MCPcopy Create free account

hub / github.com/JonasGeiping/linear_cross_entropy_loss / functions

Functions339 in github.com/JonasGeiping/linear_cross_entropy_loss

↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_dx
( x_ptr, y_ptr, A_t_ptr, lse_global_ptr, x_grad_ptr, locks_V_ptr, stride_x_N,
variants/double_recomp3.py:404
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, entropy_ptr, stride_x_N, stride
variants/final_fusion_VII.py:457
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, stride_x_N, stride_x_H, stride_
variants/final_fusion_parallel_splitN_V.py:401
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, entropy_ptr, stride_x_N, stride
variants/final_fusion_deployed.py:357
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, stride_x_N, stride_x_H, stride_
variants/final_fusion_arbitary_input.py:393
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, stride_x_N, stride_x_H, stride_
variants/final_fusion_parallel_splitN_V_usable.py:394
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( sz_ptr, x_ptr, y_ptr, A_grad_ptr, stride_x_N, stride_x_H, stride_A_H, stride
variants/chunked_highmem.py:413
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( sz_ptr, x_ptr, y_ptr, A_grad_ptr, stride_x_N, stride_x_H, stride_A_H, stride
variants/highmem.py:407
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, stride_x_N, stride_x_H, stride_
variants/final_fusion_usable_variant.py:390
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, stride_x_N, stride_x_H, stride_
variants/final_fusion_parallel_split.py:340
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, stride_x_N, stride_x_H, stride_
variants/final_fusion_parallel.py:331
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dA
( z_nv_ptr, y_ptr, x_ptr, A_grad_ptr, lse_ptr, entropy_ptr, stride_x_N, stride
linear_cross_entropy/ops.py:341
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_VII.py:355
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_parallel_splitN_V.py:310
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_deployed.py:256
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_arbitary_input.py:295
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_parallel_splitN_V_usable.py:296
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( sz_ptr, y_ptr, A_t_ptr, x_grad_ptr, stride_x_N, stride_x_H, stride_A_H, stri
variants/chunked_highmem.py:340
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( sz_ptr, y_ptr, A_t_ptr, x_grad_ptr, stride_x_N, stride_x_H, stride_A_H, stri
variants/highmem.py:338
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_usable_variant.py:292
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_parallel_split.py:252
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
variants/final_fusion_parallel.py:252
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_epilogue_dx
( z_nv_ptr, y_ptr, A_t_ptr, x_grad_ptr, lse_ptr, stride_x_N, stride_x_H, strid
linear_cross_entropy/ops.py:240
↓ 1 callersFunctionreference_torch_checkpoint
(x, y, A, default_chunk_size=512, ignore_index=-1, z_regularization=0.0, logit_scale=1.0)
variants/final_fusion_VII.py:62
↓ 1 callersMethodreset_parameters
(self)
variants/final_fusion_VII.py:1053
↓ 1 callersMethodreset_parameters
(self)
variants/final_fusion_deployed.py:886
↓ 1 callersMethodreset_parameters
(self)
variants/malek_xent.py:220
↓ 1 callersMethodreset_parameters
(self)
linear_cross_entropy/ops.py:926
↓ 1 callersFunctionsimple_bench
(fn, x, At, A, reference_loss, reference_x_grad, reference_A_grad)
bench_fusions.py:421
↓ 1 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/malek_like_with_my_fwd2b.py:33
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/malek_like_with_my_fwdNV2.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/final_fusion.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/malek_like_with_my_fwdNV.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/malek_like_with_my_fwdNV5.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/final_fusion_parallel_splitN_V.py:37
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/malek_like_with_my_fwdNV4.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/malek_like_with_my_fwd.py:36
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/malek_like_with_my_fwdNV_atomics.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/malek_like_with_my_fwdNV3.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/compile_tests_NV5.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/final_fusion_parallel_split.py:35
↓ 1 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=512)
variants/final_fusion_parallel.py:35
↓ 1 callersFunctiontorch_compiled_checkpoint
