Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/bytedance/ByteTransformer
/ functions
Functions
317 in github.com/bytedance/ByteTransformer
⨍
Functions
317
◇
Types & classes
134
Method
infer
bytetransformer/include/bert_transformer.h:215
Method
initial_offset
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/output_tile_thread_map_for_attr_mask.h:222
Method
initialize
bytetransformer/include/bert_transformer.h:188
Method
initialize
Initializes GEMM state from arguments.
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:149
Method
initialize
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:728
Function
input_compress_layernorm_kernel_launcher
bytetransformer/include/layernorm.h:393
Function
input_compress_layernorm_v2
bytetransformer/include/layernorm.h:198
Function
input_layernorm_kernel_launcher
bytetransformer/include/layernorm.h:376
Function
input_layernorm_v2
bytetransformer/include/layernorm.h:162
Function
is_source_needed
bytetransformer/include/cutlass_attention_operators.h:218
Method
is_source_needed
Returns true if source is needed
bytetransformer/include/cutlass_attention_operators.h:621
Function
layernorm
bytetransformer/include/layernorm.h:25
Function
layernorm_v2
bytetransformer/include/layernorm.h:87
Function
load
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:307
Method
load
bytetransformer/include/cutlass_attention_operators.h:838
Method
load
bytetransformer/include/cutlass_attention_operators.h:990
Method
load
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:105
Method
load
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:261
Method
load
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:479
Method
load
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:677
Method
load
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:867
Method
load
Loads a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_volta_tensor_op.h:123
Method
load
Loads a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_tensor_op.h:133
Method
load_with_byte_offset
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:233
Method
load_with_byte_offset
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:473
Method
load_with_pointer_offset
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:228
Method
load_with_pointer_offset
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:467
Function
main
unit_test/bert_transformer_test.cc:33
Method
make_align
bytetransformer/include/cutlass_attention.h:120
Method
maximum_active_blocks
Computes the maximum number of active blocks per multiprocessor
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:90
Method
operand_B_ref
Returns a TensorRef to the B operand
cutlass_contrib/include/cutlass/contrib/gemm/threadblock/mma_base.h:202
Function
operator()
bytetransformer/include/cutlass_attention_operators.h:289
Function
operator()
Perform a threadblock-scoped matrix multiply-accumulate
cutlass_contrib/include/cutlass/contrib/gemm/threadblock/mma_multistage.h:291
Method
operator()
bytetransformer/include/cutlass_attention_operators.h:83
Method
operator()
bytetransformer/include/cutlass_attention_operators.h:165
Method
operator()
bytetransformer/include/cutlass_attention_operators.h:201
Method
operator()
bytetransformer/include/cutlass_attention_operators.h:636
Method
operator()
bytetransformer/include/cutlass_attention_operators.h:863
Method
operator()
bytetransformer/include/cutlass_attention_operators.h:1010
Method
operator()
bytetransformer/include/gemm_bias_act.h:81
Method
operator()
bytetransformer/include/gemm_bias_act.h:91
Method
operator()
bytetransformer/include/gemm_bias_act.h:100
Method
operator()
bytetransformer/include/gemm_bias_act.h:187
Method
operator()
cutlass_contrib/include/cutlass/contrib/functional.h:36
Method
operator()
cutlass_contrib/include/cutlass/contrib/functional.h:46
Method
operator()
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:74
Method
operator()
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:109
Method
operator()
Runs the kernel using initialized state.
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:228
Method
operator()
Executes one GEMM
cutlass_contrib/include/cutlass/contrib/gemm/kernel/gemm_grouped.h:431
Method
operator()
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/softmax_epilogue.h:107
Function
operator++
bytetransformer/include/cutlass_attention_operators.h:507
Method
operator++
bytetransformer/include/cutlass_attention_operators.h:812
Method
operator++
bytetransformer/include/cutlass_attention_operators.h:978
Method
operator++
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:97
Method
operator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:747
Method
operator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:198
Method
operator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:436
Method
operator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:634
Function
parallel_prefix
bytetransformer/include/remove_padding.h:37
Function
print_vec
unit_test/helper.h:88
Method
push
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/epilogue.h:241
Method
push
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/epilogue.h:356
Method
run
Runs the kernel using initialized state.
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:193
Method
set_iteration_index
Overrides the internal iteration index
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:704
Method
set_iteration_index
Overrides the internal iteration index
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:893
Method
set_iteration_index
Overrides the internal iteration index
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:909
Method
set_k_partition
Functionally required for serial reduction in the epilogue
bytetransformer/include/cutlass_attention_operators.h:627
Method
set_k_partition
Functionally required for serial reduction in the epilogue
bytetransformer/include/gemm_bias_act.h:178
Function
set_kgroup_index
bytetransformer/include/cutlass_attention_operators.h:512
Method
set_kgroup_index
bytetransformer/include/cutlass_attention_operators.h:819
Method
set_kgroup_index
bytetransformer/include/cutlass_attention_operators.h:984
Function
set_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:552
Function
set_mask
< Sets the mask
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:434
Method
set_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:766
Method
set_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:955
Method
set_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:217
Method
set_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:455
Method
set_mask
< Sets the mask
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:673
Method
set_mask
< Sets the mask
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:949
Function
softmax_et_kernelLauncher
bytetransformer/include/softmax.h:398
Function
softmax_kernelLauncher
bytetransformer/include/softmax.h:319
Function
softmax_kernel_warp
bytetransformer/include/softmax.h:26
Function
softmax_kernel_warp_et
bytetransformer/include/softmax.h:108
Function
softmax_kernel_warp_half2
bytetransformer/include/softmax.h:65
Function
softmax_kernel_warp_half2_et
bytetransformer/include/softmax.h:148
Function
softmax_kernel_warp_half2_register
bytetransformer/include/softmax.h:192
Function
softmax_kernel_warp_half2_register_et
bytetransformer/include/softmax.h:245
Function
softmax_reduction_kernel
bytetransformer/include/cutlass_attention_operators.h:378
Function
softmax_reduction_kernel_launcher
bytetransformer/include/cutlass_attention_operators.h:412
Function
store
Stores a fragment to memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:368
Method
store
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:299
Method
store
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:497
Method
store
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:695
Method
store
Stores a fragment to memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:888
Method
store
Stores a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_volta_tensor_op.h:161
Method
store
Stores a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_tensor_op.h:167
Method
store_with_byte_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:273
Method
store_with_byte_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:491
Method
store_with_pointer_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:267
Method
store_with_pointer_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:485
← previous
next →
201–300 of 317, ranked by callers