Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[Performance] Faster flatten_keys #727

Merged
merged 5 commits into from
Apr 18, 2024
Merged

[Performance] Faster flatten_keys #727

merged 5 commits into from
Apr 18, 2024

Conversation

vmoens
Copy link
Contributor

@vmoens vmoens commented Apr 17, 2024

No description provided.

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Apr 17, 2024
vmoens
vmoens commented Apr 18, 2024
View reviewed changes
tensordict/base.py Outdated Show resolved Hide resolved
vmoens added 2 commits April 18, 2024 09:49
@vmoens
Merge branch 'main' into faster-flatten
d9f742e 
# Conflicts:
#	tensordict/base.py
@vmoens
amend
aa97c44
@github-actions GitHub Actions
Copy link

github-actions bot commented Apr 18, 2024 •
edited
Loading

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

Total Benchmarks: 127. Improved: $\large\color{#35bf28}7$. Worsened: $\large\color{#d91a1a}29$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_plain_set_nested 41.2170μs 17.6586μs 56.6296 KOps/s 56.9657 KOps/s $\color{#d91a1a}-0.59\%$
test_plain_set_stack_nested 46.3370μs 17.8527μs 56.0138 KOps/s 56.4005 KOps/s $\color{#d91a1a}-0.69\%$
test_plain_set_nested_inplace 58.3590μs 20.1003μs 49.7504 KOps/s 50.7481 KOps/s $\color{#d91a1a}-1.97\%$
test_plain_set_stack_nested_inplace 77.9760μs 20.1047μs 49.7396 KOps/s 50.0075 KOps/s $\color{#d91a1a}-0.54\%$
test_items 37.1790μs 2.5262μs 395.8584 KOps/s 406.2550 KOps/s $\color{#d91a1a}-2.56\%$
test_items_nested 1.9384ms 0.2732ms 3.6605 KOps/s 3.6946 KOps/s $\color{#d91a1a}-0.92\%$
test_items_nested_locked 0.4537ms 0.2858ms 3.4992 KOps/s 3.6638 KOps/s $\color{#d91a1a}-4.49\%$
test_items_nested_leaf 0.2044ms 78.1419μs 12.7972 KOps/s 12.9290 KOps/s $\color{#d91a1a}-1.02\%$
test_items_stack_nested 0.3228ms 0.2752ms 3.6331 KOps/s 3.6339 KOps/s $\color{#d91a1a}-0.02\%$
test_items_stack_nested_leaf 0.1392ms 75.3112μs 13.2782 KOps/s 12.9736 KOps/s $\color{#35bf28}+2.35\%$
test_items_stack_nested_locked 0.4856ms 0.2761ms 3.6217 KOps/s 3.6027 KOps/s $\color{#35bf28}+0.53\%$
test_keys 34.9360μs 3.7930μs 263.6404 KOps/s 258.6239 KOps/s $\color{#35bf28}+1.94\%$
test_keys_nested 0.3266ms 0.1384ms 7.2235 KOps/s 6.9943 KOps/s $\color{#35bf28}+3.28\%$
test_keys_nested_locked 0.7898ms 0.1399ms 7.1501 KOps/s 6.8142 KOps/s $\color{#35bf28}+4.93\%$
test_keys_nested_leaf 0.2041ms 0.1145ms 8.7324 KOps/s 8.0811 KOps/s $\textbf{\color{#35bf28}+8.06\%}$
test_keys_stack_nested 0.2525ms 0.1338ms 7.4714 KOps/s 6.8932 KOps/s $\textbf{\color{#35bf28}+8.39\%}$
test_keys_stack_nested_leaf 0.2063ms 0.1133ms 8.8254 KOps/s 7.8947 KOps/s $\textbf{\color{#35bf28}+11.79\%}$
