[Refactor] composite_lp_aggregate to handle log-probs aggregates globally

[vmoens]

I propose a global flag composite_lp_aggregate to handle the issue of the aggregation of log-probs.

So far, we have dealt with this using kwargs everywhere (in ProbabilisticTDModule, ProbabilisticTDSequential, CompositeDistribution and subclasses). The hierarchy of these classes and what to do when args conflict isn't easy to handle. It's also confusing for users, who I suspect will usually want to work with either collapsed or non-collapsed log-probs.

A global flag (set to True for now and False in the future) will make things easier to handle.
Globally, the v0.6.2 behaviour will not be changed all users who rely on it will be informed about the upcoming change through a warning that will tell them to set the global var to False to accommodate upcoming changes. If they set it to True, nothing will change for them but that also means that bugs will not be solved (we won't maintain the True behaviour).

When composite_lp_aggregate() == True, we'll have aggregate_probabilities=True, include_sum=True and inplace=True by default everywhere. When composite_lp_aggregate() == False, all of these will be set to False, meaning that any call to whatever.log_prob(tensordict) will return another tensordict containing the leaf log-probs.

[github-actions]

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

Total Benchmarks: 217. Improved: $\large\color{#35bf28}15$. Worsened: $\large\color{#d91a1a}27$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	49.8230μs	21.2783μs	46.9963 KOps/s	51.9469 KOps/s	$\textbf{\color{#d91a1a}-9.53\%}$
test_plain_set_stack_nested	55.7540μs	21.5773μs	46.3451 KOps/s	52.2627 KOps/s	$\textbf{\color{#d91a1a}-11.32\%}$
test_plain_set_nested_inplace	59.9320μs	22.8812μs	43.7040 KOps/s	47.3951 KOps/s	$\textbf{\color{#d91a1a}-7.79\%}$
test_plain_set_stack_nested_inplace	72.8360μs	22.9579μs	43.5579 KOps/s	47.8776 KOps/s	$\textbf{\color{#d91a1a}-9.02\%}$
test_items	27.1710μs	4.2007μs	238.0575 KOps/s	237.3068 KOps/s	$\color{#35bf28}+0.32\%$
test_items_nested	0.7366ms	0.4132ms	2.4202 KOps/s	2.5275 KOps/s	$\color{#d91a1a}-4.25\%$
test_items_nested_locked	0.7156ms	0.4093ms	2.4431 KOps/s	2.4716 KOps/s	$\color{#d91a1a}-1.15\%$
test_items_nested_leaf	0.1516ms	78.7568μs	12.6973 KOps/s	12.8632 KOps/s	$\color{#d91a1a}-1.29\%$
test_items_stack_nested	0.5882ms	0.4100ms	2.4391 KOps/s	2.5033 KOps/s	$\color{#d91a1a}-2.57\%$
test_items_stack_nested_leaf	0.1521ms	78.8587μs	12.6809 KOps/s	12.7487 KOps/s	$\color{#d91a1a}-0.53\%$
test_items_stack_nested_locked	0.6345ms	0.4116ms	2.4294 KOps/s	2.4874 KOps/s	$\color{#d91a1a}-2.33\%$
test_keys	22.6020μs	3.4875μs	286.7374 KOps/s	288.6082 KOps/s	$\color{#d91a1a}-0.65\%$
test_keys_nested	0.3129ms	0.1625ms	6.1550 KOps/s	6.0063 KOps/s	$\color{#35bf28}+2.48\%$
test_keys_nested_locked	0.6861ms	0.1682ms	5.9470 KOps/s	5.8588 KOps/s	$\color{#35bf28}+1.51\%$
test_keys_nested_leaf	0.2305ms	0.1423ms	7.0288 KOps/s	6.9533 KOps/s	$\color{#35bf28}+1.09\%$
test_keys_stack_nested	0.2879ms	0.1634ms	6.1214 KOps/s	6.0092 KOps/s	$\color{#35bf28}+1.87\%$
test_keys_stack_nested_leaf	0.2216ms	0.1418ms	7.0526 KOps/s	6.9686 KOps/s	$\color{#35bf28}+1.21\%$
test_keys_stack_nested_locked	0.3064ms	0.1681ms	5.9490 KOps/s	5.8710 KOps/s	$\color{#35bf28}+1.33\%$
test_values	5.5084μs	1.0291μs	971.7639 KOps/s	965.5300 KOps/s	$\color{#35bf28}+0.65\%$
test_values_nested	0.1142ms	61.5524μs	16.2463 KOps/s	15.9085 KOps/s	$\color{#35bf28}+2.12\%$
test_values_nested_locked	0.1125ms	61.5663μs	16.2426 KOps/s	15.9646 KOps/s	$\color{#35bf28}+1.74\%$
test_values_nested_leaf	0.1242ms	70.3801μs	14.2086 KOps/s	12.8642 KOps/s	$\textbf{\color{#35bf28}+10.45\%}$
test_values_stack_nested	0.1481ms	61.2655μs	16.3224 KOps/s	15.6874 KOps/s	$\color{#35bf28}+4.05\%$
test_values_stack_nested_leaf	0.1023ms	69.9535μs	14.2952 KOps/s	14.0240 KOps/s	$\color{#35bf28}+1.93\%$
test_values_stack_nested_locked	0.1317ms	61.4307μs	16.2785 KOps/s	15.8429 KOps/s	$\color{#35bf28}+2.75\%$
test_membership	3.4636μs	0.7184μs	1.3920 MOps/s	1.3906 MOps/s	$\color{#35bf28}+0.10\%$
test_membership_nested	23.6140μs	2.9320μs	341.0641 KOps/s	342.5849 KOps/s	$\color{#d91a1a}-0.44\%$
test_membership_nested_leaf	21.5500μs	2.9650μs	337.2717 KOps/s	339.7894 KOps/s	$\color{#d91a1a}-0.74\%$
test_membership_stacked_nested	21.1900μs	2.9376μs	340.4130 KOps/s	345.8762 KOps/s	$\color{#d91a1a}-1.58\%$
test_membership_stacked_nested_leaf	29.1240μs	2.9708μs	336.6122 KOps/s	344.7126 KOps/s	$\color{#d91a1a}-2.35\%$
test_membership_nested_last	23.7850μs	4.4172μs	226.3861 KOps/s	229.2153 KOps/s	$\color{#d91a1a}-1.23\%$
test_membership_nested_leaf_last	27.5810μs	4.4555μs	224.4439 KOps/s	227.9879 KOps/s	$\color{#d91a1a}-1.55\%$
test_membership_stacked_nested_last	27.7320μs	4.4207μs	226.2066 KOps/s	228.3774 KOps/s	$\color{#d91a1a}-0.95\%$
test_membership_stacked_nested_leaf_last	33.0810μs	4.4126μs	226.6244 KOps/s	227.2498 KOps/s	$\color{#d91a1a}-0.28\%$
test_nested_getleaf	0.1096ms	10.7878μs	92.6970 KOps/s	94.1166 KOps/s	$\color{#d91a1a}-1.51\%$
test_nested_get	32.1900μs	10.1279μs	98.7368 KOps/s	99.0653 KOps/s	$\color{#d91a1a}-0.33\%$
test_stacked_getleaf	37.0790μs	10.5626μs	94.6740 KOps/s	94.2444 KOps/s	$\color{#35bf28}+0.46\%$
test_stacked_get	33.0520μs	10.0679μs	99.3254 KOps/s	99.4643 KOps/s	$\color{#d91a1a}-0.14\%$
test_nested_getitemleaf	38.2720μs	11.1284μs	89.8604 KOps/s	89.7692 KOps/s	$\color{#35bf28}+0.10\%$
test_nested_getitem	35.7070μs	10.5291μs	94.9744 KOps/s	94.8203 KOps/s	$\color{#35bf28}+0.16\%$
test_stacked_getitemleaf	97.0810μs	11.0149μs	90.7865 KOps/s	89.9089 KOps/s	$\color{#35bf28}+0.98\%$
test_stacked_getitem	34.1230μs	10.4366μs	95.8165 KOps/s	94.6037 KOps/s	$\color{#35bf28}+1.28\%$
test_lock_nested	8.0101ms	0.4624ms	2.1626 KOps/s	1.7969 KOps/s	$\textbf{\color{#35bf28}+20.35\%}$
test_lock_stack_nested	0.7875ms	0.4316ms	2.3170 KOps/s	2.3658 KOps/s	$\color{#d91a1a}-2.06\%$
test_unlock_nested	0.9373ms	0.3775ms	2.6493 KOps/s	2.6828 KOps/s	$\color{#d91a1a}-1.25\%$
