MCPcopy Create free account

hub / github.com/bytedance/ABQ-LLM / functions

Functions14,558 in github.com/bytedance/ABQ-LLM

⨍ Functions 14,558◇ Types & classes 4,979
↓ 6 callersMethodload_state_dict
Load the state dicts of each of the models
fastertransformer/3rdparty/Megatron-LM/megatron/model/biencoder_model.py:164
↓ 6 callersMethodload_with_pointer_offset
Loads a fragment
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_pitch_linear_2dthreadtile.h:148
↓ 6 callersFunctionmetrics_from_confusion_matrix
Precision, Recall and F1 from the confusion matrix Parameters ---------- cm : tf.Tensor of type tf.int32, of shape (num_classes, num_cl
fastertransformer/examples/tensorflow/bert/bert-quantization/tf_metrics.py:168
↓ 6 callersMethodnext_tile
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/kernel/grouped_problem_visitor.h:207
↓ 6 callersMethodnumpy_type
(type)
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/test/gemm_testbed.py:216
↓ 6 callersFunctionoperator*
fastertransformer/3rdparty/cutlass/include/cutlass/half.h:796
↓ 6 callersFunctionoperator+
fastertransformer/3rdparty/cutlass/include/cutlass/half.h:769
↓ 6 callersFunctionoperator-
fastertransformer/3rdparty/cutlass/include/cutlass/half.h:778
↓ 6 callersFunctionoperator/
fastertransformer/3rdparty/cutlass/include/cutlass/half.h:805
↓ 6 callersMethodproblem_index
Gets the index of the problem
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/kernel/grouped_problem_visitor.h:143
↓ 6 callersMethodprocedural_name
(self)
fastertransformer/3rdparty/cutlass/tools/library/scripts/library.py:677
↓ 6 callersMethodreduction
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/cpp/include/epilogue/epilogue_visitor_op/visitor_op_row_reduction.h:299
↓ 6 callersMethodresetInterSize
fastertransformer/src/fastertransformer/layers/FfnLayer.h:146
↓ 6 callersMethodrun
(self, host_workspace, device_workspace, launch_config, stream=cuda.CUstream(0))
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/operation.py:94
↓ 6 callersMethodrun
Executes one test
fastertransformer/3rdparty/cutlass/test/unit/conv/device/conv2d_testbed.h:219
↓ 6 callersFunctionsafe_div
Safe division, return 0 if denominator is 0
fastertransformer/examples/tensorflow/bert/bert-quantization/tf_metrics.py:137
↓ 6 callersFunctionselect_field
(features, field)
fastertransformer/examples/pytorch/bert/bert-quantization-sparsity/run_swag.py:237
↓ 6 callersFunctionsequence_mask
(lengths, max_len=None, is_2d=True)
fastertransformer/examples/pytorch/bert/bert_example.py:36
↓ 6 callersFunctionsetClassAttr
(elem,attr)
fastertransformer/3rdparty/cutlass/docs/search/search.js:716
↓ 6 callersFunctionsplit_and_convert_process
(saved_dir, factor, key, val)
fastertransformer/examples/pytorch/llama2/utils/hugging_llamaV2_convert.py:33
↓ 6 callersMethodstep
Set lr for all parameters groups.
fastertransformer/3rdparty/Megatron-LM/megatron/learning_rates.py:94
↓ 6 callersMethodstore_with_pointer_offset
Stores a fragment
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_pitch_linear_2dthreadtile.h:189
↓ 6 callersFunctionsummarize_ft
(datapoint)
fastertransformer/examples/pytorch/t5/summarization.py:230
↓ 6 callersMethodswap
Swaps the managed objects with *this and another unique_ptr
fastertransformer/3rdparty/cutlass/include/cutlass/platform/platform.h:796
↓ 6 callersMethodthreadblock_idx
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/kernel/grouped_problem_visitor.h:148
↓ 6 callersMethodto_json_string
Serializes this instance to a JSON string.
