mindstudio-probe 1.2.1__py3-none-any.whl → 1.3.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (177) hide show
  1. {mindstudio_probe-1.2.1.dist-info → mindstudio_probe-1.3.0.dist-info}/METADATA +3 -3
  2. {mindstudio_probe-1.2.1.dist-info → mindstudio_probe-1.3.0.dist-info}/RECORD +168 -150
  3. msprobe/README.md +27 -22
  4. msprobe/core/common/const.py +129 -60
  5. msprobe/core/common/decorator.py +50 -0
  6. msprobe/core/common/exceptions.py +3 -1
  7. msprobe/core/common/file_utils.py +25 -2
  8. msprobe/core/common/inplace_ops.yaml +1 -0
  9. msprobe/core/common/utils.py +43 -33
  10. msprobe/core/compare/acc_compare.py +43 -74
  11. msprobe/core/compare/check.py +2 -6
  12. msprobe/core/compare/highlight.py +2 -0
  13. msprobe/core/compare/layer_mapping/data_scope_parser.py +1 -1
  14. msprobe/core/compare/layer_mapping/layer_mapping.py +2 -1
  15. msprobe/core/compare/merge_result/merge_result.py +16 -9
  16. msprobe/core/compare/merge_result/utils.py +81 -0
  17. msprobe/core/compare/multiprocessing_compute.py +19 -12
  18. msprobe/core/compare/npy_compare.py +30 -12
  19. msprobe/core/compare/utils.py +30 -10
  20. msprobe/core/data_dump/api_registry.py +176 -0
  21. msprobe/core/data_dump/data_collector.py +58 -13
  22. msprobe/core/data_dump/data_processor/base.py +94 -10
  23. msprobe/core/data_dump/data_processor/factory.py +3 -0
  24. msprobe/core/data_dump/data_processor/mindspore_processor.py +33 -33
  25. msprobe/core/data_dump/data_processor/pytorch_processor.py +99 -18
  26. msprobe/core/data_dump/json_writer.py +61 -40
  27. msprobe/core/grad_probe/constant.py +1 -0
  28. msprobe/core/grad_probe/grad_compare.py +1 -1
  29. msprobe/core/overflow_check/abnormal_scene.py +2 -0
  30. msprobe/docs/01.installation.md +27 -1
  31. msprobe/docs/02.config_introduction.md +27 -23
  32. msprobe/docs/03.config_examples.md +24 -0
  33. msprobe/docs/05.data_dump_PyTorch.md +103 -16
  34. msprobe/docs/06.data_dump_MindSpore.md +76 -32
  35. msprobe/docs/07.accuracy_checker_PyTorch.md +11 -1
  36. msprobe/docs/08.accuracy_checker_online_PyTorch.md +3 -1
  37. msprobe/docs/09.accuracy_checker_MindSpore.md +5 -3
  38. msprobe/docs/10.accuracy_compare_PyTorch.md +59 -33
  39. msprobe/docs/11.accuracy_compare_MindSpore.md +40 -16
  40. msprobe/docs/12.overflow_check_PyTorch.md +3 -1
  41. msprobe/docs/13.overflow_check_MindSpore.md +4 -2
  42. msprobe/docs/14.data_parse_PyTorch.md +1 -7
  43. msprobe/docs/18.online_dispatch.md +1 -1
  44. msprobe/docs/19.monitor.md +332 -273
  45. msprobe/docs/21.visualization_PyTorch.md +42 -13
  46. msprobe/docs/22.visualization_MindSpore.md +43 -13
  47. msprobe/docs/23.generate_operator_PyTorch.md +9 -9
  48. msprobe/docs/27.dump_json_instruction.md +301 -27
  49. msprobe/docs/28.debugger_save_instruction.md +94 -0
  50. msprobe/docs/28.kernel_dump_MindSpore.md +69 -0
  51. msprobe/docs/29.data_dump_MSAdapter.md +229 -0
  52. msprobe/docs/30.overflow_check_MSAdapter.md +31 -0
  53. msprobe/docs/FAQ.md +3 -11
  54. msprobe/docs/img/compare_result.png +0 -0
  55. msprobe/docs/img/merge_result.png +0 -0
  56. msprobe/docs/img/monitor/step_count_per_record.png +0 -0
  57. msprobe/docs/img/visualization/vis_browser_1.png +0 -0
  58. msprobe/docs/img/visualization/vis_match_info.png +0 -0
  59. msprobe/docs/img/visualization/vis_precision_info.png +0 -0
  60. msprobe/docs/img/visualization/vis_search_info.png +0 -0