test_unlock_stack_nested	0.5989ms	0.3484ms	2.8702 KOps/s	2.9258 KOps/s	$\color{#d91a1a}-1.90\%$
test_flatten_speed	0.4773ms	0.1038ms	9.6312 KOps/s	10.1345 KOps/s	$\color{#d91a1a}-4.97\%$
test_unflatten_speed	0.6273ms	0.5329ms	1.8764 KOps/s	1.9348 KOps/s	$\color{#d91a1a}-3.02\%$
test_common_ops	1.9100ms	0.8077ms	1.2381 KOps/s	1.3264 KOps/s	$\textbf{\color{#d91a1a}-6.66\%}$
test_creation	44.6940μs	2.5720μs	388.8034 KOps/s	404.7572 KOps/s	$\color{#d91a1a}-3.94\%$
test_creation_empty	36.8180μs	13.4418μs	74.3948 KOps/s	105.3372 KOps/s	$\textbf{\color{#d91a1a}-29.37\%}$
test_creation_nested_1	49.2720μs	16.5613μs	60.3816 KOps/s	81.7070 KOps/s	$\textbf{\color{#d91a1a}-26.10\%}$
test_creation_nested_2	74.5690μs	20.8806μs	47.8913 KOps/s	58.5738 KOps/s	$\textbf{\color{#d91a1a}-18.24\%}$
test_clone	54.9830μs	13.8258μs	72.3288 KOps/s	73.1026 KOps/s	$\color{#d91a1a}-1.06\%$
test_getitem[int]	1.2290ms	13.1004μs	76.3338 KOps/s	77.3476 KOps/s	$\color{#d91a1a}-1.31\%$
test_getitem[slice_int]	0.1392ms	25.0700μs	39.8882 KOps/s	39.9212 KOps/s	$\color{#d91a1a}-0.08\%$
test_getitem[range]	0.1657ms	48.2914μs	20.7076 KOps/s	19.8904 KOps/s	$\color{#35bf28}+4.11\%$
test_getitem[tuple]	0.1325ms	20.6038μs	48.5347 KOps/s	49.0389 KOps/s	$\color{#d91a1a}-1.03\%$
test_getitem[list]	0.1693ms	44.0739μs	22.6892 KOps/s	21.9134 KOps/s	$\color{#35bf28}+3.54\%$
test_setitem_dim[int]	56.2450μs	25.5227μs	39.1808 KOps/s	37.6849 KOps/s	$\color{#35bf28}+3.97\%$
test_setitem_dim[slice_int]	92.1010μs	52.1629μs	19.1707 KOps/s	19.0070 KOps/s	$\color{#35bf28}+0.86\%$
test_setitem_dim[range]	0.1159ms	72.7516μs	13.7454 KOps/s	13.3502 KOps/s	$\color{#35bf28}+2.96\%$
test_setitem_dim[tuple]	0.1134ms	40.9603μs	24.4139 KOps/s	23.8646 KOps/s	$\color{#35bf28}+2.30\%$
test_setitem	61.7450μs	21.9697μs	45.5173 KOps/s	51.1170 KOps/s	$\textbf{\color{#d91a1a}-10.95\%}$
test_set	65.1910μs	21.3380μs	46.8648 KOps/s	52.6455 KOps/s	$\textbf{\color{#d91a1a}-10.98\%}$
test_set_shared	4.2512ms	0.1711ms	5.8444 KOps/s	5.8271 KOps/s	$\color{#35bf28}+0.30\%$
test_update	0.3706ms	24.4082μs	40.9698 KOps/s	48.8780 KOps/s	$\textbf{\color{#d91a1a}-16.18\%}$
test_update_nested	80.8500μs	35.1263μs	28.4687 KOps/s	33.0236 KOps/s	$\textbf{\color{#d91a1a}-13.79\%}$
test_update__nested	1.0082ms	34.2226μs	29.2204 KOps/s	29.7034 KOps/s	$\color{#d91a1a}-1.63\%$
test_set_nested	78.7870μs	23.1802μs	43.1402 KOps/s	48.0020 KOps/s	$\textbf{\color{#d91a1a}-10.13\%}$
test_set_nested_new	85.4990μs	27.9400μs	35.7910 KOps/s	39.1881 KOps/s	$\textbf{\color{#d91a1a}-8.67\%}$
test_select	0.1176ms	44.5983μs	22.4224 KOps/s	24.4016 KOps/s	$\textbf{\color{#d91a1a}-8.11\%}$
test_select_nested	0.1224ms	64.6981μs	15.4564 KOps/s	16.0853 KOps/s	$\color{#d91a1a}-3.91\%$
test_exclude_nested	0.1605ms	84.0326μs	11.9001 KOps/s	12.4220 KOps/s	$\color{#d91a1a}-4.20\%$
test_empty[True]	0.5622ms	0.4087ms	2.4467 KOps/s	2.4764 KOps/s	$\color{#d91a1a}-1.20\%$
test_empty[False]	34.3040μs	1.5038μs	664.9603 KOps/s	726.7559 KOps/s	$\textbf{\color{#d91a1a}-8.50\%}$
test_unbind_speed	0.4290ms	0.2755ms	3.6295 KOps/s	3.7090 KOps/s	$\color{#d91a1a}-2.14\%$
test_unbind_speed_stack0	0.4082ms	0.2716ms	3.6821 KOps/s	3.7823 KOps/s	$\color{#d91a1a}-2.65\%$
test_unbind_speed_stack1	0.1129s	0.8294ms	1.2057 KOps/s	1.3565 KOps/s	$\textbf{\color{#d91a1a}-11.12\%}$
test_split	0.1044s	1.7957ms	556.9003 Ops/s	562.5401 Ops/s	$\color{#d91a1a}-1.00\%$
test_chunk	3.0621ms	1.6285ms	614.0454 Ops/s	554.7686 Ops/s	$\textbf{\color{#35bf28}+10.68\%}$
test_consolidate_njt[False-None]	8.3733ms	8.0165ms	124.7428 Ops/s	118.6417 Ops/s	$\textbf{\color{#35bf28}+5.14\%}$
test_creation[device0]	4.2875ms	92.7868μs	10.7774 KOps/s	10.6282 KOps/s	$\color{#35bf28}+1.40\%$
test_creation_from_tensor	0.2721ms	95.8106μs	10.4373 KOps/s	10.3061 KOps/s	$\color{#35bf28}+1.27\%$
test_add_one[memmap_tensor0]	0.1440ms	5.0565μs	197.7633 KOps/s	201.7513 KOps/s	$\color{#d91a1a}-1.98\%$
test_contiguous[memmap_tensor0]	20.3980μs	0.5183μs	1.9295 MOps/s	1.9465 MOps/s	$\color{#d91a1a}-0.87\%$
test_stack[memmap_tensor0]	53.8410μs	3.4221μs	292.2158 KOps/s	293.3690 KOps/s	$\color{#d91a1a}-0.39\%$
test_memmaptd_index	0.9493ms	0.2317ms	4.3166 KOps/s	4.1133 KOps/s	$\color{#35bf28}+4.94\%$
test_memmaptd_index_astensor	0.5901ms	0.3169ms	3.1559 KOps/s	3.0102 KOps/s	$\color{#35bf28}+4.84\%$
test_memmaptd_index_op	0.9711ms	0.6113ms	1.6359 KOps/s	1.7127 KOps/s	$\color{#d91a1a}-4.48\%$
test_serialize_model	0.1256s	0.1162s	8.6022 Ops/s	8.3469 Ops/s	$\color{#35bf28}+3.06\%$
test_serialize_model_pickle	0.4937s	0.4047s	2.4707 Ops/s	2.5328 Ops/s	$\color{#d91a1a}-2.45\%$
test_serialize_weights	0.1291s	0.1139s	8.7765 Ops/s	8.3739 Ops/s	$\color{#35bf28}+4.81\%$
test_serialize_weights_returnearly	0.2663s	0.1768s	5.6558 Ops/s	6.3680 Ops/s	$\textbf{\color{#d91a1a}-11.18\%}$
test_serialize_weights_pickle	0.5564s	0.4561s	2.1927 Ops/s	2.4815 Ops/s	$\textbf{\color{#d91a1a}-11.64\%}$
test_serialize_weights_filesystem	0.1508s	0.1444s	6.9245 Ops/s	6.8490 Ops/s	$\color{#35bf28}+1.10\%$
test_serialize_model_filesystem	0.1532s	0.1451s	6.8921 Ops/s	6.0346 Ops/s	$\textbf{\color{#35bf28}+14.21\%}$
test_reshape_pytree	82.1730μs	26.4879μs	37.7531 KOps/s	38.0296 KOps/s	$\color{#d91a1a}-0.73\%$
test_reshape_td	98.5540μs	32.6995μs	30.5815 KOps/s	29.3637 KOps/s	$\color{#35bf28}+4.15\%$
test_view_pytree	85.5680μs	26.6994μs	37.4541 KOps/s	37.7465 KOps/s	$\color{#d91a1a}-0.77\%$
test_view_td	0.1108ms	38.2904μs	26.1162 KOps/s	25.1195 KOps/s	$\color{#35bf28}+3.97\%$
test_unbind_pytree	99.2760μs	29.4235μs	33.9865 KOps/s	33.8925 KOps/s	$\color{#35bf28}+0.28\%$
test_unbind_td	0.2903ms	39.9702μs	25.0186 KOps/s	25.1026 KOps/s	$\color{#d91a1a}-0.33\%$
test_split_pytree	62.9170μs	29.0833μs	34.3840 KOps/s	34.1638 KOps/s	$\color{#35bf28}+0.64\%$
test_split_td	0.4574ms	46.5805μs	21.4682 KOps/s	22.1353 KOps/s	$\color{#d91a1a}-3.01\%$
test_add_pytree	71.1730μs	35.6355μs	28.0619 KOps/s	27.6990 KOps/s	$\color{#35bf28}+1.31\%$
test_add_td	0.1336ms	59.3027μs	16.8626 KOps/s	18.8713 KOps/s	$\textbf{\color{#d91a1a}-10.64\%}$
