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
↓ 10 callersMethodbackward
(ctx, grad_output)
variants/highmem.py:582
↓ 5 callersFunctionlinear_cross_entropy
( x, y, At, ignore_index=-100, z_regularization: float = 0.0, logit_scale: float = 1.0
variants/final_fusion_VII.py:1008
↓ 4 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=5)
variants/final_fusion_VII.py:93
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwdNV2.py:605
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/final_fusion.py:690
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/final_fusion_VII.py:1001
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/double_recomp_thinboy.py:413
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwdNV.py:473
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwdNV5.py:494
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/lock_city.py:349
↓ 3 callersMethodbackward
(ctx, grad_output, void, void2)
variants/final_fusion_parallel_splitN_V.py:777
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwd2b.py:480
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/final_fusion_arbitary_input.py:885
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/final_fusion_parallel_splitN_V_usable.py:888
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwdNV4.py:396
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwd.py:368
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/chunked_highmem.py:521
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwd2.py:424
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwdNV_atomics.py:454
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/double_recomp1.py:438
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/final_fusion_usable_variant.py:884
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/double_recomp2.py:441
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/malek_like_with_my_fwdNV3.py:729
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/compile_tests_NV5.py:494
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/final_fusion_parallel_split.py:685
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/manyway_recomp.py:434
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/double_recomp2_final_story.py:433
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/final_fusion_parallel.py:695
↓ 3 callersMethodbackward
(ctx, grad_output)
variants/double_recomp3.py:565
↓ 3 callersFunctionprovider_lookup
(provider, x, y, At, A)
bench_fusions.py:148
↓ 2 callersFunctionbaseline_torch
(x, y, A)
bench_fusions.py:31
↓ 2 callersFunctionbenchmark_with_memory_reporting
(func, quantiles, *args, **kwargs)
bench_fusions.py:304
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=50)
variants/malek_like_with_my_fwdNV2.py:50
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=5)
variants/final_fusion.py:52
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/double_recomp_thinboy.py:37
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=50)
variants/malek_like_with_my_fwdNV.py:50
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=50)
variants/malek_like_with_my_fwdNV5.py:50
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/lock_city.py:37
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=5)
variants/final_fusion_parallel_splitN_V.py:54
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=25)
variants/final_fusion_arbitary_input.py:61
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=25)
variants/final_fusion_parallel_splitN_V_usable.py:61
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/malek_like_with_my_fwdNV4.py:52
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/malek_like_with_my_fwd.py:51
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/chunked_highmem.py:36
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/malek_like_with_my_fwd2.py:33
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/highmem.py:36
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=50)
variants/malek_like_with_my_fwdNV_atomics.py:50
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/double_recomp1.py:37
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=25)
variants/final_fusion_usable_variant.py:61
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/double_recomp2.py:37
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=5)
variants/malek_like_with_my_fwdNV3.py:52
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=1)
variants/final_fusion_parallel_split.py:52
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/manyway_recomp.py:37
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/double_recomp2_final_story.py:37
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad, num_trials=1)
variants/final_fusion_parallel.py:52
↓ 2 callersFunctionsimple_bench
(fn, reference_loss, reference_x_grad, reference_A_grad)
variants/double_recomp3.py:42
↓ 2 callersFunctiontorch_checkpoint
(x, y, A, default_chunk_size=4096)
bench_fusions.py:71
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/double_recomp_thinboy.py:29
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/lock_city.py:29
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwd2b.py:25
↓ 1 callersFunctionbaseline_torch
(x, y, A, ignore_index=5, z_regularization=0.0)
variants/final_fusion_arbitary_input.py:26
↓ 1 callersFunctionbaseline_torch
(x, y, A, ignore_index=5, z_regularization=0.0)
variants/final_fusion_parallel_splitN_V_usable.py:26
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/chunked_highmem.py:28
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/malek_like_with_my_fwd2.py:25
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/highmem.py:28
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/double_recomp1.py:29
↓ 1 callersFunctionbaseline_torch
(x, y, A, ignore_index=5, z_regularization=0.0)
variants/final_fusion_usable_variant.py:26
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/double_recomp2.py:29
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/manyway_recomp.py:29
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/double_recomp2_final_story.py:29
↓ 1 callersFunctionbaseline_torch
(x, y, A)
variants/double_recomp3.py:27
↓ 1 callersFunctioncompiled_baseline
(x, y, A)
bench_fusions.py:37
↓ 1 callersFunctioncompiled_torch
(x, y, A)
variants/double_recomp3.py:33
↓ 1 callersFunctionget_reference_vals
(x, y, A)
bench_fusions.py:400
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size: int = 4096)
variants/malek_like_with_my_fwdNV2.py:611
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size: int = 4096)
variants/final_fusion.py:697
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/double_recomp_thinboy.py:459
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size=4096)
variants/malek_like_with_my_fwdNV.py:479
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size=4096)
variants/malek_like_with_my_fwdNV5.py:500
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/lock_city.py:389
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size: int = 4096)
variants/final_fusion_parallel_splitN_V.py:784
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/malek_like_with_my_fwd2b.py:486
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, z_regularization=0.0, N_chunk_size: int = 4096)
variants/final_fusion_arbitary_input.py:892
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, z_regularization=0.0, N_chunk_size: int = 4096)
variants/final_fusion_parallel_splitN_V_usable.py:895
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size: int = 4096)
variants/malek_like_with_my_fwdNV4.py:402
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/malek_like_with_my_fwd.py:374
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/chunked_highmem.py:593
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/malek_like_with_my_fwd2.py:430
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/highmem.py:636
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size=4096)
variants/malek_like_with_my_fwdNV_atomics.py:460
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/double_recomp1.py:484
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, z_regularization=0.0, N_chunk_size: int = 4096)
variants/final_fusion_usable_variant.py:891
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/double_recomp2.py:487
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size: int = 4096)
variants/malek_like_with_my_fwdNV3.py:735
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size: int = 4096)
variants/final_fusion_parallel_split.py:692
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/manyway_recomp.py:480
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/double_recomp2_final_story.py:482
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At, ignore_index=-100, N_chunk_size: int = 2048)
variants/final_fusion_parallel.py:702
↓ 1 callersFunctionlinear_cross_entropy
(x, y, At)
variants/double_recomp3.py:633
↓ 1 callersFunctionlinear_xent_bwd_kernel_matmul_t_dA
( x_ptr, y_ptr, A_t_ptr, lse_global_ptr, A_grad_ptr, locks_N_ptr, stride_x_N,
variants/double_recomp3.py:284
next →1–100 of 339, ranked by callers