  61. msprobe/docs/img/visualization/vis_show_info.png +0 -0
  62. msprobe/docs/img/visualization/vis_showcase.png +0 -0
  63. msprobe/docs/img/visualization/vis_unmatch_info.png +0 -0
  64. msprobe/mindspore/__init__.py +4 -2
  65. msprobe/mindspore/api_accuracy_checker/api_accuracy_checker.py +32 -7
  66. msprobe/mindspore/api_accuracy_checker/api_runner.py +70 -22
  67. msprobe/mindspore/api_accuracy_checker/base_compare_algorithm.py +2 -1
  68. msprobe/mindspore/api_accuracy_checker/bench_functions/flash_attention_score.py +602 -0
  69. msprobe/mindspore/api_accuracy_checker/bench_functions/fusion_operator.py +41 -0
  70. msprobe/mindspore/api_accuracy_checker/compute_element.py +47 -1
  71. msprobe/mindspore/api_accuracy_checker/data_manager.py +2 -1
  72. msprobe/mindspore/api_accuracy_checker/multi_api_accuracy_checker.py +2 -1
  73. msprobe/mindspore/api_accuracy_checker/torch_mindtorch_importer.py +130 -0
  74. msprobe/mindspore/api_accuracy_checker/type_mapping.py +24 -1
  75. msprobe/mindspore/api_accuracy_checker/utils.py +6 -1
  76. msprobe/mindspore/common/const.py +61 -0
  77. msprobe/mindspore/common/utils.py +48 -18
  78. msprobe/mindspore/compare/ms_compare.py +27 -19
  79. msprobe/mindspore/compare/ms_graph_compare.py +6 -5
  80. msprobe/mindspore/debugger/debugger_config.py +31 -6
  81. msprobe/mindspore/debugger/precision_debugger.py +45 -14
  82. msprobe/mindspore/dump/dump_tool_factory.py +5 -3
  83. msprobe/mindspore/dump/hook_cell/api_register.py +142 -0
  84. msprobe/mindspore/dump/hook_cell/hook_cell.py +9 -10
  85. msprobe/mindspore/dump/hook_cell/support_wrap_ops.yaml +24 -26
  86. msprobe/mindspore/dump/jit_dump.py +21 -15
  87. msprobe/mindspore/dym_loader/hook_dynamic_loader.cc +22 -56
  88. msprobe/mindspore/dym_loader/hook_dynamic_loader.h +0 -1
  89. msprobe/mindspore/free_benchmark/api_pynative_self_check.py +10 -6
  90. msprobe/mindspore/free_benchmark/perturbation/perturbation_factory.py +4 -2
  91. msprobe/mindspore/free_benchmark/self_check_tool_factory.py +6 -3
  92. msprobe/mindspore/grad_probe/global_context.py +2 -0
  93. msprobe/mindspore/grad_probe/grad_analyzer.py +2 -1
  94. msprobe/mindspore/grad_probe/hook.py +2 -4
  95. msprobe/mindspore/monitor/anomaly_detect.py +404 -0
  96. msprobe/mindspore/monitor/distributed/__init__.py +0 -0
  97. msprobe/mindspore/monitor/distributed/distributed_ops.yaml +15 -0
  98. msprobe/mindspore/monitor/distributed/stack_blacklist.yaml +5 -0
  99. msprobe/mindspore/monitor/distributed/wrap_distributed.py +300 -0
  100. msprobe/mindspore/monitor/features.py +63 -0
  101. msprobe/mindspore/monitor/module_hook.py +873 -0
  102. msprobe/mindspore/monitor/module_spec_verifier.py +94 -0
  103. msprobe/mindspore/monitor/utils.py +309 -0
  104. msprobe/mindspore/ms_config.py +8 -2
  105. msprobe/mindspore/overflow_check/overflow_check_tool_factory.py +5 -3
  106. msprobe/mindspore/service.py +114 -34
  107. msprobe/pytorch/__init__.py +0 -1
  108. msprobe/pytorch/api_accuracy_checker/compare/api_precision_compare.py +3 -6
  109. msprobe/pytorch/api_accuracy_checker/generate_op_script/op_generator.py +12 -7
  110. msprobe/pytorch/api_accuracy_checker/generate_op_script/operator_replication.template +2 -2
  111. msprobe/pytorch/api_accuracy_checker/run_ut/multi_run_ut.py +4 -5
  112. msprobe/pytorch/api_accuracy_checker/run_ut/run_overflow_check.py +5 -5
