels/transpose.cpp:80
↓ 3 callersFunction_to_param_dict_naive
(torch_module: torch.nn.Module)
turbo_transformers/python/turbo_transformers/layers/modeling_albert.py:43
↓ 3 callersMethodallocate
turbo_transformers/core/allocator.cpp:207
↓ 3 callersMethodfree
turbo_transformers/core/allocator.cpp:220
↓ 3 callersFunctionfree_impl
turbo_transformers/core/allocator.cpp:71
↓ 3 callersMethodfrom_torch
(torch_albert_layer: TorchAlbertLayer)
turbo_transformers/python/turbo_transformers/layers/modeling_albert.py:159
↓ 3 callersFunctionto_param_dict
(torch_module: torch.nn.Module)
turbo_transformers/python/turbo_transformers/layers/utils.py:54
↓ 2 callersFunctionBatchMatMul
turbo_transformers/layers/kernels/mat_mul.cpp:112
↓ 2 callersFunctionFill
turbo_transformers/layers/kernels/common.cpp:55
↓ 2 callersFunctionSetNumThreads
turbo_transformers/core/config.cpp:23
↓ 2 callersFunctionSoftmaxBenchmarkHelper
benchmark/turbo_transformers/layers/kernels/softmax_benchmark.cpp:28
↓ 2 callersFunctionSplitAddTransposeBenchmarkHelper
benchmark/turbo_transformers/layers/kernels/transpose_benchmark.cpp:28
↓ 2 callersMethod__call__
Decode, possibly stepwise.
turbo_transformers/python/turbo_transformers/layers/modeling_decoder.py:549
↓ 2 callersMethod__call__
( self, input_tensor: AnyTensor, attention_mask: AnyTensor, he
turbo_transformers/python/turbo_transformers/layers/modeling_albert.py:126
↓ 2 callersFunctionalign_alloc
turbo_transformers/core/memory.cpp:24
↓ 2 callersMethodalloc
turbo_transformers/core/allocator.cpp:101
↓ 2 callersFunctioncblas_sgemm_batch
turbo_transformers/core/blas_blis.cpp:17
↓ 2 callersFunctioncheck_cpu_gpu_res
turbo_transformers/layers/kernels/mat_mul_test.cpp:80
↓ 2 callersMethodcheck_dlpack
(self, use_cuda)
turbo_transformers/python/tests/tensor_conversion_test.py:22
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda, num_iter=1)
turbo_transformers/python/tests/bert_attention_test.py:66
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda, num_iter=2)
turbo_transformers/python/tests/albert_attention_test.py:59
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/bert_embedding_test.py:57
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda, num_iter=1)
turbo_transformers/python/tests/decoder_multi_headed_attention_test.py:92
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/bert_model_test.py:44
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/bert_intermediate_test.py:47
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/albert_embedding_test.py:70
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/roberta_model_test.py:45
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/sequence_pool_test.py:43
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/bert_layer_test.py:61
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/albert_model_test.py:52
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda, num_iter=1)
turbo_transformers/python/tests/decoder_transformer_decoder_layer_test.py:55
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/bert_pooler_test.py:46
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda, num_iter=1)
turbo_transformers/python/tests/positionwise_ffn_test.py:57
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/bert_output_test.py:56
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda=True)
turbo_transformers/python/tests/bert_encoder_test.py:59
↓ 2 callersMethodcheck_torch_and_turbo
(self, use_cuda)
turbo_transformers/python/tests/gpt2_model_test.py:44
↓ 2 callersFunctioncreate_test
(batch_size, key_seq_len, query_seq_len, attn_type,
turbo_transformers/python/tests/decoder_multi_headed_attention_test.py:29
↓ 2 callersMethoddisable
turbo_transformers/core/profiler.cpp:146
↓ 2 callersMethodenable
turbo_transformers/core/profiler.cpp:142
↓ 2 callersFunctionget_head_mask
# Prepare head mask if needed # 1.0 in head_mask indicate we keep the head attention_probs has shape bsz x n_heads x N x N Arguments:
turbo_transformers/python/turbo_transformers/layers/utils.py:62
↓ 2 callersFunctionis_same_shape
turbo_transformers/layers/kernels/common.cpp:25
↓ 2 callersFunctiononnxruntime_benchmark_creator
(backend: str)
benchmark/onnx_benchmark_helper.py:81
↓ 2 callersMethodoperator()
example/cpp/bert_model.cpp:242
↓ 2 callersMethodpack_parameter
(multi_headed_attn: OnmtMultiHeadedAttention, is_trans_weight: Optional[bool] = False)
turbo_transformers/python/turbo_transformers/layers/modeling_decoder.py:100
↓ 2 callersFunctiontest
example/cpp/bert_model_example.cpp:24
↓ 2 callersFunctiontest_multiple_threads
example/cpp/bert_model_example.cpp:55
↓ 2 callersFunctiontest_multiple_threads
example/cpp/bert_model_test.cpp:122
↓ 2 callersFunctionto_param_dict_convert_tt
(torch_module: torch.nn.Module)
turbo_transformers/python/turbo_transformers/layers/utils.py:47
↓ 2 callersFunctionvsTanh
turbo_transformers/core/blas_blis.cpp:42
↓ 1 callersFunctionAddBiasTransposeForScoreImpl
tranpose(2,3) input (B x seq_len x head_num * hidden_size) bias (head * hidden_size)
turbo_transformers/layers/kernels/transpose.cpp:35
↓ 1 callersFunctionBindConfig
turbo_transformers/python/pybind.cpp:54
↓ 1 callersMethodElapseSecond
benchmark/turbo_transformers/layers/kernels/benchmark_help.h:40
↓ 1 callersFunctionFill
turbo_transformers/core/tensor_test.cpp:39
↓ 1 callersMethodFusedAddBiasGELU
turbo_transformers/layers/addbias_act.h:24
↓ 1 callersMethodFusedAddBiasLayerNorm
turbo_transformers/layers/addbias_layernorm.h:24
next →1–100 of 392, ranked by callers