[Feature] DETERMINISTIC interaction mode

[vmoens]

cc @matteobettini

[github-actions]

$\color{#D29922}\textsf{\Large⚠\kern{0.2cm}\normalsize Warning}$ Result of CPU Benchmark Tests

Total Benchmarks: 144. Improved: $\large\color{#35bf28}12$. Worsened: $\large\color{#d91a1a}7$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	39.6660μs	16.9849μs	58.8759 KOps/s	58.6117 KOps/s	$\color{#35bf28}+0.45\%$
test_plain_set_stack_nested	37.8420μs	17.3269μs	57.7138 KOps/s	58.0461 KOps/s	$\color{#d91a1a}-0.57\%$
test_plain_set_nested_inplace	44.0250μs	19.3008μs	51.8114 KOps/s	51.2039 KOps/s	$\color{#35bf28}+1.19\%$
test_plain_set_stack_nested_inplace	51.0680μs	19.1445μs	52.2344 KOps/s	51.5927 KOps/s	$\color{#35bf28}+1.24\%$
test_items	19.7280μs	2.4902μs	401.5761 KOps/s	380.1613 KOps/s	$\textbf{\color{#35bf28}+5.63\%}$
test_items_nested	0.8149ms	0.2673ms	3.7416 KOps/s	3.7052 KOps/s	$\color{#35bf28}+0.98\%$
test_items_nested_locked	0.4837ms	0.2738ms	3.6530 KOps/s	3.7240 KOps/s	$\color{#d91a1a}-1.91\%$
test_items_nested_leaf	0.1465ms	77.3381μs	12.9302 KOps/s	12.8248 KOps/s	$\color{#35bf28}+0.82\%$
test_items_stack_nested	1.3021ms	0.2763ms	3.6194 KOps/s	3.7266 KOps/s	$\color{#d91a1a}-2.88\%$
test_items_stack_nested_leaf	0.1557ms	79.1600μs	12.6326 KOps/s	12.2386 KOps/s	$\color{#35bf28}+3.22\%$
test_items_stack_nested_locked	0.4742ms	0.2699ms	3.7052 KOps/s	3.7647 KOps/s	$\color{#d91a1a}-1.58\%$
test_keys	17.5340μs	3.8888μs	257.1491 KOps/s	242.1355 KOps/s	$\textbf{\color{#35bf28}+6.20\%}$
test_keys_nested	0.2291ms	0.1361ms	7.3467 KOps/s	7.2717 KOps/s	$\color{#35bf28}+1.03\%$
test_keys_nested_locked	0.7908ms	0.1402ms	7.1303 KOps/s	7.0092 KOps/s	$\color{#35bf28}+1.73\%$
test_keys_nested_leaf	0.9050ms	0.1159ms	8.6313 KOps/s	8.5300 KOps/s	$\color{#35bf28}+1.19\%$
test_keys_stack_nested	0.2716ms	0.1369ms	7.3061 KOps/s	7.2695 KOps/s	$\color{#35bf28}+0.50\%$
test_keys_stack_nested_leaf	0.2041ms	0.1162ms	8.6037 KOps/s	8.5273 KOps/s	$\color{#35bf28}+0.90\%$
test_keys_stack_nested_locked	0.2735ms	0.1418ms	7.0521 KOps/s	7.1248 KOps/s	$\color{#d91a1a}-1.02\%$
test_values	40.0890μs	1.1901μs	840.2350 KOps/s	860.9991 KOps/s	$\color{#d91a1a}-2.41\%$
test_values_nested	89.7710μs	51.2911μs	19.4966 KOps/s	19.1853 KOps/s	$\color{#35bf28}+1.62\%$
test_values_nested_locked	0.1034ms	51.5501μs	19.3986 KOps/s	19.4747 KOps/s	$\color{#d91a1a}-0.39\%$
test_values_nested_leaf	0.1128ms	46.8888μs	21.3270 KOps/s	21.3320 KOps/s	$\color{#d91a1a}-0.02\%$
test_values_stack_nested	0.1354ms	52.5087μs	19.0445 KOps/s	18.8106 KOps/s	$\color{#35bf28}+1.24\%$
test_values_stack_nested_leaf	97.8570μs	46.4477μs	21.5296 KOps/s	21.5863 KOps/s	$\color{#d91a1a}-0.26\%$
test_values_stack_nested_locked	0.1003ms	52.0128μs	19.2260 KOps/s	19.3270 KOps/s	$\color{#d91a1a}-0.52\%$
test_membership	0.1367ms	1.3484μs	741.6208 KOps/s	736.5655 KOps/s	$\color{#35bf28}+0.69\%$
test_membership_nested	94.3290μs	3.4092μs	293.3246 KOps/s	254.0131 KOps/s	$\textbf{\color{#35bf28}+15.48\%}$
test_membership_nested_leaf	22.8130μs	3.4702μs	288.1661 KOps/s	286.2878 KOps/s	$\color{#35bf28}+0.66\%$
test_membership_stacked_nested	20.4890μs	3.3991μs	294.1969 KOps/s	287.8762 KOps/s	$\color{#35bf28}+2.20\%$
test_membership_stacked_nested_leaf	25.1780μs	3.4267μs	291.8277 KOps/s	277.1602 KOps/s	$\textbf{\color{#35bf28}+5.29\%}$
test_membership_nested_last	29.3860μs	4.1770μs	239.4065 KOps/s	222.4977 KOps/s	$\textbf{\color{#35bf28}+7.60\%}$
test_membership_nested_leaf_last	35.4180μs	4.2861μs	233.3135 KOps/s	233.4525 KOps/s	$\color{#d91a1a}-0.06\%$
test_membership_stacked_nested_last	19.7570μs	4.2207μs	236.9279 KOps/s	232.3336 KOps/s	$\color{#35bf28}+1.98\%$
test_membership_stacked_nested_leaf_last	53.0110μs	4.1817μs	239.1398 KOps/s	231.0433 KOps/s	$\color{#35bf28}+3.50\%$
