MCPcopy Create free account

hub / github.com/JonasGeiping/linear_cross_entropy_loss / functions

Functions339 in github.com/JonasGeiping/linear_cross_entropy_loss

⨍ Functions 339◇ Types & classes 34↳ Endpoints 3
Functionlinear_xent_fwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H, stride_A
variants/double_recomp2.py:87
Functionlinear_xent_fwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H, stride_A
variants/manyway_recomp.py:85
Functionlinear_xent_fwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H, stride_A
variants/double_recomp2_final_story.py:87
Functionlinear_xent_fwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H, stride_A
variants/double_recomp3.py:92
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwdNV2.py:318
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/final_fusion.py:177
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, m_ptr, logit_norm_ptr,
variants/final_fusion_VII.py:236
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwdNV.py:192
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwdNV5.py:213
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/final_fusion_parallel_splitN_V.py:211
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwd2b.py:83
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, m_ptr, logit_norm_ptr,
variants/final_fusion_deployed.py:139
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwdNV4.py:117
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwd.py:116
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwd2.py:83
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, sumexp_ptr, stride_x_N, stride_
variants/malek_like_with_my_fwdNV_atomics.py:192
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/malek_like_with_my_fwdNV3.py:249
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/compile_tests_NV5.py:213
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/final_fusion_parallel_split.py:154
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, stride_x_N, stride_x_H
variants/final_fusion_parallel.py:154
Functionlinear_xent_fwd_prep_bwd_kernel_matmul_t
( x_ptr, y_ptr, A_t_ptr, z_nv_ptr, losses_ptr, lse_ptr, m_ptr, logit_norm_ptr,
linear_cross_entropy/ops.py:122
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, y_ptr, lse_ptr, stride_z_N, stride_z_V, idx_N_group, N_group: tl.const
variants/malek_like_with_my_fwdNV2.py:462
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, y_ptr, lse_ptr, stride_z_N, stride_z_V, idx_N_group, N_group: tl.const
variants/malek_like_with_my_fwdNV.py:334
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, y_ptr, lse_ptr, stride_z_N, stride_z_V, idx_N_group, N_group: tl.const
variants/malek_like_with_my_fwdNV5.py:355
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, y_ptr, lse_ptr, stride_z_N, stride_z_V, idx_N_group, N_group: tl.const
variants/malek_like_with_my_fwdNV4.py:256
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, y_ptr, lse_ptr, stride_z_N, stride_z_V, idx_N_group, N_group: tl.const
variants/malek_like_with_my_fwd.py:239
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, z_grad_ptr, y_ptr, lse_ptr, stride_z_N, stride_z_V, N_offset, V: t
variants/malek_like_with_my_fwd2.py:193
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, y_ptr, sumexp_ptr, stride_z_N, stride_z_V, idx_N_group, N_group: tl.co
variants/malek_like_with_my_fwdNV_atomics.py:321
Functionlinear_xent_mini_bwd_prologue_kernel
( z_nv_ptr, y_ptr, lse_ptr, stride_z_N, stride_z_V, idx_N_group, N_group: tl.const
variants/compile_tests_NV5.py:355
Functionlogsumexp_reduction_kernel
( lse_local_ptr, lse_global_ptr, lse_sum_ptr, reduction_ptr, z_regularization: tl.constexp
variants/final_fusion_arbitary_input.py:726
Functionlogsumexp_reduction_kernel
( lse_local_ptr, lse_global_ptr, lse_sum_ptr, reduction_ptr, z_regularization: tl.constexp
variants/final_fusion_parallel_splitN_V_usable.py:728
Functionreference_torch
(x, y, A, ignore_index=5, z_regularization=0.0, logit_scale=1.0)
variants/final_fusion_VII.py:34
Functionreference_torch
(x, y, A, ignore_index=5, z_regularization=0.0, logit_scale=1.0)
variants/final_fusion_deployed.py:11
Functionreference_torch
(x, y, A, ignore_index=5, z_regularization=0.0, logit_scale=1.0)
linear_cross_entropy/ops.py:9
Functionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=50)
variants/compile_tests_NV5.py:50
Functiontest_fn
()
bench_fusions.py:253
Functiontorch_checkpoint
(x, y, A, default_chunk_size=512, ignore_index=5)
variants/final_fusion_arbitary_input.py:38
Functiontorch_checkpoint
(x, y, A, default_chunk_size=512, ignore_index=5)
variants/final_fusion_parallel_splitN_V_usable.py:38
Functiontorch_checkpoint
(x, y, A, default_chunk_size=512, ignore_index=5)
variants/final_fusion_usable_variant.py:38
← previous301–339 of 339, ranked by callers