test_compile_add_one_nested[tensordict-compile]	0.2356ms	64.0096μs	15.6227 KOps/s	15.9695 KOps/s	$\color{#d91a1a}-2.17\%$
test_compile_add_one_nested[tensordict-eager]	1.2393ms	0.1746ms	5.7268 KOps/s	5.7161 KOps/s	$\color{#35bf28}+0.19\%$
test_compile_add_one_nested[pytree-compile]	0.1740ms	45.9232μs	21.7755 KOps/s	21.9187 KOps/s	$\color{#d91a1a}-0.65\%$
test_compile_add_one_nested[pytree-eager]	0.2848ms	0.1189ms	8.4083 KOps/s	8.3939 KOps/s	$\color{#35bf28}+0.17\%$
test_compile_copy_nested[tensordict-compile]	80.7210μs	26.6745μs	37.4890 KOps/s	38.3240 KOps/s	$\color{#d91a1a}-2.18\%$
test_compile_copy_nested[tensordict-eager]	0.1730ms	59.0825μs	16.9255 KOps/s	17.1276 KOps/s	$\color{#d91a1a}-1.18\%$
test_compile_copy_nested[pytree-compile]	0.3633ms	77.6924μs	12.8713 KOps/s	12.7929 KOps/s	$\color{#35bf28}+0.61\%$
test_compile_copy_nested[pytree-eager]	0.1266ms	66.6663μs	15.0001 KOps/s	14.9432 KOps/s	$\color{#35bf28}+0.38\%$
test_compile_add_one_flat[tensordict-compile]	0.2584ms	0.1048ms	9.5408 KOps/s	9.4630 KOps/s	$\color{#35bf28}+0.82\%$
test_compile_add_one_flat[tensordict-eager]	0.3932ms	0.2115ms	4.7280 KOps/s	4.6274 KOps/s	$\color{#35bf28}+2.17\%$
test_compile_add_one_flat[tensorclass-compile]	0.2390ms	45.4254μs	22.0141 KOps/s	21.4143 KOps/s	$\color{#35bf28}+2.80\%$
test_compile_add_one_flat[tensorclass-eager]	0.5023ms	66.4343μs	15.0525 KOps/s	14.6494 KOps/s	$\color{#35bf28}+2.75\%$
test_compile_add_one_flat[pytree-compile]	0.2503ms	0.1025ms	9.7571 KOps/s	9.6369 KOps/s	$\color{#35bf28}+1.25\%$
test_compile_add_one_flat[pytree-eager]	0.4621ms	0.1977ms	5.0582 KOps/s	4.9444 KOps/s	$\color{#35bf28}+2.30\%$
test_compile_add_self_flat[tensordict-eager]	0.5003ms	0.2278ms	4.3890 KOps/s	4.2975 KOps/s	$\color{#35bf28}+2.13\%$
test_compile_add_self_flat[tensordict-compile]	0.2106ms	0.1032ms	9.6932 KOps/s	9.0870 KOps/s	$\textbf{\color{#35bf28}+6.67\%}$
test_compile_add_self_flat[tensorclass-eager]	0.1423ms	62.1489μs	16.0904 KOps/s	15.7295 KOps/s	$\color{#35bf28}+2.29\%$
test_compile_add_self_flat[tensorclass-compile]	0.1409ms	46.7859μs	21.3740 KOps/s	21.2906 KOps/s	$\color{#35bf28}+0.39\%$
test_compile_add_self_flat[pytree-eager]	0.1943ms	0.1567ms	6.3834 KOps/s	6.3975 KOps/s	$\color{#d91a1a}-0.22\%$
test_compile_add_self_flat[pytree-compile]	0.2185ms	0.1016ms	9.8383 KOps/s	9.6278 KOps/s	$\color{#35bf28}+2.19\%$
test_compile_copy_flat[tensordict-compile]	77.4750μs	21.3447μs	46.8501 KOps/s	46.6089 KOps/s	$\color{#35bf28}+0.52\%$
test_compile_copy_flat[tensordict-eager]	0.1268ms	65.7317μs	15.2134 KOps/s	14.7783 KOps/s	$\color{#35bf28}+2.94\%$
test_compile_copy_flat[pytree-compile]	0.1516ms	77.8115μs	12.8516 KOps/s	12.4167 KOps/s	$\color{#35bf28}+3.50\%$
test_compile_copy_flat[pytree-eager]	0.1358ms	66.8626μs	14.9560 KOps/s	14.8524 KOps/s	$\color{#35bf28}+0.70\%$
test_compile_assign_and_add[tensordict-compile]	0.5603ms	0.2077ms	4.8140 KOps/s	4.8130 KOps/s	$\color{#35bf28}+0.02\%$
test_compile_assign_and_add[tensordict-eager]	2.3384ms	1.3218ms	756.5484 Ops/s	764.1802 Ops/s	$\color{#d91a1a}-1.00\%$
test_compile_assign_and_add[pytree-compile]	0.7701ms	0.2062ms	4.8499 KOps/s	4.8343 KOps/s	$\color{#35bf28}+0.32\%$
test_compile_assign_and_add[pytree-eager]	0.9175ms	0.7703ms	1.2981 KOps/s	1.2174 KOps/s	$\textbf{\color{#35bf28}+6.64\%}$
test_compile_assign_and_add_stack[compile]	0.8202ms	0.4550ms	2.1976 KOps/s	2.1689 KOps/s	$\color{#35bf28}+1.32\%$
test_compile_assign_and_add_stack[eager]	3.0584ms	2.7401ms	364.9447 Ops/s	393.7284 Ops/s	$\textbf{\color{#d91a1a}-7.31\%}$
test_compile_indexing[tensor-tensordict-compile]	0.2345ms	37.7343μs	26.5011 KOps/s	27.6066 KOps/s	$\color{#d91a1a}-4.00\%$
test_compile_indexing[tensor-tensordict-eager]	0.6436ms	33.6577μs	29.7109 KOps/s	30.1967 KOps/s	$\color{#d91a1a}-1.61\%$
test_compile_indexing[tensor-tensorclass-compile]	98.3730μs	29.9385μs	33.4018 KOps/s	34.2140 KOps/s	$\color{#d91a1a}-2.37\%$
test_compile_indexing[tensor-tensorclass-eager]	0.2096ms	23.2320μs	43.0440 KOps/s	43.3744 KOps/s	$\color{#d91a1a}-0.76\%$
test_compile_indexing[tensor-pytree-compile]	96.0100μs	30.8566μs	32.4080 KOps/s	33.6282 KOps/s	$\color{#d91a1a}-3.63\%$
test_compile_indexing[tensor-pytree-eager]	0.1091ms	23.0572μs	43.3703 KOps/s	43.8603 KOps/s	$\color{#d91a1a}-1.12\%$
test_compile_indexing[slice-tensordict-compile]	0.1392ms	52.3701μs	19.0949 KOps/s	19.0326 KOps/s	$\color{#35bf28}+0.33\%$
test_compile_indexing[slice-tensordict-eager]	0.6733ms	20.4640μs	48.8664 KOps/s	48.5273 KOps/s	$\color{#35bf28}+0.70\%$
test_compile_indexing[slice-tensorclass-compile]	0.1653ms	44.7785μs	22.3322 KOps/s	22.7484 KOps/s	$\color{#d91a1a}-1.83\%$
test_compile_indexing[slice-tensorclass-eager]	0.1077ms	18.8925μs	52.9311 KOps/s	54.2203 KOps/s	$\color{#d91a1a}-2.38\%$
test_compile_indexing[slice-pytree-compile]	0.1129ms	45.0179μs	22.2134 KOps/s	22.3369 KOps/s	$\color{#d91a1a}-0.55\%$
test_compile_indexing[slice-pytree-eager]	68.1770μs	18.7976μs	53.1983 KOps/s	54.7352 KOps/s	$\color{#d91a1a}-2.81\%$
test_compile_indexing[int-tensordict-compile]	0.2699ms	53.0842μs	18.8380 KOps/s	18.2959 KOps/s	$\color{#35bf28}+2.96\%$
test_compile_indexing[int-tensordict-eager]	1.0086ms	20.2937μs	49.2764 KOps/s	50.3167 KOps/s	$\color{#d91a1a}-2.07\%$
test_compile_indexing[int-tensorclass-compile]	0.1168ms	45.2870μs	22.0814 KOps/s	22.1497 KOps/s	$\color{#d91a1a}-0.31\%$
test_compile_indexing[int-tensorclass-eager]	83.8600μs	18.7685μs	53.2809 KOps/s	54.1608 KOps/s	$\color{#d91a1a}-1.62\%$
test_compile_indexing[int-pytree-compile]	0.1391ms	45.5874μs	21.9359 KOps/s	22.2767 KOps/s	$\color{#d91a1a}-1.53\%$
test_compile_indexing[int-pytree-eager]	88.1240μs	19.0935μs	52.3739 KOps/s	54.2129 KOps/s	$\color{#d91a1a}-3.39\%$
test_mod_add[eager]	0.1007ms	35.0369μs	28.5413 KOps/s	30.2198 KOps/s	$\textbf{\color{#d91a1a}-5.55\%}$
test_mod_add[compile]	0.1736ms	49.4849μs	20.2082 KOps/s	20.5096 KOps/s	$\color{#d91a1a}-1.47\%$
test_mod_add[compile-overhead]	0.1560ms	48.9675μs	20.4217 KOps/s	20.3653 KOps/s	$\color{#35bf28}+0.28\%$