test_nested_getleaf	32.9520μs	10.6175μs	94.1845 KOps/s	92.4920 KOps/s	$\color{#35bf28}+1.83\%$
test_nested_get	31.6710μs	10.0201μs	99.7989 KOps/s	97.6169 KOps/s	$\color{#35bf28}+2.24\%$
test_stacked_getleaf	34.1650μs	10.6255μs	94.1129 KOps/s	92.5100 KOps/s	$\color{#35bf28}+1.73\%$
test_stacked_get	37.9830μs	10.0271μs	99.7298 KOps/s	98.7417 KOps/s	$\color{#35bf28}+1.00\%$
test_nested_getitemleaf	42.8110μs	11.2802μs	88.6511 KOps/s	88.3826 KOps/s	$\color{#35bf28}+0.30\%$
test_nested_getitem	31.2200μs	10.2077μs	97.9655 KOps/s	95.9955 KOps/s	$\color{#35bf28}+2.05\%$
test_stacked_getitemleaf	29.8970μs	11.0760μs	90.2855 KOps/s	89.2160 KOps/s	$\color{#35bf28}+1.20\%$
test_stacked_getitem	26.8810μs	10.2213μs	97.8345 KOps/s	95.7262 KOps/s	$\color{#35bf28}+2.20\%$
test_lock_nested	49.2904ms	0.3907ms	2.5598 KOps/s	2.9087 KOps/s	$\textbf{\color{#d91a1a}-12.00\%}$
test_lock_stack_nested	0.5622ms	0.3110ms	3.2159 KOps/s	3.1676 KOps/s	$\color{#35bf28}+1.53\%$
test_unlock_nested	1.3483ms	0.3498ms	2.8588 KOps/s	2.7286 KOps/s	$\color{#35bf28}+4.77\%$
test_unlock_stack_nested	0.4765ms	0.3192ms	3.1331 KOps/s	3.1463 KOps/s	$\color{#d91a1a}-0.42\%$
test_flatten_speed	0.1777ms	94.4845μs	10.5837 KOps/s	10.1931 KOps/s	$\color{#35bf28}+3.83\%$
test_unflatten_speed	0.7178ms	0.4103ms	2.4372 KOps/s	2.3979 KOps/s	$\color{#35bf28}+1.64\%$
test_common_ops	4.5958ms	0.7279ms	1.3738 KOps/s	1.3742 KOps/s	$\color{#d91a1a}-0.03\%$
test_creation	16.5910μs	1.8956μs	527.5334 KOps/s	516.0628 KOps/s	$\color{#35bf28}+2.22\%$
test_creation_empty	40.1170μs	10.5029μs	95.2118 KOps/s	98.5408 KOps/s	$\color{#d91a1a}-3.38\%$
test_creation_nested_1	41.3090μs	13.2863μs	75.2653 KOps/s	76.7015 KOps/s	$\color{#d91a1a}-1.87\%$
test_creation_nested_2	39.2050μs	16.5610μs	60.3828 KOps/s	61.5797 KOps/s	$\color{#d91a1a}-1.94\%$
test_clone	72.2580μs	13.5703μs	73.6903 KOps/s	72.4733 KOps/s	$\color{#35bf28}+1.68\%$
test_getitem[int]	36.5600μs	11.5532μs	86.5564 KOps/s	85.7385 KOps/s	$\color{#35bf28}+0.95\%$
test_getitem[slice_int]	53.5720μs	23.2609μs	42.9905 KOps/s	43.0429 KOps/s	$\color{#d91a1a}-0.12\%$
test_getitem[range]	80.1030μs	60.4592μs	16.5401 KOps/s	17.0975 KOps/s	$\color{#d91a1a}-3.26\%$
test_getitem[tuple]	48.9630μs	19.2118μs	52.0512 KOps/s	49.0869 KOps/s	$\textbf{\color{#35bf28}+6.04\%}$
test_getitem[list]	0.1014ms	42.3983μs	23.5859 KOps/s	23.1024 KOps/s	$\color{#35bf28}+2.09\%$
test_setitem_dim[int]	62.4990μs	36.1535μs	27.6599 KOps/s	27.9936 KOps/s	$\color{#d91a1a}-1.19\%$
test_setitem_dim[slice_int]	98.4080μs	62.3280μs	16.0441 KOps/s	15.8558 KOps/s	$\color{#35bf28}+1.19\%$
test_setitem_dim[range]	0.1640ms	84.9313μs	11.7742 KOps/s	11.7140 KOps/s	$\color{#35bf28}+0.51\%$
test_setitem_dim[tuple]	84.9720μs	51.5642μs	19.3933 KOps/s	19.1885 KOps/s	$\color{#35bf28}+1.07\%$
test_setitem	65.8760μs	20.3972μs	49.0264 KOps/s	48.6999 KOps/s	$\color{#35bf28}+0.67\%$
test_set	0.1189ms	19.8388μs	50.4063 KOps/s	49.7247 KOps/s	$\color{#35bf28}+1.37\%$
test_set_shared	3.5475ms	0.1633ms	6.1227 KOps/s	6.8527 KOps/s	$\textbf{\color{#d91a1a}-10.65\%}$
test_update	0.5256ms	22.1560μs	45.1344 KOps/s	44.7013 KOps/s	$\color{#35bf28}+0.97\%$
test_update_nested	99.8800μs	31.4472μs	31.7993 KOps/s	32.0865 KOps/s	$\color{#d91a1a}-0.89\%$
test_update__nested	83.7030μs	25.1154μs	39.8162 KOps/s	38.2660 KOps/s	$\color{#35bf28}+4.05\%$
test_set_nested	59.2830μs	21.5019μs	46.5076 KOps/s	44.7693 KOps/s	$\color{#35bf28}+3.88\%$
test_set_nested_new	87.3260μs	25.9071μs	38.5995 KOps/s	37.0603 KOps/s	$\color{#35bf28}+4.15\%$
test_select	93.3980μs	41.0229μs	24.3766 KOps/s	23.5495 KOps/s	$\color{#35bf28}+3.51\%$
test_select_nested	0.1181ms	60.2294μs	16.6032 KOps/s	16.1520 KOps/s	$\color{#35bf28}+2.79\%$