  113. msprobe/pytorch/api_accuracy_checker/run_ut/run_ut.py +25 -6
  114. msprobe/pytorch/api_accuracy_checker/run_ut/run_ut_utils.py +28 -19
  115. msprobe/pytorch/api_accuracy_checker/tensor_transport_layer/attl.py +3 -1
  116. msprobe/pytorch/bench_functions/apply_adam.py +215 -0
  117. msprobe/pytorch/bench_functions/group_norm_silu.py +27 -0
  118. msprobe/pytorch/{parse.py → bench_functions/mish.py} +6 -4
  119. msprobe/pytorch/bench_functions/moe_gating_top_k_softmax.py +50 -0
  120. msprobe/pytorch/bench_functions/sort_v2.py +21 -0
  121. msprobe/pytorch/common/utils.py +97 -4
  122. msprobe/pytorch/debugger/debugger_config.py +19 -9
  123. msprobe/pytorch/debugger/precision_debugger.py +24 -1
  124. msprobe/pytorch/dump/module_dump/module_dump.py +4 -3
  125. msprobe/pytorch/dump/module_dump/module_processer.py +21 -35
  126. msprobe/pytorch/free_benchmark/common/utils.py +1 -1
  127. msprobe/pytorch/free_benchmark/compare/single_benchmark.py +1 -1
  128. msprobe/pytorch/free_benchmark/perturbed_layers/npu/add_noise.py +3 -3
  129. msprobe/pytorch/free_benchmark/perturbed_layers/npu/bit_noise.py +3 -3
  130. msprobe/pytorch/free_benchmark/perturbed_layers/npu/change_value.py +1 -1
  131. msprobe/pytorch/free_benchmark/perturbed_layers/npu/improve_precision.py +1 -1
  132. msprobe/pytorch/free_benchmark/result_handlers/check_handler.py +1 -1
  133. msprobe/pytorch/function_factory.py +8 -2
  134. msprobe/pytorch/grad_probe/grad_monitor.py +2 -2
  135. msprobe/pytorch/hook_module/api_register.py +131 -0
  136. msprobe/pytorch/hook_module/hook_module.py +19 -14
  137. msprobe/pytorch/hook_module/register_optimizer_hook.py +2 -1
  138. msprobe/pytorch/hook_module/support_wrap_ops.yaml +173 -75
  139. msprobe/pytorch/monitor/anomaly_detect.py +14 -29
  140. msprobe/pytorch/monitor/csv2tb.py +18 -14
  141. msprobe/pytorch/monitor/distributed/wrap_distributed.py +8 -2
  142. msprobe/pytorch/monitor/module_hook.py +238 -193
  143. msprobe/pytorch/monitor/module_metric.py +9 -6
  144. msprobe/pytorch/monitor/optimizer_collect.py +100 -67
  145. msprobe/pytorch/monitor/unittest/test_monitor.py +1 -1
  146. msprobe/pytorch/monitor/utils.py +76 -44
  147. msprobe/pytorch/online_dispatch/compare.py +0 -2
  148. msprobe/pytorch/online_dispatch/dispatch.py +9 -0
  149. msprobe/pytorch/online_dispatch/dump_compare.py +3 -0
  150. msprobe/pytorch/online_dispatch/utils.py +3 -0
  151. msprobe/pytorch/parse_tool/lib/interactive_cli.py +1 -6
  152. msprobe/pytorch/parse_tool/lib/utils.py +2 -1
  153. msprobe/pytorch/pt_config.py +30 -29
  154. msprobe/pytorch/service.py +114 -32
  155. msprobe/visualization/builder/graph_builder.py +75 -10
  156. msprobe/visualization/builder/msprobe_adapter.py +7 -6
  157. msprobe/visualization/compare/graph_comparator.py +42 -38
  158. msprobe/visualization/compare/mode_adapter.py +0 -19
  159. msprobe/visualization/graph/base_node.py +11 -3
  160. msprobe/visualization/graph/distributed_analyzer.py +71 -3
  161. msprobe/visualization/graph/graph.py +0 -11
  162. msprobe/visualization/graph/node_op.py +4 -3
  163. msprobe/visualization/graph_service.py +4 -5
  164. msprobe/visualization/utils.py +12 -35
  165. msprobe/mindspore/dump/hook_cell/api_registry.py +0 -205
  166. msprobe/mindspore/dump/hook_cell/wrap_api.py +0 -212
  167. msprobe/pytorch/hook_module/api_registry.py +0 -166
  168. msprobe/pytorch/hook_module/wrap_distributed.py +0 -75