test_mod_wrap[eager]	0.5347ms	0.2195ms	4.5561 KOps/s	4.4726 KOps/s	$\color{#35bf28}+1.87\%$
test_mod_wrap[compile]	0.5324ms	0.2082ms	4.8029 KOps/s	4.7935 KOps/s	$\color{#35bf28}+0.20\%$
test_mod_wrap[compile-overhead]	0.4206ms	0.2049ms	4.8802 KOps/s	4.8319 KOps/s	$\color{#35bf28}+1.00\%$
test_mod_wrap_and_backward[eager]	12.3741ms	10.9780ms	91.0914 Ops/s	87.3708 Ops/s	$\color{#35bf28}+4.26\%$
test_mod_wrap_and_backward[compile]	12.3040ms	10.9432ms	91.3808 Ops/s	72.8930 Ops/s	$\textbf{\color{#35bf28}+25.36\%}$
test_mod_wrap_and_backward[compile-overhead]	12.7825ms	11.0645ms	90.3789 Ops/s	77.2105 Ops/s	$\textbf{\color{#35bf28}+17.06\%}$
test_seq_add[eager]	0.4073ms	0.1200ms	8.3305 KOps/s	8.6779 KOps/s	$\color{#d91a1a}-4.00\%$
test_seq_add[compile]	0.1437ms	65.6847μs	15.2242 KOps/s	16.0620 KOps/s	$\textbf{\color{#d91a1a}-5.22\%}$
test_seq_add[compile-overhead]	0.1954ms	63.8782μs	15.6548 KOps/s	16.0536 KOps/s	$\color{#d91a1a}-2.48\%$
test_seq_wrap[eager]	0.6229ms	0.4460ms	2.2421 KOps/s	2.2555 KOps/s	$\color{#d91a1a}-0.60\%$
test_seq_wrap[compile]	0.4305ms	0.2338ms	4.2779 KOps/s	4.3815 KOps/s	$\color{#d91a1a}-2.36\%$
test_seq_wrap[compile-overhead]	0.4514ms	0.2298ms	4.3520 KOps/s	4.4140 KOps/s	$\color{#d91a1a}-1.41\%$
test_func_call_runtime[False-eager]	0.7527ms	0.5366ms	1.8636 KOps/s	1.8333 KOps/s	$\color{#35bf28}+1.65\%$
test_func_call_runtime[False-compile]	0.5471ms	0.4287ms	2.3327 KOps/s	2.3219 KOps/s	$\color{#35bf28}+0.47\%$
test_func_call_runtime[False-compile-overhead]	0.5294ms	0.4272ms	2.3411 KOps/s	2.3354 KOps/s	$\color{#35bf28}+0.24\%$
test_func_call_runtime[True-eager]	1.0641ms	0.7428ms	1.3462 KOps/s	1.3357 KOps/s	$\color{#35bf28}+0.79\%$
test_func_call_runtime[True-compile]	0.9286ms	0.4720ms	2.1184 KOps/s	2.1266 KOps/s	$\color{#d91a1a}-0.38\%$
test_func_call_runtime[True-compile-overhead]	0.6170ms	0.4729ms	2.1145 KOps/s	2.1454 KOps/s	$\color{#d91a1a}-1.44\%$
test_func_call_cm_runtime[False-eager]	0.9017ms	0.5363ms	1.8647 KOps/s	1.8590 KOps/s	$\color{#35bf28}+0.30\%$
test_func_call_cm_runtime[False-compile]	0.7954ms	0.4316ms	2.3168 KOps/s	2.3620 KOps/s	$\color{#d91a1a}-1.91\%$
test_func_call_cm_runtime[False-compile-overhead]	0.9542ms	0.4297ms	2.3273 KOps/s	2.3343 KOps/s	$\color{#d91a1a}-0.30\%$
test_func_call_cm_runtime[True-eager]	1.2443ms	0.8888ms	1.1251 KOps/s	1.1110 KOps/s	$\color{#35bf28}+1.26\%$
test_func_call_cm_runtime[True-compile]	0.8576ms	0.4955ms	2.0183 KOps/s	2.0236 KOps/s	$\color{#d91a1a}-0.26\%$
test_func_call_cm_runtime[True-compile-overhead]	0.8277ms	0.4981ms	2.0078 KOps/s	2.0226 KOps/s	$\color{#d91a1a}-0.73\%$
test_vmap_func_call_cm_runtime[eager]	2.6871ms	1.9067ms	524.4755 Ops/s	521.4270 Ops/s	$\color{#35bf28}+0.58\%$
test_vmap_func_call_cm_runtime[compile]	0.7842ms	0.5170ms	1.9344 KOps/s	1.9137 KOps/s	$\color{#35bf28}+1.08\%$
test_vmap_func_call_cm_runtime[compile-overhead]	0.9201ms	0.5216ms	1.9171 KOps/s	1.9022 KOps/s	$\color{#35bf28}+0.79\%$
test_distributed	0.2653ms	0.1259ms	7.9407 KOps/s	7.5138 KOps/s	$\textbf{\color{#35bf28}+5.68\%}$
test_tdmodule	85.9200μs	26.4485μs	37.8093 KOps/s	39.5222 KOps/s	$\color{#d91a1a}-4.33\%$
test_tdmodule_dispatch	92.7530μs	50.2864μs	19.8861 KOps/s	22.1510 KOps/s	$\textbf{\color{#d91a1a}-10.22\%}$
test_tdseq	59.5510μs	30.5913μs	32.6891 KOps/s	37.0589 KOps/s	$\textbf{\color{#d91a1a}-11.79\%}$
test_tdseq_dispatch	88.8460μs	56.8898μs	17.5778 KOps/s	19.7436 KOps/s	$\textbf{\color{#d91a1a}-10.97\%}$
test_instantiation_functorch	2.1888ms	1.5079ms	663.1709 Ops/s	647.3114 Ops/s	$\color{#35bf28}+2.45\%$
test_exec_functorch	0.3175ms	0.1816ms	5.5053 KOps/s	5.5630 KOps/s	$\color{#d91a1a}-1.04\%$
test_exec_functional_call	0.3180ms	0.1723ms	5.8034 KOps/s	5.7565 KOps/s	$\color{#35bf28}+0.81\%$
test_exec_td_decorator	0.4848ms	0.2322ms	4.3062 KOps/s	4.2154 KOps/s	$\color{#35bf28}+2.15\%$
test_vmap_mlp_speed_decorator[True-True]	0.8368ms	0.6453ms	1.5496 KOps/s	1.5512 KOps/s	$\color{#d91a1a}-0.10\%$
test_vmap_mlp_speed_decorator[True-False]	0.9173ms	0.6416ms	1.5585 KOps/s	1.5488 KOps/s	$\color{#35bf28}+0.63\%$
test_vmap_mlp_speed_decorator[False-True]	0.7814ms	0.5217ms	1.9167 KOps/s	1.8756 KOps/s	$\color{#35bf28}+2.19\%$
test_vmap_mlp_speed_decorator[False-False]	0.9348ms	0.5253ms	1.9038 KOps/s	1.8893 KOps/s	$\color{#35bf28}+0.77\%$
test_to_module_speed[True]	2.2186ms	1.3209ms	757.0715 Ops/s	743.4904 Ops/s	$\color{#35bf28}+1.83\%$
test_to_module_speed[False]	1.9408ms	1.2860ms	777.5877 Ops/s	769.1419 Ops/s	$\color{#35bf28}+1.10\%$
test_tc_init	89.0460μs	47.4220μs	21.0872 KOps/s	21.8572 KOps/s	$\color{#d91a1a}-3.52\%$
test_tc_init_nested	0.2160ms	95.6714μs	10.4524 KOps/s	10.8073 KOps/s	$\color{#d91a1a}-3.28\%$
test_tc_first_layer_tensor	17.1020μs	1.5335μs	652.1198 KOps/s	643.0244 KOps/s	$\color{#35bf28}+1.41\%$
test_tc_first_layer_nontensor	23.7950μs	4.6349μs	215.7541 KOps/s	212.6060 KOps/s	$\color{#35bf28}+1.48\%$
test_tc_second_layer_tensor	41.2370μs	2.8009μs	357.0219 KOps/s	348.7563 KOps/s	$\color{#35bf28}+2.37\%$
test_tc_second_layer_nontensor	54.4620μs	5.9851μs	167.0813 KOps/s	164.5807 KOps/s	$\color{#35bf28}+1.52\%$
test_unbind	0.2401s	14.1453ms	70.6950 Ops/s	77.6404 Ops/s	$\textbf{\color{#d91a1a}-8.95\%}$
test_full_like	12.8849ms	9.1157ms	109.7006 Ops/s	80.5217 Ops/s	$\textbf{\color{#35bf28}+36.24\%}$
test_zeros_like	3.6105ms	3.0884ms	323.7962 Ops/s	127.2016 Ops/s	$\textbf{\color{#35bf28}+154.55\%}$
test_ones_like	4.0997ms	3.4906ms	286.4838 Ops/s	124.5687 Ops/s	$\textbf{\color{#35bf28}+129.98\%}$
test_clone	6.1469ms	5.4104ms	184.8289 Ops/s	105.2500 Ops/s	$\textbf{\color{#35bf28}+75.61\%}$
test_squeeze	65.3420μs	12.1372μs	82.3914 KOps/s	82.3183 KOps/s	$\color{#35bf28}+0.09\%$
test_unsqueeze	0.1714ms	90.3071μs	11.0733 KOps/s	11.0308 KOps/s	$\color{#35bf28}+0.39\%$
test_split	0.4723ms	0.1947ms	5.1362 KOps/s	5.1149 KOps/s	$\color{#35bf28}+0.42\%$
test_permute	0.5393ms	0.1944ms	5.1430 KOps/s	4.8504 KOps/s	$\textbf{\color{#35bf28}+6.03\%}$