test_exclude_nested	0.2731ms	0.1196ms	8.3605 KOps/s	8.0052 KOps/s	$\color{#35bf28}+4.44\%$
test_empty[True]	0.4743ms	0.3898ms	2.5655 KOps/s	2.5168 KOps/s	$\color{#35bf28}+1.93\%$
test_empty[False]	6.2395μs	1.1555μs	865.4305 KOps/s	860.4843 KOps/s	$\color{#35bf28}+0.57\%$
test_unbind_speed	1.5573ms	0.2562ms	3.9037 KOps/s	3.8201 KOps/s	$\color{#35bf28}+2.19\%$
test_unbind_speed_stack0	0.4961ms	0.2572ms	3.8879 KOps/s	3.9729 KOps/s	$\color{#d91a1a}-2.14\%$
test_unbind_speed_stack1	68.8774ms	0.7348ms	1.3608 KOps/s	1.3558 KOps/s	$\color{#35bf28}+0.37\%$
test_split	67.1963ms	1.6187ms	617.7833 Ops/s	613.4495 Ops/s	$\color{#35bf28}+0.71\%$
test_chunk	73.1466ms	1.6415ms	609.1840 Ops/s	612.1814 Ops/s	$\color{#d91a1a}-0.49\%$
test_creation[device0]	0.1947ms	85.6720μs	11.6724 KOps/s	11.7757 KOps/s	$\color{#d91a1a}-0.88\%$
test_creation_from_tensor	4.1743ms	87.6779μs	11.4054 KOps/s	11.5845 KOps/s	$\color{#d91a1a}-1.55\%$
test_add_one[memmap_tensor0]	77.1270μs	5.3737μs	186.0900 KOps/s	183.5800 KOps/s	$\color{#35bf28}+1.37\%$
test_contiguous[memmap_tensor0]	10.4100μs	0.6625μs	1.5093 MOps/s	1.5838 MOps/s	$\color{#d91a1a}-4.70\%$
test_stack[memmap_tensor0]	17.8940μs	3.6924μs	270.8237 KOps/s	278.0537 KOps/s	$\color{#d91a1a}-2.60\%$
test_memmaptd_index	0.9262ms	0.2520ms	3.9686 KOps/s	3.9084 KOps/s	$\color{#35bf28}+1.54\%$
test_memmaptd_index_astensor	0.6753ms	0.3250ms	3.0772 KOps/s	3.0661 KOps/s	$\color{#35bf28}+0.36\%$
test_memmaptd_index_op	0.9547ms	0.6116ms	1.6350 KOps/s	1.6390 KOps/s	$\color{#d91a1a}-0.24\%$
test_serialize_model	0.1712s	0.1137s	8.7953 Ops/s	8.5715 Ops/s	$\color{#35bf28}+2.61\%$
test_serialize_model_pickle	0.4523s	0.3752s	2.6656 Ops/s	2.6361 Ops/s	$\color{#35bf28}+1.12\%$
test_serialize_weights	0.1857s	0.1131s	8.8438 Ops/s	8.6012 Ops/s	$\color{#35bf28}+2.82\%$
test_serialize_weights_returnearly	0.1995s	0.1345s	7.4373 Ops/s	7.9303 Ops/s	$\textbf{\color{#d91a1a}-6.22\%}$
test_serialize_weights_pickle	0.9622s	0.5753s	1.7382 Ops/s	2.4292 Ops/s	$\textbf{\color{#d91a1a}-28.45\%}$
test_serialize_weights_filesystem	0.1016s	95.0517ms	10.5206 Ops/s	10.2405 Ops/s	$\color{#35bf28}+2.74\%$
test_serialize_model_filesystem	0.1657s	99.8180ms	10.0182 Ops/s	10.0564 Ops/s	$\color{#d91a1a}-0.38\%$
test_reshape_pytree	53.0210μs	25.3116μs	39.5076 KOps/s	38.8694 KOps/s	$\color{#35bf28}+1.64\%$
test_reshape_td	69.8620μs	34.6516μs	28.8587 KOps/s	28.8067 KOps/s	$\color{#35bf28}+0.18\%$
test_view_pytree	58.1000μs	25.6374μs	39.0054 KOps/s	35.2751 KOps/s	$\textbf{\color{#35bf28}+10.57\%}$
test_view_td	95.5320μs	38.4547μs	26.0046 KOps/s	25.7627 KOps/s	$\color{#35bf28}+0.94\%$
test_unbind_pytree	76.3250μs	29.2941μs	34.1365 KOps/s	33.7609 KOps/s	$\color{#35bf28}+1.11\%$
test_unbind_td	0.4341ms	38.0541μs	26.2784 KOps/s	25.8867 KOps/s	$\color{#35bf28}+1.51\%$
test_split_pytree	65.0640μs	29.0863μs	34.3805 KOps/s	33.9467 KOps/s	$\color{#35bf28}+1.28\%$
test_split_td	0.5350ms	41.0194μs	24.3787 KOps/s	24.4765 KOps/s	$\color{#d91a1a}-0.40\%$
test_add_pytree	75.1830μs	34.8169μs	28.7217 KOps/s	28.5407 KOps/s	$\color{#35bf28}+0.63\%$
test_add_td	0.1257ms	57.1510μs	17.4975 KOps/s	17.0805 KOps/s	$\color{#35bf28}+2.44\%$
test_distributed	0.2037ms	0.1024ms	9.7612 KOps/s	9.6076 KOps/s	$\color{#35bf28}+1.60\%$
test_tdmodule	39.2250μs	18.6107μs	53.7326 KOps/s	54.7645 KOps/s	$\color{#d91a1a}-1.88\%$
test_tdmodule_dispatch	68.6210μs	36.0685μs	27.7250 KOps/s	20.6262 KOps/s	$\textbf{\color{#35bf28}+34.42\%}$
test_tdseq	42.9520μs	21.1343μs	47.3164 KOps/s	44.9526 KOps/s	$\textbf{\color{#35bf28}+5.26\%}$
test_tdseq_dispatch	72.4170μs	41.5086μs	24.0914 KOps/s	24.8587 KOps/s	$\color{#d91a1a}-3.09\%$
test_instantiation_functorch	1.7581ms	1.3619ms	734.2463 Ops/s	748.7774 Ops/s	$\color{#d91a1a}-1.94\%$