  169. msprobe/pytorch/hook_module/wrap_functional.py +0 -66
  170. msprobe/pytorch/hook_module/wrap_npu_custom.py +0 -85
  171. msprobe/pytorch/hook_module/wrap_tensor.py +0 -69
  172. msprobe/pytorch/hook_module/wrap_torch.py +0 -84
  173. msprobe/pytorch/hook_module/wrap_vf.py +0 -60
  174. {mindstudio_probe-1.2.1.dist-info → mindstudio_probe-1.3.0.dist-info}/LICENSE +0 -0
  175. {mindstudio_probe-1.2.1.dist-info → mindstudio_probe-1.3.0.dist-info}/WHEEL +0 -0
  176. {mindstudio_probe-1.2.1.dist-info → mindstudio_probe-1.3.0.dist-info}/entry_points.txt +0 -0
  177. {mindstudio_probe-1.2.1.dist-info → mindstudio_probe-1.3.0.dist-info}/top_level.txt +0 -0
@@ -149,9 +149,9 @@ tensor:
149
149
  - __bool__
150
150
  - __div__
151
151
  - __eq__
152
+ - __floordiv__
152
153
  - __ge__
153
154
  - __gt__
154
- - __getitem__
155
155
  - __iadd__
156
156
  - __iand__
157
157
  - __idiv__
@@ -160,23 +160,33 @@ tensor:
160
160
  - __imod__
161
161
  - __imul__
162
162
  - __ior__
163
+ - __ipow__
163
164
  - __irshift__
164
165
  - __isub__
165
166
  - __ixor__
167
+ - __le__
166
168
  - __lshift__
169
+ - __lt__
167
170
  - __matmul__
168
171
  - __mod__
169
172
  - __mul__
173
+ - __ne__
170
174
  - __nonzero__
171
175
  - __or__
176
+ - __pow__
172
177
  - __radd__
178
+ - __rdiv__
179
+ - __rmod__
173
180
  - __rmul__
181
+ - __ror__
182
+ - __rpow__
174
183
  - __rshift__
184
+ - __rsub__
185
+ - __rxor__
175
186
  - __setitem__
176
187
  - __sub__
177
188
  - __truediv__
178
189
  - __xor__
179
- - __pow__
180
190
  - abs
181
191
  - abs_
182
192
  - absolute
@@ -199,12 +209,14 @@ tensor:
199
209
  - addmv_
200
210
  - addr
201
211
  - addr_
212
+ - adjoint
202
213
  - align_as
203
214
  - align_to
204
215
  - all
205
216
  - allclose
206
217
  - amax
207
218
  - amin
219
+ - aminmax
208
220
  - angle
209
221
  - any
210
222
  - arccos
@@ -216,12 +228,15 @@ tensor:
216
228
  - arcsinh
217
229
  - arcsinh_
218
230
  - arctan
231
+ - arctan2
232
+ - arctan2_
219
233
  - arctan_
220
234
  - arctanh
221
235
  - arctanh_
222
236
  - argmax
223
237
  - argmin
224
238
  - argsort
239
+ - argwhere
225
240
  - asin
226
241
  - asin_
227
242
  - asinh
@@ -236,39 +251,51 @@ tensor:
236
251
  - baddbmm_
237
252
  - bernoulli
238
253
  - bernoulli_
254
+ - bfloat16
239
255
  - bincount
240
256
  - bitwise_and
241
257
  - bitwise_and_
258
+ - bitwise_left_shift
259
+ - bitwise_left_shift_
242
260
  - bitwise_not
243
261
  - bitwise_not_
244
262
  - bitwise_or
245
263
  - bitwise_or_
264
+ - bitwise_right_shift
265
+ - bitwise_right_shift_
246
266
  - bitwise_xor
247
267
  - bitwise_xor_
248
268
  - bmm
269
+ - bool
249
270
  - broadcast_to
271
+ - byte
250
272
  - cauchy_
251
273
  - ceil
252
274
  - ceil_
275
+ - cfloat
276
+ - char
253
277
  - cholesky
278
+ - cholesky_inverse
279
+ - cholesky_solve
254
280
  - chunk
255
281
  - clamp
256
- - cholesky_solve
257
- - cholesky_inverse
258
282
  - clamp_
259
283
  - clamp_max
260
284
  - clamp_max_
261
- - clip
262
285
  - clamp_min
263
286
  - clamp_min_
287
+ - clip
264
288
  - clip_
289
+ - conj_physical
265
290
  - copysign
266
291
  - copysign_
292
+ - corrcoef
267
293
  - cos
268
294
  - cos_
269
295
  - cosh
270
296
  - cosh_
271
297
  - count_nonzero
298
+ - cov
272
299
  - cummax
273
300
  - cummin
274
301
  - cumprod
@@ -282,20 +309,23 @@ tensor:
282
309
  - diag_embed
283
310
  - diagflat
284
311