test_stack	31.4204ms	25.9278ms	38.5686 Ops/s	39.6751 Ops/s	$\color{#d91a1a}-2.79\%$
test_cat	31.6721ms	25.6890ms	38.9271 Ops/s	39.5518 Ops/s	$\color{#d91a1a}-1.58\%$

[github-actions]

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

Total Benchmarks: 229. Improved: $\large\color{#35bf28}17$. Worsened: $\large\color{#d91a1a}10$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	28.8010μs	12.6360μs	79.1390 KOps/s	79.4643 KOps/s	$\color{#d91a1a}-0.41\%$
test_plain_set_stack_nested	40.5000μs	12.8580μs	77.7728 KOps/s	78.6394 KOps/s	$\color{#d91a1a}-1.10\%$
test_plain_set_nested_inplace	0.4025ms	13.7000μs	72.9926 KOps/s	73.2416 KOps/s	$\color{#d91a1a}-0.34\%$
test_plain_set_stack_nested_inplace	0.2016ms	13.7940μs	72.4955 KOps/s	73.1424 KOps/s	$\color{#d91a1a}-0.88\%$
test_items	45.5410μs	2.9214μs	342.3045 KOps/s	339.9020 KOps/s	$\color{#35bf28}+0.71\%$
test_items_nested	0.7671ms	0.3610ms	2.7700 KOps/s	2.7465 KOps/s	$\color{#35bf28}+0.86\%$
test_items_nested_locked	0.7598ms	0.3644ms	2.7439 KOps/s	2.7350 KOps/s	$\color{#35bf28}+0.33\%$
test_items_nested_leaf	0.4767ms	58.3399μs	17.1409 KOps/s	17.2987 KOps/s	$\color{#d91a1a}-0.91\%$
test_items_stack_nested	0.7748ms	0.3655ms	2.7358 KOps/s	2.7238 KOps/s	$\color{#35bf28}+0.44\%$
test_items_stack_nested_leaf	0.1031ms	60.1812μs	16.6165 KOps/s	16.7964 KOps/s	$\color{#d91a1a}-1.07\%$
test_items_stack_nested_locked	0.7714ms	0.3657ms	2.7341 KOps/s	2.7283 KOps/s	$\color{#35bf28}+0.21\%$
test_keys	0.4084ms	3.4814μs	287.2423 KOps/s	289.3488 KOps/s	$\color{#d91a1a}-0.73\%$
test_keys_nested	0.4827ms	87.7129μs	11.4008 KOps/s	11.3857 KOps/s	$\color{#35bf28}+0.13\%$
test_keys_nested_locked	0.7921ms	93.7056μs	10.6717 KOps/s	10.6989 KOps/s	$\color{#d91a1a}-0.25\%$
test_keys_nested_leaf	0.4738ms	79.2086μs	12.6249 KOps/s	12.9069 KOps/s	$\color{#d91a1a}-2.19\%$
test_keys_stack_nested	0.1221ms	89.1559μs	11.2163 KOps/s	11.2851 KOps/s	$\color{#d91a1a}-0.61\%$
test_keys_stack_nested_leaf	0.4974ms	80.6762μs	12.3952 KOps/s	12.6455 KOps/s	$\color{#d91a1a}-1.98\%$
test_keys_stack_nested_locked	0.5134ms	95.4641μs	10.4751 KOps/s	10.6555 KOps/s	$\color{#d91a1a}-1.69\%$
test_values	67.3895μs	0.8615μs	1.1607 MOps/s	1.1658 MOps/s	$\color{#d91a1a}-0.43\%$
test_values_nested	0.4364ms	37.6510μs	26.5597 KOps/s	26.9254 KOps/s	$\color{#d91a1a}-1.36\%$
test_values_nested_locked	0.4364ms	39.1383μs	25.5504 KOps/s	25.9511 KOps/s	$\color{#d91a1a}-1.54\%$
test_values_nested_leaf	0.4380ms	41.9690μs	23.8271 KOps/s	24.0856 KOps/s	$\color{#d91a1a}-1.07\%$
test_values_stack_nested	0.4498ms	38.2031μs	26.1759 KOps/s	26.2754 KOps/s	$\color{#d91a1a}-0.38\%$
test_values_stack_nested_leaf	81.9120μs	42.7058μs	23.4160 KOps/s	23.7881 KOps/s	$\color{#d91a1a}-1.56\%$
test_values_stack_nested_locked	0.4378ms	39.9701μs	25.0187 KOps/s	25.3387 KOps/s	$\color{#d91a1a}-1.26\%$
test_membership	20.7304μs	0.5117μs	1.9542 MOps/s	1.9668 MOps/s	$\color{#d91a1a}-0.64\%$
test_membership_nested	0.2119ms	1.9704μs	507.5000 KOps/s	492.8007 KOps/s	$\color{#35bf28}+2.98\%$
test_membership_nested_leaf	0.2154ms	1.9876μs	503.1275 KOps/s	509.0535 KOps/s	$\color{#d91a1a}-1.16\%$
test_membership_stacked_nested	20.9100μs	2.0272μs	493.2829 KOps/s	486.6987 KOps/s	$\color{#35bf28}+1.35\%$
test_membership_stacked_nested_leaf	0.4079ms	2.0196μs	495.1362 KOps/s	491.5500 KOps/s	$\color{#35bf28}+0.73\%$
test_membership_nested_last	41.5600μs	3.0468μs	328.2143 KOps/s	323.9584 KOps/s	$\color{#35bf28}+1.31\%$
test_membership_nested_leaf_last	0.3996ms	3.0843μs	324.2175 KOps/s	327.2971 KOps/s	$\color{#d91a1a}-0.94\%$
test_membership_stacked_nested_last	37.4110μs	3.7814μs	264.4519 KOps/s	260.5690 KOps/s	$\color{#35bf28}+1.49\%$
test_membership_stacked_nested_leaf_last	44.8510μs	3.8310μs	261.0315 KOps/s	265.5028 KOps/s	$\color{#d91a1a}-1.68\%$
test_nested_getleaf	0.4277ms	6.1580μs	162.3916 KOps/s	162.8336 KOps/s	$\color{#d91a1a}-0.27\%$
test_nested_get	0.4083ms	5.8630μs	170.5600 KOps/s	171.6812 KOps/s	$\color{#d91a1a}-0.65\%$
test_stacked_getleaf	49.1610μs	6.1631μs	162.2555 KOps/s	163.8031 KOps/s	$\color{#d91a1a}-0.94\%$
test_stacked_get	33.0210μs	5.8258μs	171.6494 KOps/s	173.6668 KOps/s	$\color{#d91a1a}-1.16\%$
test_nested_getitemleaf	53.5010μs	6.3853μs	156.6103 KOps/s	156.9453 KOps/s	$\color{#d91a1a}-0.21\%$
test_nested_getitem	0.4268ms	6.0931μs	164.1214 KOps/s	164.9916 KOps/s	$\color{#d91a1a}-0.53\%$
test_stacked_getitemleaf	33.6510μs	6.4240μs	155.6661 KOps/s	156.7016 KOps/s	$\color{#d91a1a}-0.66\%$
test_stacked_getitem	0.4305ms	6.1217μs	163.3527 KOps/s	163.1128 KOps/s	$\color{#35bf28}+0.15\%$
test_lock_nested	0.7787ms	0.3695ms	2.7061 KOps/s	2.6529 KOps/s	$\color{#35bf28}+2.01\%$
test_lock_stack_nested	0.4156ms	0.3387ms	2.9528 KOps/s	2.9711 KOps/s	$\color{#d91a1a}-0.62\%$
test_unlock_nested	0.7001ms	0.3120ms	3.2054 KOps/s	3.2222 KOps/s	$\color{#d91a1a}-0.52\%$
test_unlock_stack_nested	0.3961ms	0.2791ms	3.5826 KOps/s	3.6163 KOps/s	$\color{#d91a1a}-0.93\%$
test_flatten_speed	0.4957ms	74.8266μs	13.3642 KOps/s	13.4923 KOps/s	$\color{#d91a1a}-0.95\%$
test_unflatten_speed	0.7295ms	0.3171ms	3.1535 KOps/s	3.1560 KOps/s	$\color{#d91a1a}-0.08\%$
test_common_ops	1.6485ms	0.6364ms	1.5713 KOps/s	1.6045 KOps/s	$\color{#d91a1a}-2.07\%$
test_creation	0.1054ms	1.7302μs	577.9818 KOps/s	578.7302 KOps/s	$\color{#d91a1a}-0.13\%$
test_creation_empty	43.9700μs	9.0000μs	111.1116 KOps/s	111.7880 KOps/s	$\color{#d91a1a}-0.61\%$
test_creation_nested_1	35.9500μs	10.6815μs	93.6201 KOps/s	93.7699 KOps/s	$\color{#d91a1a}-0.16\%$
test_creation_nested_2	55.6410μs	13.4337μs	74.4396 KOps/s	74.7594 KOps/s	$\color{#d91a1a}-0.43\%$
test_clone	95.7110μs	10.8261μs	92.3696 KOps/s	93.6198 KOps/s	$\color{#d91a1a}-1.34\%$
test_getitem[int]	1.4978ms	10.5823μs	94.4972 KOps/s	94.3335 KOps/s	$\color{#35bf28}+0.17\%$
test_getitem[slice_int]	0.1058ms	20.8162μs	48.0394 KOps/s	48.8339 KOps/s	$\color{#d91a1a}-1.63\%$
test_getitem[range]	0.1526ms	37.6808μs	26.5387 KOps/s	27.1785 KOps/s	$\color{#d91a1a}-2.35\%$
test_getitem[tuple]	0.1120ms	18.0116μs	55.5197 KOps/s	55.7979 KOps/s	$\color{#d91a1a}-0.50\%$