test_instantiation_td	1.5532ms	1.0299ms	970.9503 Ops/s	965.1835 Ops/s	$\color{#35bf28}+0.60\%$
test_exec_functorch	0.3550ms	0.1638ms	6.1050 KOps/s	6.2422 KOps/s	$\color{#d91a1a}-2.20\%$
test_exec_functional_call	0.2444ms	0.1513ms	6.6087 KOps/s	6.7799 KOps/s	$\color{#d91a1a}-2.53\%$
test_exec_td	0.2906ms	0.1476ms	6.7751 KOps/s	6.8712 KOps/s	$\color{#d91a1a}-1.40\%$
test_exec_td_decorator	0.8109ms	0.2245ms	4.4548 KOps/s	4.5450 KOps/s	$\color{#d91a1a}-1.98\%$
test_vmap_mlp_speed[True-True]	0.7180ms	0.4904ms	2.0390 KOps/s	2.0731 KOps/s	$\color{#d91a1a}-1.64\%$
test_vmap_mlp_speed[True-False]	0.7846ms	0.4905ms	2.0388 KOps/s	2.0018 KOps/s	$\color{#35bf28}+1.85\%$
test_vmap_mlp_speed[False-True]	0.5942ms	0.3996ms	2.5024 KOps/s	2.5673 KOps/s	$\color{#d91a1a}-2.53\%$
test_vmap_mlp_speed[False-False]	0.7676ms	0.3980ms	2.5126 KOps/s	2.5655 KOps/s	$\color{#d91a1a}-2.06\%$
test_vmap_mlp_speed_decorator[True-True]	1.1789ms	0.5568ms	1.7960 KOps/s	1.7965 KOps/s	$\color{#d91a1a}-0.03\%$
test_vmap_mlp_speed_decorator[True-False]	0.8956ms	0.5541ms	1.8046 KOps/s	1.6819 KOps/s	$\textbf{\color{#35bf28}+7.30\%}$
test_vmap_mlp_speed_decorator[False-True]	1.7430ms	0.4689ms	2.1326 KOps/s	2.1974 KOps/s	$\color{#d91a1a}-2.95\%$
test_vmap_mlp_speed_decorator[False-False]	0.7452ms	0.4548ms	2.1989 KOps/s	2.2041 KOps/s	$\color{#d91a1a}-0.24\%$
test_to_module_speed[True]	2.6855ms	1.7063ms	586.0497 Ops/s	595.3279 Ops/s	$\color{#d91a1a}-1.56\%$
test_to_module_speed[False]	2.5758ms	1.6773ms	596.1916 Ops/s	600.6650 Ops/s	$\color{#d91a1a}-0.74\%$
test_tc_init	56.7670μs	29.2001μs	34.2464 KOps/s	34.9283 KOps/s	$\color{#d91a1a}-1.95\%$
test_tc_init_nested	0.1470ms	62.2777μs	16.0571 KOps/s	16.2937 KOps/s	$\color{#d91a1a}-1.45\%$
test_tc_first_layer_tensor	3.8973μs	0.6846μs	1.4607 MOps/s	1.3975 MOps/s	$\color{#35bf28}+4.52\%$
test_tc_first_layer_nontensor	1.9011μs	0.6738μs	1.4841 MOps/s	1.4743 MOps/s	$\color{#35bf28}+0.67\%$
test_tc_second_layer_tensor	0.1232ms	1.8531μs	539.6282 KOps/s	539.0945 KOps/s	$\color{#35bf28}+0.10\%$
test_tc_second_layer_nontensor	22.8530μs	1.6432μs	608.5631 KOps/s	670.5113 KOps/s	$\textbf{\color{#d91a1a}-9.24\%}$
test_unbind	85.5244ms	6.4600ms	154.7994 Ops/s	187.7161 Ops/s	$\textbf{\color{#d91a1a}-17.54\%}$
test_full_like	16.7672ms	11.2110ms	89.1983 Ops/s	122.4476 Ops/s	$\textbf{\color{#d91a1a}-27.15\%}$
test_zeros_like	6.9307ms	6.0744ms	164.6260 Ops/s	172.4498 Ops/s	$\color{#d91a1a}-4.54\%$
test_ones_like	7.4221ms	6.4057ms	156.1113 Ops/s	140.2829 Ops/s	$\textbf{\color{#35bf28}+11.28\%}$
test_clone	14.2208ms	8.7043ms	114.8853 Ops/s	118.6913 Ops/s	$\color{#d91a1a}-3.21\%$
test_squeeze	65.3740μs	13.8783μs	72.0551 KOps/s	68.3577 KOps/s	$\textbf{\color{#35bf28}+5.41\%}$
test_unsqueeze	0.4248ms	62.5165μs	15.9958 KOps/s	16.5435 KOps/s	$\color{#d91a1a}-3.31\%$
test_split	0.1872ms	0.1156ms	8.6537 KOps/s	8.6700 KOps/s	$\color{#d91a1a}-0.19\%$
test_permute	0.1840ms	0.1287ms	7.7700 KOps/s	7.6335 KOps/s	$\color{#35bf28}+1.79\%$
test_stack	28.9411ms	23.4806ms	42.5884 Ops/s	42.6253 Ops/s	$\color{#d91a1a}-0.09\%$
test_cat	27.9652ms	23.5592ms	42.4462 Ops/s	43.7791 Ops/s	$\color{#d91a1a}-3.04\%$

[github-actions]

$\color{#D29922}\textsf{\Large⚠\kern{0.2cm}\normalsize Warning}$ Result of GPU Benchmark Tests

Total Benchmarks: 152. Improved: $\large\color{#35bf28}6$. Worsened: $\large\color{#d91a1a}30$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	33.4200μs	10.6606μs	93.8035 KOps/s	104.6444 KOps/s	$\textbf{\color{#d91a1a}-10.36\%}$
test_plain_set_stack_nested	38.1110μs	10.8504μs	92.1622 KOps/s	104.1689 KOps/s	$\textbf{\color{#d91a1a}-11.53\%}$
test_plain_set_nested_inplace	70.8920μs	11.9268μs	83.8450 KOps/s	92.9727 KOps/s	$\textbf{\color{#d91a1a}-9.82\%}$