  - diagonal
312
+ - diagonal_scatter
285
313
  - diff
286
- - dist
287
314
  - digamma
288
315
  - digamma_
316
+ - dist
289
317
  - div
290
318
  - div_
291
319
  - divide
292
320
  - divide_
293
321
  - dot
322
+ - double
323
+ - dsplit
294
324
  - eig
295
325
  - eq
296
326
  - eq_
297
- - erf
298
327
  - equal
328
+ - erf
299
329
  - erf_
300
330
  - erfc
301
331
  - erfc_
@@ -304,18 +334,21 @@ tensor:
304
334
  - exp
305
335
  - exp2
306
336
  - exp2_
307
- - expm1
308
337
  - exp_
338
+ - expand
339
+ - expand_as
340
+ - expm1
309
341
  - expm1_
310
342
  - exponential_
311
343
  - fill_
312
- - fix
313
344
  - fill_diagonal_
345
+ - fix
314
346
  - fix_
347
+ - flatten
315
348
  - flip
316
349
  - fliplr
317
- - flatten
318
350
  - flipud
351
+ - float
319
352
  - float_power
320
353
  - float_power_
321
354
  - floor
@@ -328,6 +361,7 @@ tensor:
328
361
  - fmod_
329
362
  - frac
330
363
  - frac_
364
+ - frexp
331
365
  - gather
332
366
  - gcd
333
367
  - gcd_
@@ -338,31 +372,37 @@ tensor:
338
372
  - ger
339
373
  - greater
340
374
  - greater_
341
- - gt
342
- - gt_
343
375
  - greater_equal
344
376
  - greater_equal_
377
+ - gt
378
+ - gt_
379
+ - half
345
380
  - hardshrink
346
381
  - heaviside
347
382
  - heaviside_
348
383
  - histc
384
+ - histogram
385
+ - hsplit
349
386
  - hypot
350
387
  - hypot_
388
+ - i0
389
+ - i0_
351
390
  - igamma
352
391
  - igamma_
353
392
  - igammac
354
393
  - igammac_
355
394
  - index_add
356
395
  - index_add_
357
- - inverse
358
396
  - index_copy
359
397
  - index_copy_
360
398
  - index_fill
361
399
  - index_fill_
362
400
  - index_put
363
401
  - index_put_
364
- - inner
365
402
  - index_select
403
+ - inner
404
+ - int
405
+ - inverse
366
406
  - isclose
367
407
  - isfinite
368
408
  - isinf
@@ -380,7 +420,6 @@ tensor:
380
420
  - le_
381
421
  - lerp
382
422
  - lerp_
383
- - where
384
423
  - less
385
424
  - less_
386
425
  - less_equal
@@ -397,43 +436,47 @@ tensor:
397
436
  - log_
398
437
  - log_normal_
399
438
  - log_softmax
400
- - logcumsumexp
401
- - logdet
402
439
  - logaddexp
403
440
  - logaddexp2
441
+ - logcumsumexp
442
+ - logdet
404
443
  - logical_and
405
444
  - logical_and_
406
445
  - logical_not
407
- - logit
408
446
  - logical_not_
409
447
  - logical_or
410
448
  - logical_or_
411
449
  - logical_xor
412
450
  - logical_xor_
451
+ - logit
413
452
  - logit_
414
453
  - logsumexp
454
+ - long
415
455
  - lstsq
416
456
  - lt
417
457
  - lt_
458
+ - lu
418
459
  - lu_solve
419
460
  - map2_
420
461
  - map_
421
462
  - masked_fill
422
- - matmul
423
463
  - masked_fill_
424
464
  - masked_scatter
425
465
  - masked_scatter_
426
466
  - masked_select
467
+ - matmul
427
468
  - matrix_exp
469
+ - matrix_power
428
470
  - max
429
471
  - maximum
430
472
  - mean
431
- - matrix_power
432
473
  - median
433
474
  - min
434
475
  - minimum
435
476
  - mm
436
477
  - mode
478
+ - moveaxis
479
+ - movedim
437
480
  - msort
438
481
  - mul
439
482
  - mul_
@@ -443,6 +486,11 @@ tensor:
443
486
  - mv
444
487
  - mvlgamma
445
488
  - mvlgamma_
489
+ - nan_to_num
490
+ - nan_to_num_
491
+ - nanmean
492
+ - nanmedian
493
+ - nanquantile
446
494
  - nansum
447
495
  - narrow
448
496
  - narrow_copy
@@ -452,20 +500,29 @@ tensor:
452
500
  - neg_
453
501
  - negative
454
502
  - negative_
503
+ - nextafter
504
+ - nextafter_
455
505
  - nonzero
456
506
  - norm
457
507
  - normal_
458