fastertransformer/examples/pytorch/bert/bert-quantization-sparsity/modeling.py:300
↓ 6 callersMethodtok_encode
(self, string: str)
algorithm/lm_eval/base.py:147
↓ 6 callersMethodtok_encode
(self, string: str)
algorithm/models/models_utils.py:184
↓ 6 callersMethodtrace
Returns the sum of diagonal elements
fastertransformer/3rdparty/cutlass/include/cutlass/matrix.h:587
↓ 6 callersFunctiontransform_trt_mask
fastertransformer/src/fastertransformer/th_op/swin/SwinOp.cc:173
↓ 6 callersFunctionunpack_nemo_ckpt
( nemo_archive_path: typing.Union[str, pathlib.Path], out_dir_path: typing.Union[str, pathlib.Path], )
fastertransformer/examples/pytorch/nemo.py:29
↓ 5 callersFunctionCreateTrmmOperator
(manifest, layouts, side_modes, fill_modes, diag_types, tile_descriptions, data_type, \ alignment_constraint
fastertransformer/3rdparty/cutlass/tools/library/scripts/generator.py:239
↓ 5 callersFunctionCutlassUnitTestProblemCount
fastertransformer/3rdparty/cutlass/test/unit/common/filter_architecture.cpp:101
↓ 5 callersMethodInt8Gemm
fastertransformer/src/fastertransformer/utils/cublasMMWrapper.cc:1075
↓ 5 callersFunctionMaxAlignment
(fmt)
fastertransformer/3rdparty/cutlass/tools/library/scripts/rt.py:53
↓ 5 callersMethod__init__
(self, operation: 'GemmOperation')
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/gemm_operation.py:525
↓ 5 callersMethod__init__
(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.)
fastertransformer/examples/pytorch/swin/Swin-Transformer-Quantization/SwinTransformer/models/swin_mlp.py:16
↓ 5 callersFunction__internal_halfraw_to_float
fastertransformer/tests/unittests/fp8_gemm_test/2022_03_21__fp8_stride_batch_example/include/cuda_fp8.hpp:284
↓ 5 callersMethod_check_and_set
Auxiliary function for checking the values in the checkpoint and setting them.
fastertransformer/3rdparty/Megatron-LM/megatron/learning_rates.py:114
↓ 5 callersMethod_map
(self, func)
fastertransformer/examples/pytorch/gpt/utils/gpt_decoder.py:180
↓ 5 callersFunction_set_cuda_rng_state
Sets the random number generator state of the current GPU. Argumentss: new_state (torch.ByteTensor): The desired state This function
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/random.py:39
↓ 5 callersMethodadd
Track the rng state.
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/random.py:213
↓ 5 callersMethodadd_pointer_offset
Adds a pointer offset in units of Element
fastertransformer/3rdparty/cutlass/examples/42_fused_multi_head_attention/iterators/predicated_tile_access_iterator_residual_last.h:809
↓ 5 callersMethodadd_pointer_offset
Adds a pointer offset in units of Element
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_tensor_op_sm70.h:477
↓ 5 callersMethodadd_pointer_offset
Adds a pointer offset in units of Element
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/predicated_tile_access_iterator.h:949
↓ 5 callersMethodadd_tile_offset
Adds a tile offset
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_tensor_op_sm70.h:483
↓ 5 callersMethodadd_tile_offset
Advances an iterator along logical dimensions of matrix in units of whole tiles
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/predicated_tile_access_iterator.h:956
↓ 5 callersFunctionbackward_step
Backward step through passed-in output tensor. If last stage, output_tensor_grad is None, otherwise gradient of loss with respect to stage's
fastertransformer/3rdparty/Megatron-LM/megatron/schedules.py:140
↓ 5 callersMethodbatchedGemm
fastertransformer/src/fastertransformer/utils/gemm.cc:372
↓ 5 callersFunctionbert_position_ids
(token_ids)
fastertransformer/3rdparty/Megatron-LM/megatron/model/bert_model.py:48
↓ 5 callersFunctionbuild_data_loader
Data loader. Note that batch-size is the local (per GPU) batch-size.