test_plain_set_stack_nested_inplace	41.1210μs	11.8214μs	84.5927 KOps/s	93.9902 KOps/s	$\textbf{\color{#d91a1a}-10.00\%}$
test_items	36.2610μs	2.9200μs	342.4671 KOps/s	340.9665 KOps/s	$\color{#35bf28}+0.44\%$
test_items_nested	0.3779ms	0.3017ms	3.3147 KOps/s	3.3687 KOps/s	$\color{#d91a1a}-1.60\%$
test_items_nested_locked	0.3631ms	0.2965ms	3.3731 KOps/s	3.3407 KOps/s	$\color{#35bf28}+0.97\%$
test_items_nested_leaf	90.9020μs	63.8468μs	15.6625 KOps/s	15.7552 KOps/s	$\color{#d91a1a}-0.59\%$
test_items_stack_nested	0.3690ms	0.2971ms	3.3655 KOps/s	3.3590 KOps/s	$\color{#35bf28}+0.19\%$
test_items_stack_nested_leaf	94.1220μs	64.8780μs	15.4136 KOps/s	15.4670 KOps/s	$\color{#d91a1a}-0.35\%$
test_items_stack_nested_locked	0.3460ms	0.2970ms	3.3669 KOps/s	3.3152 KOps/s	$\color{#35bf28}+1.56\%$
test_keys	26.9500μs	3.4463μs	290.1661 KOps/s	288.8490 KOps/s	$\color{#35bf28}+0.46\%$
test_keys_nested	0.1145ms	55.3609μs	18.0633 KOps/s	18.0013 KOps/s	$\color{#35bf28}+0.34\%$
test_keys_nested_locked	2.5695ms	59.4904μs	16.8094 KOps/s	16.7180 KOps/s	$\color{#35bf28}+0.55\%$
test_keys_nested_leaf	76.8710μs	46.8738μs	21.3339 KOps/s	21.1253 KOps/s	$\color{#35bf28}+0.99\%$
test_keys_stack_nested	92.9220μs	55.2153μs	18.1109 KOps/s	18.0042 KOps/s	$\color{#35bf28}+0.59\%$
test_keys_stack_nested_leaf	93.7420μs	46.9320μs	21.3074 KOps/s	21.1178 KOps/s	$\color{#35bf28}+0.90\%$
test_keys_stack_nested_locked	89.7520μs	59.8029μs	16.7216 KOps/s	16.7586 KOps/s	$\color{#d91a1a}-0.22\%$
test_values	5.8418μs	0.8444μs	1.1843 MOps/s	1.1709 MOps/s	$\color{#35bf28}+1.15\%$
test_values_nested	93.3920μs	29.5109μs	33.8858 KOps/s	33.5344 KOps/s	$\color{#35bf28}+1.05\%$
test_values_nested_locked	57.1210μs	31.1616μs	32.0908 KOps/s	32.0068 KOps/s	$\color{#35bf28}+0.26\%$
test_values_nested_leaf	0.1781ms	26.2440μs	38.1039 KOps/s	38.4405 KOps/s	$\color{#d91a1a}-0.88\%$
test_values_stack_nested	57.1710μs	30.3925μs	32.9028 KOps/s	32.7746 KOps/s	$\color{#35bf28}+0.39\%$
test_values_stack_nested_leaf	62.5710μs	27.2646μs	36.6776 KOps/s	36.7988 KOps/s	$\color{#d91a1a}-0.33\%$
test_values_stack_nested_locked	55.1610μs	31.6830μs	31.5627 KOps/s	31.2365 KOps/s	$\color{#35bf28}+1.04\%$
test_membership	1.9881μs	0.6185μs	1.6167 MOps/s	1.6102 MOps/s	$\color{#35bf28}+0.40\%$
test_membership_nested	27.2110μs	2.2551μs	443.4471 KOps/s	440.5549 KOps/s	$\color{#35bf28}+0.66\%$
test_membership_nested_leaf	16.0655μs	2.1960μs	455.3721 KOps/s	453.3567 KOps/s	$\color{#35bf28}+0.44\%$
test_membership_stacked_nested	26.4900μs	2.2740μs	439.7607 KOps/s	439.7855 KOps/s	$-0.01\%$
test_membership_stacked_nested_leaf	28.1410μs	2.2664μs	441.2338 KOps/s	440.9663 KOps/s	$\color{#35bf28}+0.06\%$
test_membership_nested_last	29.2510μs	2.8025μs	356.8266 KOps/s	350.5947 KOps/s	$\color{#35bf28}+1.78\%$
test_membership_nested_leaf_last	0.1832ms	2.8161μs	355.1023 KOps/s	359.0402 KOps/s	$\color{#d91a1a}-1.10\%$
test_membership_stacked_nested_last	0.2087ms	4.1007μs	243.8621 KOps/s	356.4438 KOps/s	$\textbf{\color{#d91a1a}-31.58\%}$
test_membership_stacked_nested_leaf_last	28.6000μs	4.1493μs	241.0056 KOps/s	360.4639 KOps/s	$\textbf{\color{#d91a1a}-33.14\%}$
test_nested_getleaf	0.1687ms	6.4759μs	154.4181 KOps/s	153.2067 KOps/s	$\color{#35bf28}+0.79\%$
test_nested_get	0.2040ms	6.0846μs	164.3482 KOps/s	162.8555 KOps/s	$\color{#35bf28}+0.92\%$
test_stacked_getleaf	0.1883ms	6.4006μs	156.2365 KOps/s	156.1527 KOps/s	$\color{#35bf28}+0.05\%$
test_stacked_get	0.1885ms	5.9896μs	166.9568 KOps/s	166.2037 KOps/s	$\color{#35bf28}+0.45\%$
test_nested_getitemleaf	0.2039ms	6.5692μs	152.2253 KOps/s	150.1736 KOps/s	$\color{#35bf28}+1.37\%$
test_nested_getitem	47.1610μs	6.1684μs	162.1174 KOps/s	159.9524 KOps/s	$\color{#35bf28}+1.35\%$
