Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwdNV2.py:504
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # ignores all negative integers .
variants/double_recomp_thinboy.py:378
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwdNV.py:376
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwdNV5.py:397
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/final_fusion_parallel_splitN_V.py:675
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwd2b.py:373
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100,
z_regularization=0.0,
variants/final_fusion_arbitary_input.py:762
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100,
z_regularization=0.0,
variants/final_fusion_parallel_splitN_V_usable.py:765
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwdNV4.py:298
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwd.py:281
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # ignores all negative integers .
variants/chunked_highmem.py:487
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwd2.py:339
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwdNV_atomics.py:363
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100,
z_regularization=0.0,
variants/final_fusion_usable_variant.py:759
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/malek_like_with_my_fwdNV3.py:619
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/compile_tests_NV5.py:397
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/final_fusion_parallel_split.py:588
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # ignores all negative integers .
variants/double_recomp2_final_story.py:398
Methodforward(
ctx,
x,
y,
At,
ignore_index=-100, # code ignores all negative integ
variants/final_fusion_parallel.py:600
Functionlinear_cross_entropy(
x,
y,
At,
ignore_index=-100,
z_regularization: float = 0.0,
logit_scale: float = 1.0
variants/final_fusion_deployed.py:833
Functionlinear_cross_entropy(
x,
y,
At,
ignore_index=-100,
z_regularization: float = 0.0,
logit_scale: float = 1.0
linear_cross_entropy/ops.py:814
Functionlinear_xent_bwd_dispatcher(
logits_ptr,
y_ptr,
x_ptr,
A_t_ptr,
x_grad,
At_grad,
lse_global,
logit_entrop
variants/final_fusion_VII.py:672
Functionlinear_xent_bwd_dispatcher(
logits_ptr,
y_ptr,
x_ptr,
A_t_ptr,
x_grad,
At_grad,
lse_global,
stride_x_N,
variants/final_fusion_parallel_splitN_V.py:591
Functionlinear_xent_bwd_dispatcher(
logits_ptr,
y_ptr,
x_ptr,
A_t_ptr,
x_grad,
At_grad,
lse_global,
logit_entrop
variants/final_fusion_deployed.py:572
Functionlinear_xent_bwd_dispatcher(
logits_ptr,
y_ptr,
x_ptr,
A_t_ptr,
x_grad,
At_grad,
lse_global,
stride_x_N,
variants/final_fusion_arbitary_input.py:597
Functionlinear_xent_bwd_dispatcher(
logits_ptr,
y_ptr,
x_ptr,
A_t_ptr,
x_grad,
At_grad,
lse_global,
stride_x_N,
variants/final_fusion_parallel_splitN_V_usable.py:598
Functionlinear_xent_bwd_dispatcher(
logits_ptr,
y_ptr,
x_ptr,
A_t_ptr,
x_grad,
At_grad,
lse_global,
stride_x_N,
variants/final_fusion_usable_variant.py:594
Functionlinear_xent_bwd_dispatcher(
logits,
y,
x,
At,
x_grad,
At_grad,
lse_global,
stride_x_N,
stride_x_H,
variants/final_fusion_parallel_split.py:510
Functionlinear_xent_bwd_dispatcher(
logits,
y,
x,
At,
x_grad,
At_grad,
lse_global,
stride_x_N,
stride_x_H,
variants/final_fusion_parallel.py:524
Functionlinear_xent_bwd_dispatcher(
logits_ptr,
y_ptr,
x_ptr,
A_t_ptr,
x_grad,
At_grad,
lse_global,
logit_entrop
linear_cross_entropy/ops.py:557
Functionlinear_xent_bwd_kernel_matmul_t(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
A_grad_ptr,
x_grad_ptr,
locks_N_ptr,
variants/lock_city.py:184
Functionlinear_xent_bwd_kernel_matmul_t_dA(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
A_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp_thinboy.py:183
Functionlinear_xent_bwd_kernel_matmul_t_dA(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
A_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp1.py:185
Functionlinear_xent_bwd_kernel_matmul_t_dA(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
A_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp2.py:186
Functionlinear_xent_bwd_kernel_matmul_t_dA(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
A_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/manyway_recomp.py:180
Functionlinear_xent_bwd_kernel_matmul_t_dA(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
A_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp2_final_story.py:186
Functionlinear_xent_bwd_kernel_matmul_t_dA_dx(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
x_grad_ptr,
A_grad_ptr,
locks_N_ptr,
variants/double_recomp3.py:192
Functionlinear_xent_bwd_kernel_matmul_t_dx(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
x_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp_thinboy.py:290
Functionlinear_xent_bwd_kernel_matmul_t_dx(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
x_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp1.py:306
Functionlinear_xent_bwd_kernel_matmul_t_dx(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
x_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp2.py:295
Functionlinear_xent_bwd_kernel_matmul_t_dx(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
x_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/manyway_recomp.py:300
Functionlinear_xent_bwd_kernel_matmul_t_dx(
x_ptr,
y_ptr,
A_t_ptr,
lse_global_ptr,
x_grad_ptr,
stride_x_N,
stride_x_H,
s
variants/double_recomp2_final_story.py:298
Functionlinear_xent_bwd_kernel_matmul_t_epilogue(
sz_ptr,
x_ptr,
y_ptr,
A_t_ptr,
x_grad_ptr,
A_grad_ptr,
stride_x_N,
stride_x_
variants/chunked_highmem.py:265
Functionlinear_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.py:492
Functionlinear_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/malek_like_with_my_fwd2b.py:294
Functionlinear_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/malek_like_with_my_fwdNV3.py:540
Functionlinear_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.py:334
Functionlinear_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/malek_like_with_my_fwd2b.py:197
Functionlinear_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/malek_like_with_my_fwd2.py:259
Functionlinear_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/malek_like_with_my_fwdNV3.py:406
Functionlinear_xent_bwd_kernel_matmul_t_prologue(
sz_ptr,
x_ptr,
A_t_ptr,
lse_global_ptr,
stride_x_N,
stride_x_H,
stride_A_H,
variants/chunked_highmem.py:180
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_recomp_thinboy.py:87
Functionlinear_xent_fwd_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_arbitary_input.py:192
Functionlinear_xent_fwd_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_usable.py:192
Functionlinear_xent_fwd_kernel_matmul_t(
x_ptr,
y_ptr,
A_t_ptr,
loss_ptr,
lse_ptr,
stride_x_N,
stride_x_H,
stride_A_H
variants/chunked_highmem.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_recomp1.py:87
Functionlinear_xent_fwd_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_usable_variant.py:192