508
  - not_equal
459
509
  - not_equal_
510
+ - numpy
511
+ - orgqr
512
+ - ormqr
513
+ - outer
460
514
  - permute
461
515
  - pinverse
462
516
  - polygamma
517
+ - polygamma_
463
518
  - pow
464
519
  - pow_
465
- - polygamma_
466
520
  - prelu
467
521
  - prod
468
522
  - put_
523
+ - q_zero_point
524
+ - qr
525
+ - quantile
469
526
  - rad2deg
470
527
  - rad2deg_
471
528
  - ravel
@@ -474,15 +531,16 @@ tensor:
474
531
  - relu
475
532
  - relu_
476
533
  - remainder
477
- - repeat_interleave
478
- - reshape
479
534
  - remainder_
480
535
  - renorm
481
536
  - renorm_
482
537
  - repeat
538
+ - repeat_interleave
539
+ - reshape
483
540
  - reshape_as
484
541
  - resize_
485
542
  - resize_as_
543
+ - resolve_neg
486
544
  - roll
487
545
  - rot90
488
546
  - round
@@ -496,6 +554,7 @@ tensor:
496
554
  - select
497
555
  - sgn
498
556
  - sgn_
557
+ - short
499
558
  - sigmoid
500
559
  - sigmoid_
501
560
  - sign
@@ -507,11 +566,13 @@ tensor:
507
566
  - sinc_
508
567
  - sinh
509
568
  - sinh_
569
+ - slice_scatter
510
570
  - slogdet
511
571
  - smm
512
572
  - softmax
513
573
  - solve
514
574
  - sort
575
+ - split
515
576
  - split_with_sizes
516
577
  - sqrt
517
578
  - sqrt_
@@ -521,21 +582,29 @@ tensor:
521
582
  - squeeze_
522
583
  - sspaddmm
523
584
  - std
585
+ - stft
586
+ - stride
524
587
  - sub
525
588
  - sub_
589
+ - subtract
526
590
  - sum
527
591
  - sum_to_size
528
592
  - svd
593
+ - swapaxes
594
+ - swapdims
595
+ - swapdims_
529
596
  - symeig
530
597
  - t
531
598
  - t_
532
599
  - take
600
+ - take_along_dim
533
601
  - tan
534
602
  - tan_
535
603
  - tanh
536
604
  - tanh_
537
605
  - tensor_split
538
606
  - tile
607
+ - to
539
608
  - topk
540
609
  - transpose
541
610
  - transpose_
@@ -543,8 +612,8 @@ tensor:
543
612
  - tril
544
613
  - tril_
545
614
  - triu
546
- - true_divide
547
615
  - triu_
616
+ - true_divide
548
617
  - true_divide_
549
618
  - trunc
550
619
  - trunc_
@@ -552,37 +621,20 @@ tensor:
552
621
  - unbind
553
622
  - unflatten
554
623
  - unfold
624
+ - unique
625
+ - unique_consecutive
555
626
  - unsafe_chunk
556
- - unsqueeze
557
627
  - unsafe_split
558
628
  - unsafe_split_with_sizes
629
+ - unsqueeze
630
+ - unsqueeze_
559
631
  - var
560
632
  - vdot
561
- - unsqueeze_
562
633
  - view_as
634
+ - vsplit
635
+ - where
563
636
  - xlogy
564
637
  - xlogy_
565
- - split
566
- - stft
567
- - nan_to_num
568
- - dsplit
569
- - orgqr
570
- - bitwise_left_shift_
571
- - arctan2
572
- - histogram
573
- - q_zero_point
574
- - adjoint
575
- - ormqr
576
- - bitwise_right_shift_
577
- - nanquantile
578
- - lu
579
- - quantile
580
- - arctan2_
581
- - qr
582
- - diagonal_scatter
583
- - corrcoef
584
- - vsplit
585
- - aminmax
586
638
 
587
639
  torch:
588
640
  - linalg.norm
@@ -642,13 +694,14 @@ torch:
642
694
  - addmv
643
695
  - addmv_
644
696
  - addr
645
- - amax
646
697
  - affine_grid_generator
647
698
  - align_tensors
648
699
  - all
649
700
  - alpha_dropout
650
- - amin
651
701
  - alpha_dropout_
702
+ - amax
703
+ - amin
704
+ - aminmax
652
705
  - angle
653
706
  - any
654
707
  - arange
@@ -661,12 +714,14 @@ torch:
661
714
  - arcsinh
662
715
  - arcsinh_
663
716
  - arctan
717
+ - arctan2
664
718
  - arctan_
665
719
  - arctanh
666
720
  - arctanh_
667
721
  - argmax
668
722
  - argmin
669
723
  - argsort
724
+ - argwhere
670
725
  - asin
671
726
  - asin_
672
727
  - asinh
