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
↓ 3 callersFunctionget_model
(model_type)
fastertransformer/3rdparty/Megatron-LM/tools/merge_mp_partitions.py:110
↓ 3 callersFunctionget_name_prefix
(tensor_name)
fastertransformer/examples/tensorflow/bert/bert-quantization/ft-tensorflow-quantization/ft_tensorflow_quantization/python/calib/calibrator.py:332
↓ 3 callersMethodget_pooled_output
(self)
fastertransformer/examples/tensorflow/bert/tensorflow_bert/bert/modeling.py:234
↓ 3 callersMethodget_sequence_output
Gets final hidden layer of encoder. Returns: float Tensor of shape [batch_size, seq_length, hidden_size] corresponding to the final h
fastertransformer/examples/tensorflow/bert/tensorflow_bert/bert/modeling.py:237
↓ 3 callersFunctionget_shape_list
(tensor, expected_rank=None, name=None)
fastertransformer/examples/tensorflow/encoder/utils/encoder.py:135
↓ 3 callersMethodget_states
Get rng states. Copy the dictionary so we have direct pointers to the states, not just a pointer to the dictionary.
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/random.py:200
↓ 3 callersMethodget_tensor_ref
( self, tensor, dtype, tensor_layout, problem_size, operand)
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/gemm_operation.py:223
↓ 3 callersFunctionget_tensorboard_writer
Return tensorboard writer. It can be None so no need to check if it is initialized.
fastertransformer/3rdparty/Megatron-LM/megatron/global_vars.py:63
↓ 3 callersMethodget_test_examples
Gets a collection of `InputExample`s for prediction.
fastertransformer/examples/tensorflow/bert/tensorflow_bert/bert/run_classifier.py:188
↓ 3 callersMethodget_train_examples
Gets a collection of `InputExample`s for the train set.
fastertransformer/examples/pytorch/bert/bert-quantization-sparsity/processors/glue.py:57
↓ 3 callersMethodget_workspace_size
(self, arguments)
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/gemm_operation.py:731
↓ 3 callersFunctionget_world_size
()
fastertransformer/examples/pytorch/vit/ViT-quantization/ViT-pytorch/utils/dist_util.py:10
↓ 3 callersFunctiongpt_generate_fn
()
fastertransformer/examples/pytorch/gpt/multi_gpu_gpt_example.py:301
↓ 3 callersFunctionhandle_exceptions
(model_config: PretrainedConfig, param_name: str, param: torch.nn.
fastertransformer/examples/pytorch/gpt/utils/huggingface_bloom_convert.py:205
↓ 3 callersMethodhas_validation_docs
Whether the task has a validation set
algorithm/lm_eval/base.py:459
↓ 3 callersFunctionhash_args
(attr, args)
algorithm/lm_eval/base.py:763
↓ 3 callersMethodids_tensor
Creates a random int32 tensor of the shape within the vocab size.
fastertransformer/examples/tensorflow/bert/tensorflow_bert/bert/modeling_test.py:147
↓ 3 callersFunctionigemm_config
fastertransformer/src/fastertransformer/utils/gemm_test/encoder_igemm_func.cc:1013
↓ 3 callersFunctioninit_onmt_cache
(layer_num, memory_bank)
fastertransformer/examples/pytorch/decoder/utils/decoder.py:48
↓ 3 callersFunctioninit_op_cache
(layer_num, batch_size, beam_width, max_seq_len, \ decoding_max_seq_len, head_num, size_per_
fastertransformer/examples/pytorch/decoder/utils/decoder.py:30
↓ 3 callersMethodinit_state
Initialize decoder state.