test_stacked_getitemleaf	30.6910μs	6.5420μs	152.8578 KOps/s	152.0594 KOps/s	$\color{#35bf28}+0.53\%$
test_stacked_getitem	48.5210μs	6.0775μs	164.5415 KOps/s	162.2702 KOps/s	$\color{#35bf28}+1.40\%$
test_lock_nested	58.2551ms	0.3940ms	2.5383 KOps/s	2.5355 KOps/s	$\color{#35bf28}+0.11\%$
test_lock_stack_nested	0.3991ms	0.2920ms	3.4242 KOps/s	3.3890 KOps/s	$\color{#35bf28}+1.04\%$
test_unlock_nested	59.9287ms	0.3952ms	2.5304 KOps/s	2.5036 KOps/s	$\color{#35bf28}+1.07\%$
test_unlock_stack_nested	0.4066ms	0.3001ms	3.3325 KOps/s	3.2986 KOps/s	$\color{#35bf28}+1.03\%$
test_flatten_speed	0.3154ms	79.1217μs	12.6388 KOps/s	12.7795 KOps/s	$\color{#d91a1a}-1.10\%$
test_unflatten_speed	0.3259ms	0.2613ms	3.8273 KOps/s	3.8300 KOps/s	$\color{#d91a1a}-0.07\%$
test_common_ops	1.6978ms	0.6498ms	1.5390 KOps/s	1.6205 KOps/s	$\textbf{\color{#d91a1a}-5.03\%}$
test_creation	0.1109ms	1.4652μs	682.5127 KOps/s	696.6021 KOps/s	$\color{#d91a1a}-2.02\%$
test_creation_empty	93.8320μs	8.6230μs	115.9690 KOps/s	157.4603 KOps/s	$\textbf{\color{#d91a1a}-26.35\%}$
test_creation_nested_1	35.3310μs	10.3140μs	96.9559 KOps/s	125.0231 KOps/s	$\textbf{\color{#d91a1a}-22.45\%}$
test_creation_nested_2	0.1036ms	12.2186μs	81.8424 KOps/s	100.6253 KOps/s	$\textbf{\color{#d91a1a}-18.67\%}$
test_clone	92.3520μs	12.1082μs	82.5883 KOps/s	80.4669 KOps/s	$\color{#35bf28}+2.64\%$
test_getitem[int]	28.0210μs	10.9360μs	91.4407 KOps/s	92.5230 KOps/s	$\color{#d91a1a}-1.17\%$
test_getitem[slice_int]	0.1534ms	21.6483μs	46.1930 KOps/s	46.6034 KOps/s	$\color{#d91a1a}-0.88\%$
test_getitem[range]	64.7010μs	51.4210μs	19.4473 KOps/s	20.8312 KOps/s	$\textbf{\color{#d91a1a}-6.64\%}$
test_getitem[tuple]	0.1263ms	19.2153μs	52.0419 KOps/s	51.2310 KOps/s	$\color{#35bf28}+1.58\%$
test_getitem[list]	0.1310ms	36.5471μs	27.3619 KOps/s	27.5599 KOps/s	$\color{#d91a1a}-0.72\%$
test_setitem_dim[int]	51.0110μs	30.1881μs	33.1257 KOps/s	36.2995 KOps/s	$\textbf{\color{#d91a1a}-8.74\%}$
test_setitem_dim[slice_int]	0.1101ms	52.5998μs	19.0115 KOps/s	20.7065 KOps/s	$\textbf{\color{#d91a1a}-8.19\%}$
test_setitem_dim[range]	0.2002ms	69.0351μs	14.4854 KOps/s	15.2659 KOps/s	$\textbf{\color{#d91a1a}-5.11\%}$
test_setitem_dim[tuple]	78.5110μs	44.9620μs	22.2410 KOps/s	23.6957 KOps/s	$\textbf{\color{#d91a1a}-6.14\%}$
test_setitem	53.4410μs	16.7995μs	59.5255 KOps/s	60.7306 KOps/s	$\color{#d91a1a}-1.98\%$
test_set	53.8610μs	16.1390μs	61.9618 KOps/s	65.3917 KOps/s	$\textbf{\color{#d91a1a}-5.25\%}$
test_set_shared	2.1275ms	0.1565ms	6.3878 KOps/s	6.3639 KOps/s	$\color{#35bf28}+0.38\%$
test_update	0.5052ms	19.3314μs	51.7292 KOps/s	56.6287 KOps/s	$\textbf{\color{#d91a1a}-8.65\%}$
test_update_nested	0.1560ms	24.2533μs	41.2315 KOps/s	45.4391 KOps/s	$\textbf{\color{#d91a1a}-9.26\%}$
test_update__nested	0.1614ms	22.1114μs	45.2254 KOps/s	43.5262 KOps/s	$\color{#35bf28}+3.90\%$
test_set_nested	0.1234ms	17.2295μs	58.0398 KOps/s	59.1436 KOps/s	$\color{#d91a1a}-1.87\%$
test_set_nested_new	0.1531ms	20.4459μs	48.9095 KOps/s	51.6618 KOps/s	$\textbf{\color{#d91a1a}-5.33\%}$
test_select	0.1618ms	33.0823μs	30.2277 KOps/s	31.8775 KOps/s	$\textbf{\color{#d91a1a}-5.18\%}$
test_select_nested	1.0486ms	45.9656μs	21.7554 KOps/s	21.7454 KOps/s	$\color{#35bf28}+0.05\%$
test_exclude_nested	0.1304ms	96.1206μs	10.4036 KOps/s	10.1362 KOps/s	$\color{#35bf28}+2.64\%$
test_empty[True]	0.3563ms	0.3021ms	3.3106 KOps/s	3.2959 KOps/s	$\color{#35bf28}+0.45\%$
test_empty[False]	4.2661μs	0.8403μs	1.1900 MOps/s	1.1986 MOps/s	$\color{#d91a1a}-0.71\%$
test_to	0.1027ms	75.5226μs	13.2411 KOps/s	13.0539 KOps/s	$\color{#35bf28}+1.43\%$
test_to_nonblocking	0.2138ms	59.8256μs	16.7152 KOps/s	16.4153 KOps/s	$\color{#35bf28}+1.83\%$
test_unbind_speed	0.9182ms	0.2586ms	3.8676 KOps/s	3.7744 KOps/s	$\color{#35bf28}+2.47\%$