(x, y, A, default_chunk_size=4096)
bench_fusions.py:56
↓ 1 callersFunctiontorch_mixed_precision
(x, y, A)
bench_fusions.py:42
Method__init__
( self, in_features: int, out_features: int, device=None, dtype=None,
variants/final_fusion_VII.py:1022
Method__init__
( self, in_features: int, out_features: int, device=None, dtype=None,
variants/final_fusion_deployed.py:850
Method__init__
( self, dim: int, n_classes: int, n_loop_iters: int = 1, ignore_index:
variants/malek_xent.py:205
Method__init__
( self, in_features: int, out_features: int, device=None, dtype=None,
linear_cross_entropy/ops.py:890
Function_inner_function
(x_block, y_block, A, num_blocks)
bench_fusions.py:51
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/malek_like_with_my_fwdNV2.py:31
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/final_fusion.py:31
Function_inner_function
(x_block, y_block, A, num_blocks, ignore_index=-1, z_regularization=0.0, logit_scale=1.0)
variants/final_fusion_VII.py:45
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/malek_like_with_my_fwdNV.py:31
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/malek_like_with_my_fwdNV5.py:31
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/final_fusion_parallel_splitN_V.py:33
Function_inner_function
(x_block, y_block, A, num_blocks, ignore_index=5)
variants/final_fusion_arbitary_input.py:34
Function_inner_function
(x_block, y_block, A, num_blocks, ignore_index=5)
variants/final_fusion_parallel_splitN_V_usable.py:34
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/malek_like_with_my_fwdNV4.py:31
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/malek_like_with_my_fwd.py:32
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/malek_like_with_my_fwdNV_atomics.py:31
Function_inner_function
(x_block, y_block, A, num_blocks, ignore_index=5)
variants/final_fusion_usable_variant.py:34
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/malek_like_with_my_fwdNV3.py:31
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/compile_tests_NV5.py:31
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/final_fusion_parallel_split.py:31
Function_inner_function
(x_block, y_block, A, num_blocks)
variants/final_fusion_parallel.py:31
Methodbackward
(ctx, grad_output)
variants/final_fusion_deployed.py:826
Methodbackward
(ctx, grad_output)
variants/malek_xent.py:187
Methodbackward
(ctx, grad_output, void0, void1, void2, void3)
linear_cross_entropy/ops.py:807
Functionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwdNV2.py:26
Functionbaseline_torch
(x, y, A)
variants/final_fusion.py:26
Functionbaseline_torch
(x, y, A, ignore_index=5, z_regularization=0.0, logit_scale=1.0)
variants/final_fusion_VII.py:25
Functionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwdNV.py:26
Functionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwdNV5.py:26
Functionbaseline_torch
(x, y, A)
variants/final_fusion_parallel_splitN_V.py:28
Functionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwdNV4.py:26
Functionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwd.py:27
Functionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwdNV_atomics.py:26
Functionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwdNV3.py:26
Functionbaseline_torch
(x, y, A)
variants/compile_tests_NV5.py:26
Functionbaseline_torch
(x, y, A)
variants/final_fusion_parallel_split.py:26
Functionbaseline_torch
(x, y, A)
variants/final_fusion_parallel.py:26
Functionbenchmark
(H=2048, V=131072, N=(4096 * 4), provider="torch", mode="fwd", acc="fp16")
bench_fusions.py:232
Functioncompiled_helper_version
(x, y, At, lse_global, xgrad, Atgrad)
variants/highmem.py:475
Functioncosim
(x, y)
bench_fusions.py:22
Functioncosim
(x, y)
variants/malek_like_with_my_fwdNV2.py:18
Functioncosim
(x, y)
variants/final_fusion.py:18
Functioncosim
(x, y)
variants/final_fusion_VII.py:17
Functioncosim
(x, y)
variants/double_recomp_thinboy.py:20
Functioncosim
(x, y)
variants/malek_like_with_my_fwdNV.py:18
Functioncosim
(x, y)
variants/malek_like_with_my_fwdNV5.py:18
Functioncosim
(x, y)
variants/lock_city.py:20
Functioncosim
(x, y)
variants/final_fusion_parallel_splitN_V.py:20
Functioncosim
(x, y)
variants/malek_like_with_my_fwd2b.py:16
Functioncosim
(x, y)
variants/final_fusion_arbitary_input.py:18
Functioncosim
(x, y)
variants/final_fusion_parallel_splitN_V_usable.py:18
Functioncosim
(x, y)
variants/malek_like_with_my_fwdNV4.py:18
Functioncosim
(x, y)
variants/malek_like_with_my_fwd.py:19
Functioncosim
(x, y)
variants/chunked_highmem.py:19
Functioncosim
(x, y)
variants/malek_like_with_my_fwd2.py:16
Functioncosim
(x, y)
variants/highmem.py:19
← previousnext →101–200 of 339, ranked by callers