test_getitem[list]	0.2467ms	34.0303μs	29.3856 KOps/s	30.6461 KOps/s	$\color{#d91a1a}-4.11\%$
test_setitem_dim[int]	39.0510μs	19.8906μs	50.2749 KOps/s	52.9328 KOps/s	$\textbf{\color{#d91a1a}-5.02\%}$
test_setitem_dim[slice_int]	86.1810μs	38.9423μs	25.6790 KOps/s	26.3875 KOps/s	$\color{#d91a1a}-2.69\%$
test_setitem_dim[range]	79.8720μs	53.2165μs	18.7912 KOps/s	19.0993 KOps/s	$\color{#d91a1a}-1.61\%$
test_setitem_dim[tuple]	54.7810μs	32.6934μs	30.5872 KOps/s	31.3968 KOps/s	$\color{#d91a1a}-2.58\%$
test_setitem	0.2072ms	15.6575μs	63.8673 KOps/s	63.9478 KOps/s	$\color{#d91a1a}-0.13\%$
test_set	0.1008ms	15.3735μs	65.0470 KOps/s	65.7743 KOps/s	$\color{#d91a1a}-1.11\%$
test_set_shared	1.5221ms	0.1515ms	6.6026 KOps/s	6.6049 KOps/s	$\color{#d91a1a}-0.04\%$
test_update	1.0424ms	18.4463μs	54.2114 KOps/s	50.9721 KOps/s	$\textbf{\color{#35bf28}+6.35\%}$
test_update_nested	96.9620μs	23.8896μs	41.8592 KOps/s	41.2168 KOps/s	$\color{#35bf28}+1.56\%$
test_update__nested	0.1296ms	25.7101μs	38.8951 KOps/s	39.5396 KOps/s	$\color{#d91a1a}-1.63\%$
test_set_nested	97.6210μs	16.0920μs	62.1427 KOps/s	60.5397 KOps/s	$\color{#35bf28}+2.65\%$
test_set_nested_new	93.9920μs	19.1217μs	52.2966 KOps/s	52.5143 KOps/s	$\color{#d91a1a}-0.41\%$
test_select	0.1183ms	31.0543μs	32.2016 KOps/s	32.3276 KOps/s	$\color{#d91a1a}-0.39\%$
test_select_nested	88.4320μs	43.6341μs	22.9178 KOps/s	22.9291 KOps/s	$\color{#d91a1a}-0.05\%$
test_exclude_nested	0.1065ms	63.0735μs	15.8545 KOps/s	16.3064 KOps/s	$\color{#d91a1a}-2.77\%$
test_empty[True]	0.4329ms	0.3006ms	3.3265 KOps/s	3.3690 KOps/s	$\color{#d91a1a}-1.26\%$
test_empty[False]	39.9987μs	0.8108μs	1.2334 MOps/s	1.2122 MOps/s	$\color{#35bf28}+1.75\%$
test_to	86.4720μs	54.6660μs	18.2929 KOps/s	16.9875 KOps/s	$\textbf{\color{#35bf28}+7.68\%}$
test_to_nonblocking	0.1991ms	47.3760μs	21.1077 KOps/s	20.0496 KOps/s	$\textbf{\color{#35bf28}+5.28\%}$
test_unbind_speed	1.7436ms	0.2325ms	4.3004 KOps/s	4.3092 KOps/s	$\color{#d91a1a}-0.20\%$
test_unbind_speed_stack0	0.3706ms	0.2338ms	4.2779 KOps/s	4.3236 KOps/s	$\color{#d91a1a}-1.06\%$
test_unbind_speed_stack1	93.1921ms	0.6576ms	1.5207 KOps/s	1.5175 KOps/s	$\color{#35bf28}+0.21\%$
test_split	95.6654ms	1.5887ms	629.4518 Ops/s	583.1948 Ops/s	$\textbf{\color{#35bf28}+7.93\%}$
test_chunk	95.2452ms	1.5826ms	631.8824 Ops/s	695.6165 Ops/s	$\textbf{\color{#d91a1a}-9.16\%}$
test_consolidate[False-None]	97.2070ms	2.9156ms	342.9864 Ops/s	340.8040 Ops/s	$\color{#35bf28}+0.64\%$
test_consolidate[default-None]	1.8280ms	1.6674ms	599.7328 Ops/s	601.2249 Ops/s	$\color{#d91a1a}-0.25\%$
test_consolidate[reduce-overhead-None]	1.8853ms	1.7064ms	586.0240 Ops/s	588.6689 Ops/s	$\color{#d91a1a}-0.45\%$
test_consolidate_njt[False-None]	6.5827ms	6.4777ms	154.3758 Ops/s	154.0551 Ops/s	$\color{#35bf28}+0.21\%$
test_to[False-False-None]	1.8404ms	1.6891ms	592.0373 Ops/s	588.5170 Ops/s	$\color{#35bf28}+0.60\%$
test_to[True-False-None]	1.7702ms	1.3313ms	751.1506 Ops/s	772.8905 Ops/s	$\color{#d91a1a}-2.81\%$
test_to[within-False-None]	4.2128ms	4.0807ms	245.0550 Ops/s	243.3478 Ops/s	$\color{#35bf28}+0.70\%$
test_to[True-default-None]	5.4571ms	5.3122ms	188.2474 Ops/s	189.7531 Ops/s	$\color{#d91a1a}-0.79\%$
test_to_njt[False-False-None]	6.9885ms	6.8366ms	146.2717 Ops/s	144.2820 Ops/s	$\color{#35bf28}+1.38\%$
test_to_njt[True-False-None]	5.6663ms	5.4499ms	183.4908 Ops/s	184.7188 Ops/s	$\color{#d91a1a}-0.66\%$
test_to_njt[within-False-None]	12.2051ms	12.0071ms	83.2844 Ops/s	82.8790 Ops/s	$\color{#35bf28}+0.49\%$
test_creation[device0]	0.3745ms	79.7022μs	12.5467 KOps/s	11.9475 KOps/s	$\textbf{\color{#35bf28}+5.02\%}$
test_creation_from_tensor	0.5421ms	83.4786μs	11.9791 KOps/s	11.5037 KOps/s	$\color{#35bf28}+4.13\%$
test_add_one[memmap_tensor0]	0.4343ms	6.8663μs	145.6382 KOps/s	148.7211 KOps/s	$\color{#d91a1a}-2.07\%$
test_contiguous[memmap_tensor0]	1.7840μs	0.4043μs	2.4733 MOps/s	2.4972 MOps/s	$\color{#d91a1a}-0.96\%$
test_stack[memmap_tensor0]	37.0510μs	4.3499μs	229.8908 KOps/s	237.8936 KOps/s	$\color{#d91a1a}-3.36\%$
test_memmaptd_index	1.6517ms	0.2542ms	3.9341 KOps/s	4.1236 KOps/s	$\color{#d91a1a}-4.60\%$
test_memmaptd_index_astensor	0.9365ms	0.3157ms	3.1676 KOps/s	3.2712 KOps/s	$\color{#d91a1a}-3.17\%$
test_memmaptd_index_op	1.0024ms	0.6031ms	1.6580 KOps/s	1.7271 KOps/s	$\color{#d91a1a}-4.00\%$
test_serialize_model	0.1312s	0.1302s	7.6816 Ops/s	7.6507 Ops/s	$\color{#35bf28}+0.40\%$
test_serialize_model_pickle	1.3504s	1.1924s	0.8386 Ops/s	0.8225 Ops/s	$\color{#35bf28}+1.97\%$
test_serialize_weights	0.1322s	0.1303s	7.6763 Ops/s	7.6865 Ops/s	$\color{#d91a1a}-0.13\%$
test_serialize_weights_returnearly	0.5040s	71.9870ms	13.8914 Ops/s	14.6701 Ops/s	$\textbf{\color{#d91a1a}-5.31\%}$
test_serialize_weights_pickle	1.3771s	1.1946s	0.8371 Ops/s	0.8199 Ops/s	$\color{#35bf28}+2.09\%$
test_reshape_pytree	0.1127ms	21.8656μs	45.7339 KOps/s	45.1356 KOps/s	$\color{#35bf28}+1.33\%$
test_reshape_td	48.5900μs	25.8960μs	38.6160 KOps/s	36.4544 KOps/s	$\textbf{\color{#35bf28}+5.93\%}$
test_view_pytree	0.1719ms	21.8560μs	45.7540 KOps/s	46.2352 KOps/s	$\color{#d91a1a}-1.04\%$
test_view_td	0.1576ms	30.2989μs	33.0045 KOps/s	30.9353 KOps/s	$\textbf{\color{#35bf28}+6.69\%}$
test_unbind_pytree	0.1339ms	27.6445μs	36.1736 KOps/s	36.4197 KOps/s	$\color{#d91a1a}-0.68\%$
test_unbind_td	0.7664ms	36.1885μs	27.6331 KOps/s	27.6710 KOps/s	$\color{#d91a1a}-0.14\%$
test_split_pytree	0.1495ms	29.7568μs	33.6058 KOps/s	34.0359 KOps/s	$\color{#d91a1a}-1.26\%$
test_split_td	0.9804ms	38.1604μs	26.2052 KOps/s	25.7727 KOps/s	$\color{#35bf28}+1.68\%$
test_add_pytree	0.1079ms	34.6243μs	28.8815 KOps/s	28.9885 KOps/s	$\color{#d91a1a}-0.37\%$
test_add_td	88.2720μs	49.9785μs	20.0086 KOps/s	19.8795 KOps/s	$\color{#35bf28}+0.65\%$
test_compile_add_one_nested[tensordict-compile]	0.1705ms	0.1193ms	8.3839 KOps/s	8.0859 KOps/s	$\color{#35bf28}+3.69\%$
test_compile_add_one_nested[tensordict-eager]	0.2739ms	0.1312ms	7.6206 KOps/s	7.5067 KOps/s	$\color{#35bf28}+1.52\%$
test_compile_add_one_nested[pytree-compile]	0.2377ms	94.9048μs	10.5369 KOps/s	10.4138 KOps/s	$\color{#35bf28}+1.18\%$