fastertransformer/examples/pytorch/decoding/utils/decoding.py:293
↓ 3 callersMethodinitialize
Initializes data structures
fastertransformer/3rdparty/cutlass/test/unit/gemm/device/testbed_gemm_with_reduction.h:186
↓ 3 callersMethodinitialize
Initializes data structures
fastertransformer/3rdparty/cutlass/test/unit/gemm/device/testbed_gemm_with_broadcast.h:195
↓ 3 callersMethodinitialize
fastertransformer/3rdparty/cutlass/examples/13_two_tensor_op_fusion/b2b_interleaved_conv2d_run.h:496
↓ 3 callersMethodinitialize_word_embeddings
(self, init_method_normal)
fastertransformer/3rdparty/Megatron-LM/megatron/model/module.py:63
↓ 3 callersFunctioninvokeAddFusedQKVBiasTranspose
fastertransformer/src/fastertransformer/kernels/unfused_attention_kernels.h:117
↓ 3 callersFunctionisCudaSuccess
engine/test/test_mma/test_mma.h:47
↓ 3 callersFunctionisCudaSuccess
engine/test/test_wmma/test_wmma.h:47
↓ 3 callersFunctionisPerfectSquare
(x)
fastertransformer/3rdparty/Megatron-LM/megatron/model/vision/vit_backbone.py:63
↓ 3 callersMethodis_complex
(self)
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/gemm_operation.py:819
↓ 3 callersFunctionj
(bw,e)
fastertransformer/3rdparty/cutlass/docs/jquery.js:16
↓ 3 callersFunctionlayer_norm
(input_tensor, name=None)
fastertransformer/examples/tensorflow/encoder/utils/encoder.py:21
↓ 3 callersFunctionlayer_norm
Run layer normalization on the last dimension of the tensor.
fastertransformer/examples/tensorflow/bert/tensorflow_bert/bert/modeling.py:362
↓ 3 callersFunctionlayer_norm
Run layer normalization on the last dimension of the tensor.
fastertransformer/examples/tensorflow/bert/bert-quantization/modeling.py:385
↓ 3 callersMethodload
(self, ckpt_path)
fastertransformer/examples/pytorch/gptneox/utils/gptneox.py:237
↓ 3 callersMethodload_from
(self, weights)
fastertransformer/examples/pytorch/vit/ViT-quantization/ViT-pytorch/models/modeling.py:283
↓ 3 callersFunctionload_state_dict
Load weights from model file `safetensors` or `pytorch binary` is supported # Args. file_path: model file path, ends with .bin or .
fastertransformer/examples/pytorch/gpt/utils/huggingface_bloom_convert.py:308
↓ 3 callersMethodload_state_dict
(self, state_dict)
fastertransformer/3rdparty/Megatron-LM/megatron/optimizer/optimizer.py:555
↓ 3 callersFunctionload_the_npy_file
fastertransformer/examples/cpp/xlnet/cnpy.cpp:201
↓ 3 callersMethodload_with_byte_offset
Loads a fragment from memory
fastertransformer/3rdparty/cutlass/examples/42_fused_multi_head_attention/iterators/predicated_tile_iterator_residual_last.h:881
↓ 3 callersMethodload_with_byte_offset
Loads a fragment from memory
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator.h:786
↓ 3 callersFunctionload_with_pointer_offset
Loads a fragment from memory with additional logical offset
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/warp/mma_complex_tensor_op_tile_iterator_sm80.h:284
↓ 3 callersFunctionload_with_pointer_offset
Load
fastertransformer/3rdparty/cutlass/include/cutlass/epilogue/warp/tile_iterator_simt.h:227
↓ 3 callersFunctionload_with_pointer_offset
Load
fastertransformer/3rdparty/cutlass/include/cutlass/epilogue/warp/tile_iterator_tensor_op.h:210
↓ 3 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/t5_dataset.py:233
↓ 3 callersMethodmake_dataset
( directory: str, class_to_idx: Dict[str, int], data_per_class_fraction: float,
fastertransformer/3rdparty/Megatron-LM/megatron/data/image_folder.py:178
↓ 3 callersMethodmantissa
Returns the mantissa
fastertransformer/3rdparty/cutlass/include/cutlass/tfloat32.h:174
↓ 3 callersMethodmantissa
Returns the mantissa
fastertransformer/3rdparty/cutlass/include/cutlass/half.h:492
↓ 3 callersMethodmantissa
Returns the mantissa
fastertransformer/3rdparty/cutlass/include/cutlass/bfloat16.h:188
↓ 3 callersFunctionmap_layers_to_multi_gpus
(layers)
algorithm/parallel_utils.py:173
↓ 3 callersMethodmap_state
(self, fn)
fastertransformer/examples/pytorch/decoding/utils/decoding.py:298
↓ 3 callersFunctionnorm
Layer normalizes :obj:`inputs`.