fastertransformer/3rdparty/Megatron-LM/tasks/vision/finetune_utils.py:73
↓ 5 callersFunctionbuild_dataset
(is_train, config)
fastertransformer/examples/pytorch/swin/Swin-Transformer-Quantization/SwinTransformer/data/build.py:98
↓ 5 callersFunctionbuild_loader
(config, args)
fastertransformer/examples/pytorch/vit/ViT-quantization/data.py:69
↓ 5 callersFunctionbuild_scheduler
(config, optimizer, n_iter_per_epoch)
fastertransformer/examples/pytorch/swin/Swin-Transformer-Quantization/SwinTransformer/lr_scheduler.py:14
↓ 5 callersFunctionbuild_tokenizer
Initialize tokenizer.
fastertransformer/3rdparty/Megatron-LM/megatron/tokenizer/tokenizer.py:25
↓ 5 callersFunctioncached_path
Given something that might be a URL (or might be a local path), determine which. If it's a URL, download the file and cache it, and retur
fastertransformer/examples/pytorch/bert/bert-quantization-sparsity/file_utils.py:97
↓ 5 callersFunctioncalGemmTestBufSizeInByte
fastertransformer/src/fastertransformer/utils/gemm_test/gemm_func.cc:855
↓ 5 callersFunctioncompute_ngrams_insert_dict
(args, text, ngrams)
fastertransformer/3rdparty/Megatron-LM/tools/openwebtext/filter_ngrams.py:194
↓ 5 callersMethodcuLaunchKernel
fastertransformer/3rdparty/trt_fused_multihead_attention/cudaDriverWrapper.cpp:132
↓ 5 callersMethodcuda
(self)
fastertransformer/examples/pytorch/gptneox/utils/gptneox.py:247
↓ 5 callersMethodcustom
(start, end)
fastertransformer/3rdparty/Megatron-LM/megatron/model/transformer.py:626
↓ 5 callersFunctiondeserialize
fastertransformer/src/fastertransformer/models/bert_fp8/serialize.hpp:57
↓ 5 callersFunctiondeserialize_h2d
fastertransformer/src/fastertransformer/models/bert_fp8/serialize.hpp:77
↓ 5 callersMethodemit
(self)
fastertransformer/3rdparty/cutlass/tools/library/scripts/rt.py:533
↓ 5 callersMethodencode
(self, text)
fastertransformer/examples/tensorflow/gpt/utils/gpt_token_encoder.py:134
↓ 5 callersFunctionevaluate_and_print_results
Helper function to evaluate and dump results on screen.
fastertransformer/3rdparty/Megatron-LM/megatron/training.py:811
↓ 5 callersFunctionfast_max
fastertransformer/3rdparty/cutlass/include/cutlass/fast_math.h:612
↓ 5 callersMethodflops
(self)
fastertransformer/examples/pytorch/swin/Swin-Transformer-Quantization/models/swin_transformer.py:463
↓ 5 callersMethodfork
Fork the cuda rng state, perform operations, and exit with the original state.
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/random.py:231
↓ 5 callersFunctionft_encoder_opennmt
Run the bert transformer layer by FasterTransformer. Args: inputs: A tf.Tensor with shape [batch_size, seq_len, hidden_dimension].
fastertransformer/examples/tensorflow/encoder/utils/encoder.py:203
↓ 5 callersMethodget
Returns a pointer
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/predicated_tile_access_iterator.h:962
↓ 5 callersMethodgetOnDiag
Return if the address in on the diagonal
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/predicated_tile_access_iterator_triangular_matrix.h:874
↓ 5 callersFunctiongetTensorRef
(tensor, tensor_layout, conv_kind, problem_size, operand)
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/test/conv2d_testbed.py:44
↓ 5 callersFunctiongetUnfusedAttentionType
fastertransformer/src/fastertransformer/layers/attention_layers/BaseAttentionLayer.h:130
↓ 5 callersFunctionget_cuda_rng_tracker
Get cuda rng tracker.