test_unbind_speed_stack0	0.4042ms	0.2631ms	3.8006 KOps/s	3.7378 KOps/s	$\color{#35bf28}+1.68\%$
test_unbind_speed_stack1	75.7023ms	0.8046ms	1.2429 KOps/s	1.2277 KOps/s	$\color{#35bf28}+1.23\%$
test_split	75.9697ms	1.7349ms	576.3885 Ops/s	569.0045 Ops/s	$\color{#35bf28}+1.30\%$
test_chunk	1.7057ms	1.5978ms	625.8596 Ops/s	617.6477 Ops/s	$\color{#35bf28}+1.33\%$
test_creation[device0]	0.3108ms	82.0647μs	12.1855 KOps/s	12.1822 KOps/s	$\color{#35bf28}+0.03\%$
test_creation_from_tensor	0.4619ms	83.0665μs	12.0385 KOps/s	12.0654 KOps/s	$\color{#d91a1a}-0.22\%$
test_add_one[memmap_tensor0]	0.4652ms	7.8485μs	127.4134 KOps/s	120.7102 KOps/s	$\textbf{\color{#35bf28}+5.55\%}$
test_contiguous[memmap_tensor0]	0.1956ms	0.6466μs	1.5464 MOps/s	1.5320 MOps/s	$\color{#35bf28}+0.94\%$
test_stack[memmap_tensor0]	35.9100μs	5.2292μs	191.2339 KOps/s	187.3580 KOps/s	$\color{#35bf28}+2.07\%$
test_memmaptd_index	1.0444ms	0.2939ms	3.4029 KOps/s	3.3637 KOps/s	$\color{#35bf28}+1.17\%$
test_memmaptd_index_astensor	0.6127ms	0.3599ms	2.7786 KOps/s	2.7731 KOps/s	$\color{#35bf28}+0.20\%$
test_memmaptd_index_op	78.3806ms	0.7671ms	1.3035 KOps/s	1.4840 KOps/s	$\textbf{\color{#d91a1a}-12.16\%}$
test_serialize_model	0.1364s	0.1322s	7.5650 Ops/s	7.6205 Ops/s	$\color{#d91a1a}-0.73\%$
test_serialize_model_pickle	1.3507s	1.2118s	0.8252 Ops/s	0.8234 Ops/s	$\color{#35bf28}+0.22\%$
test_serialize_weights	0.1319s	0.1302s	7.6777 Ops/s	7.0389 Ops/s	$\textbf{\color{#35bf28}+9.07\%}$
test_serialize_weights_returnearly	79.2730ms	68.7984ms	14.5352 Ops/s	12.4285 Ops/s	$\textbf{\color{#35bf28}+16.95\%}$
test_serialize_weights_pickle	1.3590s	1.2131s	0.8243 Ops/s	0.8232 Ops/s	$\color{#35bf28}+0.13\%$
test_reshape_pytree	0.1704ms	24.3612μs	41.0489 KOps/s	40.4989 KOps/s	$\color{#35bf28}+1.36\%$
test_reshape_td	59.3110μs	28.9786μs	34.5082 KOps/s	34.2216 KOps/s	$\color{#35bf28}+0.84\%$
test_view_pytree	0.1465ms	23.6826μs	42.2252 KOps/s	42.1642 KOps/s	$\color{#35bf28}+0.14\%$
test_view_td	0.1539ms	32.3329μs	30.9282 KOps/s	31.1189 KOps/s	$\color{#d91a1a}-0.61\%$
test_unbind_pytree	0.1250ms	33.1144μs	30.1983 KOps/s	30.0392 KOps/s	$\color{#35bf28}+0.53\%$
test_unbind_td	0.4272ms	39.9321μs	25.0425 KOps/s	24.1175 KOps/s	$\color{#35bf28}+3.84\%$
test_split_pytree	63.5210μs	33.3030μs	30.0273 KOps/s	29.2383 KOps/s	$\color{#35bf28}+2.70\%$
test_split_td	0.4838ms	38.9099μs	25.7004 KOps/s	25.4652 KOps/s	$\color{#35bf28}+0.92\%$
test_add_pytree	0.1083ms	38.1991μs	26.1786 KOps/s	25.2569 KOps/s	$\color{#35bf28}+3.65\%$
test_add_td	0.1667ms	53.9143μs	18.5479 KOps/s	20.3580 KOps/s	$\textbf{\color{#d91a1a}-8.89\%}$
test_distributed	0.8831ms	0.1155ms	8.6603 KOps/s	8.8916 KOps/s	$\color{#d91a1a}-2.60\%$
test_tdmodule	32.3400μs	13.5320μs	73.8988 KOps/s	81.8832 KOps/s	$\textbf{\color{#d91a1a}-9.75\%}$
test_tdmodule_dispatch	0.3436ms	27.7933μs	35.9799 KOps/s	41.9516 KOps/s	$\textbf{\color{#d91a1a}-14.23\%}$
test_tdseq	34.7900μs	15.5044μs	64.4979 KOps/s	69.6624 KOps/s	$\textbf{\color{#d91a1a}-7.41\%}$
test_tdseq_dispatch	50.6010μs	30.1488μs	33.1688 KOps/s	36.8202 KOps/s	$\textbf{\color{#d91a1a}-9.92\%}$
test_instantiation_functorch	1.5820ms	1.4339ms	697.4034 Ops/s	685.4485 Ops/s	$\color{#35bf28}+1.74\%$
test_instantiation_td	78.3114ms	1.0581ms	945.0585 Ops/s	1.0219 KOps/s	$\textbf{\color{#d91a1a}-7.52\%}$
test_exec_functorch	0.2658ms	0.1510ms	6.6204 KOps/s	6.3741 KOps/s	$\color{#35bf28}+3.86\%$
test_exec_functional_call	0.3188ms	0.1452ms	6.8862 KOps/s	6.7358 KOps/s	$\color{#35bf28}+2.23\%$
test_exec_td	0.3166ms	0.1472ms	6.7932 KOps/s	6.7771 KOps/s	$\color{#35bf28}+0.24\%$
test_exec_td_decorator	0.5835ms	0.2050ms	4.8778 KOps/s	4.7873 KOps/s	$\color{#35bf28}+1.89\%$
test_vmap_mlp_speed[True-True]	1.3146ms	0.6247ms	1.6007 KOps/s	1.6425 KOps/s	$\color{#d91a1a}-2.55\%$