fastertransformer/examples/tensorflow/decoder/utils/decoder.py:28
↓ 3 callersMethodnormalize_string
(self, s)
algorithm/lm_eval/decontamination/janitor.py:207
↓ 3 callersMethodoperator++
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_tensor_op_sm70.h:498
↓ 3 callersMethodoperator++
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_access_iterator_tensor_op_sm80.h:588
↓ 3 callersFunctionparaName
fastertransformer/examples/cpp/xlnet/xlnet_correctness_example.cc:133
↓ 3 callersFunctionparallel_lm_logits
LM logits using word embedding weights.
fastertransformer/3rdparty/Megatron-LM/megatron/model/language_model.py:29
↓ 3 callersFunctionparam2file
(pt_param, layer_id, save_dir, dest_key)
fastertransformer/examples/pytorch/gptj/utils/gptj_ckpt_convert.py:90
↓ 3 callersFunctionparam2file
(pt_param, layer_id, save_dir, dest_key)
fastertransformer/examples/pytorch/gptj/utils/huggingface_gptj_ckpt_convert.py:21
↓ 3 callersFunctionparam_is_not_shared
(param)
fastertransformer/3rdparty/Megatron-LM/megatron/model/module.py:32
↓ 3 callersFunctionparam_is_not_tensor_parallel_duplicate
(param)
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/layers.py:47
↓ 3 callersFunctionprint_abs_mean
(node, tensor, info = "")
fastertransformer/examples/tensorflow/common_utils/common.py:126
↓ 3 callersFunctionprint_split_stats
(name, index)
fastertransformer/3rdparty/Megatron-LM/megatron/data/dataset_utils.py:503
↓ 3 callersFunctionprint_split_stats
(name, index)
fastertransformer/3rdparty/Megatron-LM/megatron/data/gpt_dataset.py:96
↓ 3 callersMethodprocedural_name
The full procedural name indicates architecture, extended name, tile size, and layout.
fastertransformer/3rdparty/cutlass/tools/library/scripts/rank_k_operation.py:136
↓ 3 callersMethodprocedural_name
The full procedural name indicates architecture, extended name, tile size, and layout.
fastertransformer/3rdparty/cutlass/tools/library/scripts/trmm_operation.py:143
↓ 3 callersMethodprocedural_name
The full procedural name indicates architecture, extended name, tile size, and layout.
fastertransformer/3rdparty/cutlass/tools/library/scripts/conv2d_operation.py:122
↓ 3 callersMethodprocedural_name
The full procedural name indicates architecture, extended name, tile size, and layout.
fastertransformer/3rdparty/cutlass/tools/library/scripts/symm_operation.py:142
↓ 3 callersMethodprocedural_name
The full procedural name indicates architecture, extended name, tile size, and layout.
fastertransformer/3rdparty/cutlass/tools/library/scripts/rank_2k_operation.py:138
↓ 3 callersMethodprocedural_name
The full procedural name indicates architecture, extended name, tile size, and layout.
fastertransformer/3rdparty/cutlass/tools/library/scripts/conv3d_operation.py:100
↓ 3 callersMethodprocedural_name
The full procedural name indicates architecture, extended name, tile size, and layout.
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/gemm_operation.py:911
↓ 3 callersFunctionquantize
(mat, act_range)
fastertransformer/examples/pytorch/gpt/utils/huggingface_opt_convert.py:47
↓ 3 callersFunctionrandom_tensor
(shape, dtype, device, mean=0, std=1)
fastertransformer/tests/gemm_dequantize/th_gemm_dequantize.py:4
↓ 3 callersFunctionrandom_tensor
(shape, dtype, device, mean=0, std=1)
fastertransformer/tests/gemm_dequantize/single_test.py:5
↓ 3 callersFunctionrebuild_tokenizer
(args)
fastertransformer/3rdparty/Megatron-LM/megatron/global_vars.py:135
↓ 3 callersFunctionrecv_forward
(tensor_shapes, timers)
fastertransformer/3rdparty/Megatron-LM/megatron/schedules.py:506
↓ 3 callersMethodref
Returns a TensorRef to the operand
fastertransformer/3rdparty/cutlass/examples/42_fused_multi_head_attention/gemm/custom_mma_base.h:114
↓ 3 callersMethodreload_model_params
(self)
fastertransformer/3rdparty/Megatron-LM/megatron/optimizer/optimizer.py:547
↓ 3 callersFunctionreorder_meta
fastertransformer/3rdparty/cutlass/tools/util/include/cutlass/util/host_reorder.h:86
↓ 3 callersMethodreset
Set to the initial state (no tracker).