test_compile_add_one_nested[pytree-eager]	0.3348ms	0.1474ms	6.7853 KOps/s	6.6847 KOps/s	$\color{#35bf28}+1.50\%$
test_compile_copy_nested[tensordict-compile]	0.1859ms	24.8034μs	40.3170 KOps/s	41.0853 KOps/s	$\color{#d91a1a}-1.87\%$
test_compile_copy_nested[tensordict-eager]	65.9910μs	29.1124μs	34.3496 KOps/s	34.0429 KOps/s	$\color{#35bf28}+0.90\%$
test_compile_copy_nested[pytree-compile]	0.4405ms	63.6862μs	15.7020 KOps/s	15.3392 KOps/s	$\color{#35bf28}+2.37\%$
test_compile_copy_nested[pytree-eager]	87.1720μs	48.7590μs	20.5090 KOps/s	20.1237 KOps/s	$\color{#35bf28}+1.91\%$
test_compile_add_one_flat[tensordict-compile]	0.1806ms	0.1402ms	7.1333 KOps/s	6.9994 KOps/s	$\color{#35bf28}+1.91\%$
test_compile_add_one_flat[tensordict-eager]	0.3496ms	0.2160ms	4.6289 KOps/s	4.6042 KOps/s	$\color{#35bf28}+0.54\%$
test_compile_add_one_flat[tensorclass-compile]	0.2413ms	96.9231μs	10.3175 KOps/s	10.3134 KOps/s	$\color{#35bf28}+0.04\%$
test_compile_add_one_flat[tensorclass-eager]	0.1973ms	54.5533μs	18.3307 KOps/s	18.3919 KOps/s	$\color{#d91a1a}-0.33\%$
test_compile_add_one_flat[pytree-compile]	0.1737ms	0.1344ms	7.4402 KOps/s	7.4327 KOps/s	$\color{#35bf28}+0.10\%$
test_compile_add_one_flat[pytree-eager]	0.6278ms	0.4731ms	2.1137 KOps/s	2.0922 KOps/s	$\color{#35bf28}+1.03\%$
test_compile_add_self_flat[tensordict-eager]	0.4086ms	0.2603ms	3.8421 KOps/s	3.8423 KOps/s	$-0.00\%$
test_compile_add_self_flat[tensordict-compile]	0.2586ms	0.1413ms	7.0788 KOps/s	7.0932 KOps/s	$\color{#d91a1a}-0.20\%$
test_compile_add_self_flat[tensorclass-eager]	0.2249ms	66.2841μs	15.0866 KOps/s	14.9813 KOps/s	$\color{#35bf28}+0.70\%$
test_compile_add_self_flat[tensorclass-compile]	0.1520ms	99.7629μs	10.0238 KOps/s	10.1902 KOps/s	$\color{#d91a1a}-1.63\%$
test_compile_add_self_flat[pytree-eager]	0.5449ms	0.4038ms	2.4762 KOps/s	2.4854 KOps/s	$\color{#d91a1a}-0.37\%$
test_compile_add_self_flat[pytree-compile]	0.1733ms	0.1345ms	7.4327 KOps/s	7.5231 KOps/s	$\color{#d91a1a}-1.20\%$
test_compile_copy_flat[tensordict-compile]	87.5010μs	19.4273μs	51.4740 KOps/s	54.8891 KOps/s	$\textbf{\color{#d91a1a}-6.22\%}$
test_compile_copy_flat[tensordict-eager]	60.4210μs	31.2913μs	31.9578 KOps/s	31.4302 KOps/s	$\color{#35bf28}+1.68\%$
test_compile_copy_flat[pytree-compile]	0.1071ms	70.0123μs	14.2832 KOps/s	14.3504 KOps/s	$\color{#d91a1a}-0.47\%$
test_compile_copy_flat[pytree-eager]	0.1551ms	50.5605μs	19.7783 KOps/s	19.4746 KOps/s	$\color{#35bf28}+1.56\%$
test_compile_assign_and_add[tensordict-compile]	1.5853ms	0.3896ms	2.5670 KOps/s	2.2511 KOps/s	$\textbf{\color{#35bf28}+14.03\%}$
test_compile_assign_and_add[tensordict-eager]	2.7948ms	2.5729ms	388.6630 Ops/s	382.9095 Ops/s	$\color{#35bf28}+1.50\%$
test_compile_assign_and_add[pytree-compile]	1.5779ms	0.4376ms	2.2851 KOps/s	2.2970 KOps/s	$\color{#d91a1a}-0.52\%$
test_compile_assign_and_add[pytree-eager]	2.8266ms	2.5815ms	387.3773 Ops/s	386.6520 Ops/s	$\color{#35bf28}+0.19\%$
test_compile_indexing[tensor-tensordict-compile]	0.4861ms	0.1179ms	8.4805 KOps/s	8.7881 KOps/s	$\color{#d91a1a}-3.50\%$
test_compile_indexing[tensor-tensordict-eager]	0.6014ms	80.7422μs	12.3851 KOps/s	12.4136 KOps/s	$\color{#d91a1a}-0.23\%$
test_compile_indexing[tensor-tensorclass-compile]	0.5126ms	0.1102ms	9.0715 KOps/s	9.6022 KOps/s	$\textbf{\color{#d91a1a}-5.53\%}$
test_compile_indexing[tensor-tensorclass-eager]	2.7094ms	71.8635μs	13.9153 KOps/s	15.0180 KOps/s	$\textbf{\color{#d91a1a}-7.34\%}$
test_compile_indexing[tensor-pytree-compile]	0.1597ms	0.1101ms	9.0827 KOps/s	9.5506 KOps/s	$\color{#d91a1a}-4.90\%$
test_compile_indexing[tensor-pytree-eager]	0.2410ms	71.7396μs	13.9393 KOps/s	14.9660 KOps/s	$\textbf{\color{#d91a1a}-6.86\%}$
test_compile_indexing[slice-tensordict-compile]	0.2434ms	0.1004ms	9.9632 KOps/s	10.1621 KOps/s	$\color{#d91a1a}-1.96\%$
test_compile_indexing[slice-tensordict-eager]	0.1408ms	16.9706μs	58.9255 KOps/s	58.6836 KOps/s	$\color{#35bf28}+0.41\%$
test_compile_indexing[slice-tensorclass-compile]	0.2640ms	98.9054μs	10.1107 KOps/s	10.5874 KOps/s	$\color{#d91a1a}-4.50\%$
test_compile_indexing[slice-tensorclass-eager]	0.1647ms	16.4262μs	60.8784 KOps/s	64.7708 KOps/s	$\textbf{\color{#d91a1a}-6.01\%}$
test_compile_indexing[slice-pytree-compile]	0.2850ms	98.6066μs	10.1413 KOps/s	10.5658 KOps/s	$\color{#d91a1a}-4.02\%$
test_compile_indexing[slice-pytree-eager]	0.1126ms	15.8216μs	63.2046 KOps/s	64.9604 KOps/s	$\color{#d91a1a}-2.70\%$
test_compile_indexing[int-tensordict-compile]	0.2804ms	0.1026ms	9.7506 KOps/s	10.0463 KOps/s	$\color{#d91a1a}-2.94\%$
test_compile_indexing[int-tensordict-eager]	0.5518ms	16.7304μs	59.7715 KOps/s	59.1362 KOps/s	$\color{#35bf28}+1.07\%$
test_compile_indexing[int-tensorclass-compile]	0.2520ms	96.1915μs	10.3959 KOps/s	10.4995 KOps/s	$\color{#d91a1a}-0.99\%$
test_compile_indexing[int-tensorclass-eager]	96.3020μs	15.7474μs	63.5027 KOps/s	65.5357 KOps/s	$\color{#d91a1a}-3.10\%$
test_compile_indexing[int-pytree-compile]	0.2233ms	95.9462μs	10.4225 KOps/s	10.5175 KOps/s	$\color{#d91a1a}-0.90\%$
test_compile_indexing[int-pytree-eager]	92.8520μs	15.9271μs	62.7862 KOps/s	65.4563 KOps/s	$\color{#d91a1a}-4.08\%$
test_mod_add[eager]	0.1873ms	37.5566μs	26.6265 KOps/s	26.0395 KOps/s	$\color{#35bf28}+2.25\%$
test_mod_add[compile]	0.3390ms	78.4650μs	12.7445 KOps/s	12.5689 KOps/s	$\color{#35bf28}+1.40\%$
test_mod_add[compile-overhead]	0.3216ms	0.1634ms	6.1183 KOps/s	5.7544 KOps/s	$\textbf{\color{#35bf28}+6.33\%}$
test_mod_wrap[eager]	0.3778ms	0.2453ms	4.0772 KOps/s	3.9936 KOps/s	$\color{#35bf28}+2.09\%$
test_mod_wrap[compile]	0.3974ms	0.2787ms	3.5879 KOps/s	3.5583 KOps/s	$\color{#35bf28}+0.83\%$
test_mod_wrap[compile-overhead]	6.9500ms	3.7215ms	268.7102 Ops/s	265.9146 Ops/s	$\color{#35bf28}+1.05\%$
test_mod_wrap_and_backward[eager]	1.5921ms	1.3996ms	714.4700 Ops/s	675.9173 Ops/s	$\textbf{\color{#35bf28}+5.70\%}$
test_mod_wrap_and_backward[compile]	1.4128ms	1.2497ms	800.1712 Ops/s	730.0133 Ops/s	$\textbf{\color{#35bf28}+9.61\%}$