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/random.py:255
↓ 5 callersMethodget_labels
Gets the list of labels for this data set.
fastertransformer/examples/pytorch/bert/bert-quantization-sparsity/processors/glue.py:65
↓ 5 callersMethodget_original
(self, newarr)
algorithm/lm_eval/utils.py:132
↓ 5 callersFunctionget_rank
()
fastertransformer/examples/pytorch/bert/bert-quantization-sparsity/utils.py:20
↓ 5 callersMethodget_reordered
(self)
algorithm/lm_eval/utils.py:129
↓ 5 callersFunctionigemm_config_INT8IO
fastertransformer/src/fastertransformer/utils/gemm_test/swin_igemm_func.cc:136
↓ 5 callersFunctioninfer_latency_sli
(threshold)
fastertransformer/examples/pytorch/bert/bert-quantization-sparsity/run_glue.py:875
↓ 5 callersMethodinitialize
Initializes the workspace
fastertransformer/3rdparty/cutlass/tools/library/src/gemm_operation.h:268
↓ 5 callersFunctioninitialize_distributed
Initialize torch.distributed.
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/tests/commons.py:42
↓ 5 callersMethodinverse
fastertransformer/3rdparty/cutlass/include/cutlass/layout/tensor_op_multiplicand_sm70.h:375
↓ 5 callersMethodinverse
fastertransformer/3rdparty/cutlass/include/cutlass/layout/tensor_op_multiplicand_sm75.h:301
↓ 5 callersFunctionisUnPaddedMHA
fastertransformer/src/fastertransformer/layers/attention_layers/BaseAttentionLayer.h:120
↓ 5 callersFunctionlayer_norm
Run layer normalization on the last dimension of the tensor.
fastertransformer/examples/tensorflow/bert/tensorflow_bert/my_modeling.py:368
↓ 5 callersMethodload
Loads a fragment from memory at the location pointed to by the iterator.
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/warp/mma_tensor_op_tile_iterator_sm80.h:2369
↓ 5 callersMethodload_from_bin
(self, ckpt_path, model_type)
fastertransformer/examples/pytorch/t5/utils/ft_decoding.py:168
↓ 5 callersMethodload_from_bin
(self, ckpt_path, model_type)
fastertransformer/examples/pytorch/t5/utils/ft_encoder.py:134
↓ 5 callersMethodload_state_dict
Load the state dicts of each of the models
fastertransformer/3rdparty/Megatron-LM/megatron/model/realm_model.py:105
↓ 5 callersMethodload_state_dict
Customized load.
fastertransformer/3rdparty/Megatron-LM/megatron/model/t5_model.py:195
↓ 5 callersFunctionload_with_byte_offset
Loads a fragment from memory
fastertransformer/3rdparty/cutlass/examples/42_fused_multi_head_attention/iterators/epilogue_predicated_tile_iterator.h:339
↓ 5 callersFunctionload_with_pointer_offset
Loads a fragment from memory with additional logical offset
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/warp/mma_tensor_op_tile_iterator_sm80.h:301
↓ 5 callersMethodload_with_pointer_offset
Loads a fragment from memory
fastertransformer/3rdparty/cutlass/examples/42_fused_multi_head_attention/iterators/predicated_tile_iterator_residual_last.h:875
↓ 5 callersMethodload_with_pointer_offset
Loads a fragment from memory with additional logical offset
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/warp/mma_tensor_op_tile_iterator_sm80.h:2376
↓ 5 callersMethodload_with_pointer_offset
Loads a fragment from memory
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_tensor_op_sm70.h:507
↓ 5 callersMethodload_with_pointer_offset
Loads a fragment from memory
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator.h:780
↓ 5 callersMethodload_with_pointer_offset
Loads a fragment
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_pitch_linear.h:339
↓ 5 callersFunctionmake_attention_mask
Returns a 2-dimensional (2-D) attention mask :param source_block: 1-D array :param target_block: 1-D array
fastertransformer/3rdparty/Megatron-LM/megatron/data/biencoder_dataset_utils.py:12
← previousnext →701–800 of 14,558, ranked by callers