@@ -687,13 +742,13 @@ torch:
687
742
  - batch_norm_elemt
688
743
  - batch_norm_gather_stats
689
744
  - batch_norm_gather_stats_with_counts
690
- - bernoulli
691
745
  - batch_norm_stats
692
746
  - batch_norm_update_stats
747
+ - bernoulli
693
748
  - bilinear
749
+ - binary_cross_entropy_with_logits
694
750
  - bincount
695
751
  - binomial
696
- - binary_cross_entropy_with_logits
697
752
  - bitwise_and
698
753
  - bitwise_not
699
754
  - bitwise_or
@@ -739,9 +794,9 @@ torch:
739
794
  - conv_transpose1d
740
795
  - conv_transpose2d
741
796
  - conv_transpose3d
742
- - cos
743
797
  - convolution
744
798
  - copysign
799
+ - cos
745
800
  - cos_
746
801
  - cosh
747
802
  - cosh_
@@ -755,14 +810,16 @@ torch:
755
810
  - cummin
756
811
  - cumprod
757
812
  - cumsum
813
+ - cumulative_trapezoid
758
814
  - deg2rad
759
815
  - deg2rad_
760
816
  - det
761
817
  - diag
762
818
  - diag_embed
763
- - diff
764
819
  - diagflat
765
820
  - diagonal
821
+ - diagonal_scatter
822
+ - diff
766
823
  - digamma
767
824
  - dist
768
825
  - div
@@ -771,12 +828,15 @@ torch:
771
828
  - dropout
772
829
  - dropout_
773
830
  - dsmm
831
+ - dsplit
774
832
  - dstack
775
833
  - eig
776
834
  - einsum
777
835
  - embedding
778
836
  - embedding_bag
779
837
  - embedding_renorm_
838
+ - empty
839
+ - empty_like
780
840
  - eq
781
841
  - equal
782
842
  - erf
@@ -791,12 +851,12 @@ torch:
791
851
  - expm1
792
852
  - expm1_
793
853
  - eye
794
- - feature_dropout
795
854
  - feature_alpha_dropout
796
855
  - feature_alpha_dropout_
856
+ - feature_dropout
797
857
  - feature_dropout_
798
- - fix
799
858
  - fill_
859
+ - fix
800
860
  - fix_
801
861
  - flatten
802
862
  - flip
@@ -811,8 +871,9 @@ torch:
811
871
  - fmod
812
872
  - frac
813
873
  - frac_
814
- - full
874
+ - frexp
815
875
  - frobenius_norm
876
+ - full
816
877
  - full_like
817
878
  - gather
818
879
  - gcd
@@ -824,8 +885,8 @@ torch:
824
885
  - greater_equal
825
886
  - grid_sampler
826
887
  - grid_sampler_2d
827
- - group_norm
828
888
  - grid_sampler_3d
889
+ - group_norm
829
890
  - gru
830
891
  - gru_cell
831
892
  - gt
@@ -835,23 +896,29 @@ torch:
835
896
  - heaviside
836
897
  - hinge_embedding_loss
837
898
  - histc
899
+ - histogram
900
+ - histogramdd
838
901
  - hsmm
902
+ - hsplit
839
903
  - hspmm
840
904
  - hstack
841
905
  - hypot
906
+ - i0
907
+ - i0_
842
908
  - igamma
843
909
  - igammac
844
910
  - index_add
845
911
  - index_copy
846
- - inner
847
912
  - index_fill
848
913
  - index_put
849
914
  - index_put_
850
915
  - index_select
916
+ - inner
851
917
  - instance_norm
852
918
  - inverse
853
919
  - isclose
854
920
  - isfinite
921
+ - isin
855
922
  - isinf
856
923
  - isnan
857
924
  - isneginf
@@ -879,8 +946,8 @@ torch:
879
946
  - log1p_
880
947
  - log2
881
948
  - log2_
882
- - log_softmax
883
949
  - log_
950
+ - log_softmax
884
951
  - logaddexp
885
952
  - logaddexp2
886
953
  - logcumsumexp
@@ -899,18 +966,18 @@ torch:
899
966
  - lt
900
967
  - lu_solve
901
968
  - lu_unpack
902
- - masked_fill
903
969
  - margin_ranking_loss
970
+ - masked_fill
904
971
  - masked_scatter
905
972
  - masked_select
906
- - matrix_exp
907
973
  - matmul
974
+ - matrix_exp
908
975
  - matrix_power
909
976
  - matrix_rank
910
977
  - max
911
978
  - max_pool1d
912
- - max_pool2d
913
979
  - max_pool1d_with_indices
980
+ - max_pool2d
914
981
  - max_pool3d
915
982
  - maximum