test_vmap_mlp_speed[True-False]	0.8305ms	0.6249ms	1.6003 KOps/s	1.6510 KOps/s	$\color{#d91a1a}-3.07\%$
test_vmap_mlp_speed[False-True]	0.8463ms	0.5547ms	1.8028 KOps/s	1.8470 KOps/s	$\color{#d91a1a}-2.39\%$
test_vmap_mlp_speed[False-False]	0.7439ms	0.5627ms	1.7773 KOps/s	1.8292 KOps/s	$\color{#d91a1a}-2.84\%$
test_vmap_mlp_speed_decorator[True-True]	1.0368ms	0.6662ms	1.5010 KOps/s	1.5060 KOps/s	$\color{#d91a1a}-0.33\%$
test_vmap_mlp_speed_decorator[True-False]	0.8576ms	0.6645ms	1.5048 KOps/s	1.5061 KOps/s	$\color{#d91a1a}-0.09\%$
test_vmap_mlp_speed_decorator[False-True]	0.7914ms	0.6027ms	1.6592 KOps/s	1.6788 KOps/s	$\color{#d91a1a}-1.17\%$
test_vmap_mlp_speed_decorator[False-False]	0.8169ms	0.6016ms	1.6623 KOps/s	1.6750 KOps/s	$\color{#d91a1a}-0.76\%$
test_vmap_transformer_speed[True-True]	8.7128ms	8.2808ms	120.7608 Ops/s	120.1925 Ops/s	$\color{#35bf28}+0.47\%$
test_vmap_transformer_speed[True-False]	8.6301ms	8.2781ms	120.8003 Ops/s	120.6015 Ops/s	$\color{#35bf28}+0.16\%$
test_vmap_transformer_speed[False-True]	8.6247ms	8.2392ms	121.3711 Ops/s	121.2077 Ops/s	$\color{#35bf28}+0.13\%$
test_vmap_transformer_speed[False-False]	8.6884ms	8.1958ms	122.0134 Ops/s	121.6030 Ops/s	$\color{#35bf28}+0.34\%$
test_vmap_transformer_speed_decorator[True-True]	21.0073ms	19.9852ms	50.0370 Ops/s	49.8526 Ops/s	$\color{#35bf28}+0.37\%$
test_vmap_transformer_speed_decorator[True-False]	20.3819ms	19.9371ms	50.1577 Ops/s	50.0069 Ops/s	$\color{#35bf28}+0.30\%$
test_vmap_transformer_speed_decorator[False-True]	20.7662ms	19.9207ms	50.1990 Ops/s	50.2138 Ops/s	$\color{#d91a1a}-0.03\%$
test_vmap_transformer_speed_decorator[False-False]	20.8112ms	19.8950ms	50.2639 Ops/s	50.1948 Ops/s	$\color{#35bf28}+0.14\%$
test_to_module_speed[True]	1.3929ms	1.2628ms	791.9195 Ops/s	779.6518 Ops/s	$\color{#35bf28}+1.57\%$
test_to_module_speed[False]	1.3917ms	1.2709ms	786.8214 Ops/s	799.6234 Ops/s	$\color{#d91a1a}-1.60\%$
test_tc_init	50.6710μs	24.3407μs	41.0835 KOps/s	54.1200 KOps/s	$\textbf{\color{#d91a1a}-24.09\%}$
test_tc_init_nested	0.1064ms	49.4966μs	20.2034 KOps/s	25.3640 KOps/s	$\textbf{\color{#d91a1a}-20.35\%}$
test_tc_first_layer_tensor	0.8645μs	0.3386μs	2.9530 MOps/s	2.9672 MOps/s	$\color{#d91a1a}-0.48\%$
test_tc_first_layer_nontensor	2.9244μs	0.3527μs	2.8351 MOps/s	2.8676 MOps/s	$\color{#d91a1a}-1.13\%$
test_tc_second_layer_tensor	5.8318μs	0.8424μs	1.1871 MOps/s	1.1874 MOps/s	$\color{#d91a1a}-0.03\%$
test_tc_second_layer_nontensor	3.5010μs	0.7169μs	1.3950 MOps/s	1.3836 MOps/s	$\color{#35bf28}+0.82\%$
test_unbind	0.1059s	7.6070ms	131.4585 Ops/s	208.0919 Ops/s	$\textbf{\color{#d91a1a}-36.83\%}$
test_full_like	16.0835ms	15.5660ms	64.2427 Ops/s	44.4294 Ops/s	$\textbf{\color{#35bf28}+44.59\%}$
test_zeros_like	17.0425ms	16.1763ms	61.8189 Ops/s	59.9711 Ops/s	$\color{#35bf28}+3.08\%$
test_ones_like	16.8309ms	16.1121ms	62.0650 Ops/s	60.3122 Ops/s	$\color{#35bf28}+2.91\%$
test_clone	17.8434ms	17.3053ms	57.7858 Ops/s	56.0869 Ops/s	$\color{#35bf28}+3.03\%$
test_squeeze	0.1778ms	9.4526μs	105.7915 KOps/s	108.6076 KOps/s	$\color{#d91a1a}-2.59\%$
test_unsqueeze	0.2247ms	50.8601μs	19.6618 KOps/s	20.5042 KOps/s	$\color{#d91a1a}-4.11\%$
test_split	0.2071ms	93.5986μs	10.6839 KOps/s	10.4663 KOps/s	$\color{#35bf28}+2.08\%$
test_permute	0.2351ms	0.1035ms	9.6613 KOps/s	9.0269 KOps/s	$\textbf{\color{#35bf28}+7.03\%}$
test_stack	51.4211ms	50.8746ms	19.6562 Ops/s	18.6397 Ops/s	$\textbf{\color{#35bf28}+5.45\%}$
test_cat	51.6123ms	50.8517ms	19.6650 Ops/s	19.0448 Ops/s	$\color{#35bf28}+3.26\%$

[matteobettini]

Suggested change

	if interaction_type is InteractionType.MODE:
	elif interaction_type is InteractionType.MODE:

[vmoens]

not necessary, the function either returns or raises just before