fastertransformer/3rdparty/Megatron-LM/megatron/mpu/random.py:195
↓ 3 callersFunctionreshape_from_matrix
Reshapes a rank 2 tensor back to its original rank >= 2 tensor.
fastertransformer/examples/tensorflow/bert/tensorflow_bert/my_modeling.py:956
↓ 3 callersFunctionreshape_to_matrix
Reshapes a >= rank 2 tensor to a rank 2 tensor (i.e., a matrix).
fastertransformer/examples/tensorflow/bert/utils/bert.py:299
↓ 3 callersFunctionreshape_to_matrix
Reshapes a >= rank 2 tensor to a rank 2 tensor (i.e., a matrix).
fastertransformer/examples/tensorflow/bert/tensorflow_bert/bert/modeling.py:932
↓ 3 callersFunctionreshape_to_matrix
Reshapes a >= rank 2 tensor to a rank 2 tensor (i.e., a matrix).
fastertransformer/examples/tensorflow/bert/bert-quantization/modeling.py:1025
↓ 3 callersFunctionreverse_pyt_forward
(attn_windows, shift_size, window_size, H, W)
fastertransformer/examples/pytorch/swin/Swin-Transformer-Quantization/SwinTransformer/kernels/window_process/unit_test.py:108
↓ 3 callersMethodrun
(self, problem_size, split_k_mode=cutlass.conv.SplitKMode.Serial, alpha=1.0, beta=0.0)
fastertransformer/3rdparty/cutlass/tools/library/scripts/pycutlass/src/pycutlass/test/conv2d_testbed.py:417
↓ 3 callersMethodrun
Executes one test
fastertransformer/3rdparty/cutlass/test/unit/gemm/device/testbed_rank2k_universal.h:309
↓ 3 callersMethodrun
Runs the test
fastertransformer/3rdparty/cutlass/test/unit/gemm/warp/testbed.h:195
↓ 3 callersMethodrun
Executes one test
fastertransformer/3rdparty/cutlass/test/unit/conv/device/conv3d_testbed.h:212
↓ 3 callersMethodrun
Executes one test
fastertransformer/3rdparty/cutlass/test/unit/conv/device/conv2d_testbed_interleaved.h:190
↓ 3 callersMethodrun_decode
(self, batch_size=4, beam_width=1, max_input_length=4
fastertransformer/tests/unittests/th_op/test_th_decode_op.py:85
↓ 3 callersMethodset
fastertransformer/src/fastertransformer/utils/gpu_buf.h:45
↓ 3 callersMethodset_epoch
(self, epoch)
fastertransformer/3rdparty/Megatron-LM/megatron/data/data_samplers.py:121
↓ 3 callersMethodset_epoch
(self, epoch)
fastertransformer/examples/pytorch/swin/Swin-Transformer-Quantization/SwinTransformer/data/samplers.py:28
↓ 3 callersFunctionset_global_variables
Set args, tokenizer, tensorboard-writer, adlr-autoresume, and timers.
fastertransformer/3rdparty/Megatron-LM/megatron/global_vars.py:89
↓ 3 callersMethodset_hf_plugin_mode
(self, is_plugin)
fastertransformer/examples/pytorch/longformer/model.py:80
↓ 3 callersMethodset_input_tensor
See megatron.model.transformer.set_input_tensor()
fastertransformer/3rdparty/Megatron-LM/megatron/model/t5_model.py:128
↓ 3 callersMethodset_iteration_index
Overrides the internal iteration index
fastertransformer/3rdparty/cutlass/include/cutlass/transform/threadblock/regular_tile_access_iterator_tensor_op.h:680
↓ 3 callersMethodset_iteration_index
Overrides the internal iteration index
fastertransformer/3rdparty/cutlass/include/cutlass/epilogue/threadblock/predicated_tile_iterator.h:920
↓ 3 callersMethodset_kgroup_index
Notify the iterator which k-group it is currently pointing to. This does not advance the iterator. Rather, it overrides its internal tracking with co
fastertransformer/3rdparty/cutlass/include/cutlass/gemm/warp/mma_complex_tensor_op_tile_iterator_sm80.h:570
← previousnext →1,201–1,300 of 14,558, ranked by callers