916
983
  - mean
@@ -929,18 +996,20 @@ torch:
929
996
  - mvlgamma
930
997
  - nan_to_num
931
998
  - nan_to_num_
999
+ - nanmean
932
1000
  - nanmedian
1001
+ - nanquantile
933
1002
  - nansum
934
1003
  - narrow
1004
+ - narrow_copy
935
1005
  - native_batch_norm
936
1006
  - native_group_norm
937
- - narrow_copy
938
1007
  - native_layer_norm
939
1008
  - native_norm
940
1009
  - ne
941
1010
  - neg
942
- - negative
943
1011
  - neg_
1012
+ - negative
944
1013
  - negative_
945
1014
  - nextafter
946
1015
  - nonzero
@@ -972,30 +1041,31 @@ torch:
972
1041
  - ravel
973
1042
  - real
974
1043
  - reciprocal
975
- - relu
976
1044
  - reciprocal_
1045
+ - relu
977
1046
  - relu_
978
1047
  - remainder
979
1048
  - renorm
980
1049
  - repeat_interleave
981
1050
  - reshape
982
1051
  - resize_as_
1052
+ - resolve_neg
983
1053
  - roll
984
1054
  - rot90
985
1055
  - round
986
1056
  - round_
1057
+ - row_stack
987
1058
  - rrelu
988
1059
  - rrelu_
989
1060
  - rsqrt
990
- - row_stack
991
1061
  - rsqrt_
992
1062
  - rsub
993
1063
  - saddmm
994
1064
  - scalar_tensor
995
1065
  - scatter
996
- - select
997
1066
  - scatter_add
998
1067
  - searchsorted
1068
+ - select
999
1069
  - selu
1000
1070
  - selu_
1001
1071
  - sgn
@@ -1015,12 +1085,12 @@ torch:
1015
1085
  - solve
1016
1086
  - sort
1017
1087
  - sparse_coo_tensor
1018
- - square
1019
1088
  - split
1020
1089
  - split_with_sizes
1021
1090
  - spmm
1022
1091
  - sqrt
1023
1092
  - sqrt_
1093
+ - square
1024
1094
  - square_
1025
1095
  - squeeze
1026
1096
  - sspaddmm
@@ -1042,8 +1112,8 @@ torch:
1042
1112
  - tan_
1043
1113
  - tanh
1044
1114
  - tanh_
1045
- - tensordot
1046
1115
  - tensor_split
1116
+ - tensordot
1047
1117
  - threshold
1048
1118
  - threshold_
1049
1119
  - tile
@@ -1059,19 +1129,21 @@ torch:
1059
1129
  - true_divide
1060
1130
  - trunc
1061
1131
  - trunc_
1062
- - unique_consecutive
1063
- - xlogy
1064
1132
  - unbind
1133
+ - unflatten
1134
+ - unique_consecutive
1065
1135
  - unsafe_chunk
1066
1136
  - unsafe_split
1067
- - vander
1068
- - var
1069
- - vdot
1070
1137
  - unsafe_split_with_sizes
1071
1138
  - unsqueeze
1139
+ - vander
1140
+ - var
1072
1141
  - var_mean
1142
+ - vdot
1143
+ - vsplit
1073
1144
  - vstack
1074
1145
  - where
1146
+ - xlogy
1075
1147
  - xlogy_
1076
1148
 
1077
1149
  _VF:
@@ -1165,6 +1237,27 @@ torch_npu:
1165
1237
  - npu_moe_finalize_routing
1166
1238
  - npu_moe_gating_top_k_softmax
1167
1239
  - npu_trans_quant_param
1240
+ - npu_gelu
1241
+ - npu_ffn
1242
+ - npu_quant_matmul
1243
+ - npu_format_cast_
1244
+ - npu_dynamic_quant
1245
+ - npu_moe_compute_expert_tokens
1246
+ - npu_weight_quant_batchmatmul
1247
+ - npu_dynamic_quant_asymmetric
1248
+ - npu_grouped_matmul
1249
+ - npu_quant_scatter_
1250
+ - npu_group_quant
1251
+ - npu_fused_infer_attention_score
1252
+ - npu_quantize
1253
+ - npu_fast_gelu
1254
+ - npu_weight_quant_batchmatmul
1255
+ - scatter_update
1256
+ - scatter_update_
1257
+ - npu_moe_init_routing
1258
+ - npu_scatter_nd_update_
1259
+ - npu_scatter_nd_update
1260
+ - npu_prefetch
1168
1261
 
1169
1262
  aten:
1170
1263
  - signbit
@@ -1911,4 +2004,9 @@ distributed:
1911
2004
  - all_to_all_single
1912
2005
  - all_to_all
1913
2006
  - all_gather_into_tensor
1914
- - reduce_scatter_tensor
2007
+ - reduce_scatter_tensor
2008
+ - batch_isend_irecv
2009
+
2010
+ npu_distributed:
2011
+ - isend
2012
+ - irecv