test_mod_wrap_and_backward[compile-overhead]	1.4289ms	0.9250ms	1.0811 KOps/s	954.4387 Ops/s	$\textbf{\color{#35bf28}+13.27\%}$
test_seq_add[eager]	0.3274ms	0.1185ms	8.4361 KOps/s	8.3344 KOps/s	$\color{#35bf28}+1.22\%$
test_seq_add[compile]	0.2448ms	87.9137μs	11.3748 KOps/s	11.0080 KOps/s	$\color{#35bf28}+3.33\%$
test_seq_add[compile-overhead]	0.3290ms	0.1273ms	7.8580 KOps/s	7.8081 KOps/s	$\color{#35bf28}+0.64\%$
test_seq_wrap[eager]	0.6148ms	0.4167ms	2.4000 KOps/s	2.3664 KOps/s	$\color{#35bf28}+1.42\%$
test_seq_wrap[compile]	0.4658ms	0.2959ms	3.3801 KOps/s	3.3650 KOps/s	$\color{#35bf28}+0.45\%$
test_seq_wrap[compile-overhead]	0.3636ms	0.2222ms	4.5005 KOps/s	4.3119 KOps/s	$\color{#35bf28}+4.37\%$
test_func_call_runtime[False-eager]	0.9297ms	0.7559ms	1.3229 KOps/s	1.2526 KOps/s	$\textbf{\color{#35bf28}+5.61\%}$
test_func_call_runtime[False-compile]	0.9456ms	0.7267ms	1.3761 KOps/s	1.3612 KOps/s	$\color{#35bf28}+1.09\%$
test_func_call_runtime[False-compile-overhead]	0.4165ms	0.3586ms	2.7889 KOps/s	2.7698 KOps/s	$\color{#35bf28}+0.69\%$
test_func_call_runtime[True-eager]	1.0391ms	0.9003ms	1.1107 KOps/s	1.1193 KOps/s	$\color{#d91a1a}-0.77\%$
test_func_call_runtime[True-compile]	0.9677ms	0.7552ms	1.3242 KOps/s	1.3310 KOps/s	$\color{#d91a1a}-0.51\%$
test_func_call_runtime[True-compile-overhead]	0.5184ms	0.3778ms	2.6469 KOps/s	2.6199 KOps/s	$\color{#35bf28}+1.03\%$
test_func_call_cm_runtime[False-eager]	0.8441ms	0.7175ms	1.3937 KOps/s	1.2641 KOps/s	$\textbf{\color{#35bf28}+10.25\%}$
test_func_call_cm_runtime[False-compile]	0.9000ms	0.7343ms	1.3619 KOps/s	1.2523 KOps/s	$\textbf{\color{#35bf28}+8.75\%}$
test_func_call_cm_runtime[False-compile-overhead]	0.5090ms	0.3622ms	2.7607 KOps/s	2.7358 KOps/s	$\color{#35bf28}+0.91\%$
test_func_call_cm_runtime[True-eager]	1.0861ms	0.9852ms	1.0150 KOps/s	934.9338 Ops/s	$\textbf{\color{#35bf28}+8.57\%}$
test_func_call_cm_runtime[True-compile]	0.9376ms	0.7807ms	1.2810 KOps/s	1.2645 KOps/s	$\color{#35bf28}+1.31\%$
test_func_call_cm_runtime[True-compile-overhead]	0.5052ms	0.4084ms	2.4488 KOps/s	2.4433 KOps/s	$\color{#35bf28}+0.23\%$
test_vmap_func_call_cm_runtime[eager]	2.6843ms	2.1460ms	465.9733 Ops/s	474.7593 Ops/s	$\color{#d91a1a}-1.85\%$
test_vmap_func_call_cm_runtime[compile]	1.0389ms	0.8502ms	1.1762 KOps/s	1.1508 KOps/s	$\color{#35bf28}+2.20\%$
test_vmap_func_call_cm_runtime[compile-overhead]	0.5343ms	0.4069ms	2.4573 KOps/s	2.4490 KOps/s	$\color{#35bf28}+0.34\%$
test_distributed	4.6210ms	0.2414ms	4.1428 KOps/s	8.1015 KOps/s	$\textbf{\color{#d91a1a}-48.86\%}$
test_tdmodule	57.5410μs	20.4778μs	48.8334 KOps/s	50.1350 KOps/s	$\color{#d91a1a}-2.60\%$
test_tdmodule_dispatch	78.1110μs	36.4521μs	27.4332 KOps/s	27.4947 KOps/s	$\color{#d91a1a}-0.22\%$
test_tdseq	58.5310μs	21.0201μs	47.5734 KOps/s	46.7700 KOps/s	$\color{#35bf28}+1.72\%$
test_tdseq_dispatch	69.6010μs	39.1402μs	25.5492 KOps/s	25.3246 KOps/s	$\color{#35bf28}+0.89\%$
test_instantiation_functorch	1.6474ms	1.5321ms	652.7131 Ops/s	657.8127 Ops/s	$\color{#d91a1a}-0.78\%$
test_exec_functorch	0.1996ms	0.1464ms	6.8286 KOps/s	7.0815 KOps/s	$\color{#d91a1a}-3.57\%$
test_exec_functional_call	0.2803ms	0.1403ms	7.1297 KOps/s	7.3007 KOps/s	$\color{#d91a1a}-2.34\%$
test_exec_td_decorator	0.3859ms	0.1869ms	5.3515 KOps/s	5.4101 KOps/s	$\color{#d91a1a}-1.08\%$
test_vmap_mlp_speed_decorator[True-True]	0.8357ms	0.6811ms	1.4682 KOps/s	1.4658 KOps/s	$\color{#35bf28}+0.17\%$
test_vmap_mlp_speed_decorator[True-False]	0.8126ms	0.6826ms	1.4651 KOps/s	1.4634 KOps/s	$\color{#35bf28}+0.12\%$
test_vmap_mlp_speed_decorator[False-True]	0.7365ms	0.6180ms	1.6181 KOps/s	1.5969 KOps/s	$\color{#35bf28}+1.33\%$
test_vmap_mlp_speed_decorator[False-False]	0.7747ms	0.6206ms	1.6114 KOps/s	1.5894 KOps/s	$\color{#35bf28}+1.38\%$
test_vmap_transformer_speed_decorator[True-True]	19.1584ms	19.0321ms	52.5428 Ops/s	52.2756 Ops/s	$\color{#35bf28}+0.51\%$
test_vmap_transformer_speed_decorator[True-False]	19.1677ms	19.0325ms	52.5416 Ops/s	51.8572 Ops/s	$\color{#35bf28}+1.32\%$
test_vmap_transformer_speed_decorator[False-True]	19.7835ms	18.9602ms	52.7421 Ops/s	52.6647 Ops/s	$\color{#35bf28}+0.15\%$
test_vmap_transformer_speed_decorator[False-False]	19.3346ms	18.9270ms	52.8345 Ops/s	52.6136 Ops/s	$\color{#35bf28}+0.42\%$
test_to_module_speed[True]	1.6296ms	0.9638ms	1.0376 KOps/s	1.0408 KOps/s	$\color{#d91a1a}-0.31\%$
test_to_module_speed[False]	1.4790ms	0.9439ms	1.0594 KOps/s	1.0500 KOps/s	$\color{#35bf28}+0.90\%$
test_tc_init	74.2220μs	36.6187μs	27.3084 KOps/s	27.0592 KOps/s	$\color{#35bf28}+0.92\%$
test_tc_init_nested	0.2272ms	75.4294μs	13.2574 KOps/s	13.6972 KOps/s	$\color{#d91a1a}-3.21\%$
test_tc_first_layer_tensor	4.8900μs	0.6945μs	1.4400 MOps/s	1.2373 MOps/s	$\textbf{\color{#35bf28}+16.38\%}$
test_tc_first_layer_nontensor	37.2200μs	2.2313μs	448.1684 KOps/s	444.6312 KOps/s	$\color{#35bf28}+0.80\%$
test_tc_second_layer_tensor	7.6550μs	1.4085μs	709.9756 KOps/s	700.6774 KOps/s	$\color{#35bf28}+1.33\%$
test_tc_second_layer_nontensor	40.6800μs	2.9624μs	337.5594 KOps/s	334.0619 KOps/s	$\color{#35bf28}+1.05\%$
test_unbind	0.2292s	9.9001ms	101.0086 Ops/s	145.2471 Ops/s	$\textbf{\color{#d91a1a}-30.46\%}$
test_full_like	10.5238ms	9.4334ms	106.0061 Ops/s	105.5727 Ops/s	$\color{#35bf28}+0.41\%$
test_zeros_like	5.2722ms	4.3482ms	229.9820 Ops/s	233.6216 Ops/s	$\color{#d91a1a}-1.56\%$
test_ones_like	5.3862ms	4.2632ms	234.5671 Ops/s	232.0524 Ops/s	$\color{#35bf28}+1.08\%$
test_clone	7.0348ms	6.5907ms	151.7279 Ops/s	150.9094 Ops/s	$\color{#35bf28}+0.54\%$
test_squeeze	0.1546ms	9.8031μs	102.0086 KOps/s	104.6865 KOps/s	$\color{#d91a1a}-2.56\%$
test_unsqueeze	0.2628ms	73.5846μs	13.5898 KOps/s	13.3166 KOps/s	$\color{#35bf28}+2.05\%$
test_split	0.4078ms	0.1693ms	5.9060 KOps/s	6.1728 KOps/s	$\color{#d91a1a}-4.32\%$
test_permute	0.3642ms	0.1858ms	5.3823 KOps/s	5.4279 KOps/s	$\color{#d91a1a}-0.84\%$
test_stack	51.6608ms	50.9426ms	19.6299 Ops/s	19.3348 Ops/s	$\color{#35bf28}+1.53\%$
test_cat	51.5917ms	50.7316ms	19.7116 Ops/s	19.5790 Ops/s	$\color{#35bf28}+0.68\%$