test_keys_stack_nested_locked 0.2696ms 0.1381ms 7.2419 KOps/s 6.6474 KOps/s $\textbf{\color{#35bf28}+8.94\%}$
test_values 23.9045μs 1.1595μs 862.4136 KOps/s 861.9713 KOps/s $\color{#35bf28}+0.05\%$
test_values_nested 0.1142ms 50.5380μs 19.7871 KOps/s 19.9309 KOps/s $\color{#d91a1a}-0.72\%$
test_values_nested_locked 0.1051ms 50.9054μs 19.6443 KOps/s 19.9264 KOps/s $\color{#d91a1a}-1.42\%$
test_values_nested_leaf 0.1107ms 45.5222μs 21.9673 KOps/s 22.0482 KOps/s $\color{#d91a1a}-0.37\%$
test_values_stack_nested 0.1237ms 51.9357μs 19.2546 KOps/s 19.4256 KOps/s $\color{#d91a1a}-0.88\%$
test_values_stack_nested_leaf 0.1002ms 45.3305μs 22.0602 KOps/s 22.4447 KOps/s $\color{#d91a1a}-1.71\%$
test_values_stack_nested_locked 0.1112ms 51.4457μs 19.4380 KOps/s 19.7086 KOps/s $\color{#d91a1a}-1.37\%$
test_membership 37.5910μs 1.3673μs 731.3905 KOps/s 777.3359 KOps/s $\textbf{\color{#d91a1a}-5.91\%}$
test_membership_nested 36.2370μs 3.4741μs 287.8451 KOps/s 296.5440 KOps/s $\color{#d91a1a}-2.93\%$
test_membership_nested_leaf 25.2770μs 3.4942μs 286.1871 KOps/s 295.7432 KOps/s $\color{#d91a1a}-3.23\%$
test_membership_stacked_nested 46.6370μs 3.4945μs 286.1603 KOps/s 298.2834 KOps/s $\color{#d91a1a}-4.06\%$
test_membership_stacked_nested_leaf 27.5720μs 3.4595μs 289.0576 KOps/s 293.0608 KOps/s $\color{#d91a1a}-1.37\%$
test_membership_nested_last 40.5050μs 4.2728μs 234.0398 KOps/s 239.5227 KOps/s $\color{#d91a1a}-2.29\%$
test_membership_nested_leaf_last 41.9290μs 4.2866μs 233.2855 KOps/s 233.8806 KOps/s $\color{#d91a1a}-0.25\%$
test_membership_stacked_nested_last 44.2730μs 13.4345μs 74.4350 KOps/s 172.0594 KOps/s $\textbf{\color{#d91a1a}-56.74\%}$
test_membership_stacked_nested_leaf_last 44.4430μs 13.4339μs 74.4387 KOps/s 172.3182 KOps/s $\textbf{\color{#d91a1a}-56.80\%}$
test_nested_getleaf 88.6050μs 10.8557μs 92.1171 KOps/s 94.3460 KOps/s $\color{#d91a1a}-2.36\%$
test_nested_get 40.7670μs 10.0976μs 99.0336 KOps/s 99.1298 KOps/s $\color{#d91a1a}-0.10\%$
test_stacked_getleaf 43.6710μs 10.6102μs 94.2488 KOps/s 94.6514 KOps/s $\color{#d91a1a}-0.43\%$
test_stacked_get 71.6840μs 10.1587μs 98.4373 KOps/s 99.9837 KOps/s $\color{#d91a1a}-1.55\%$
test_nested_getitemleaf 47.4790μs 11.2098μs 89.2076 KOps/s 90.1139 KOps/s $\color{#d91a1a}-1.01\%$
test_nested_getitem 47.3590μs 10.2827μs 97.2507 KOps/s 96.6618 KOps/s $\color{#35bf28}+0.61\%$
test_stacked_getitemleaf 41.9680μs 11.0440μs 90.5468 KOps/s 90.6824 KOps/s $\color{#d91a1a}-0.15\%$
test_stacked_getitem 74.2290μs 10.2327μs 97.7255 KOps/s 95.9078 KOps/s $\color{#35bf28}+1.90\%$
test_lock_nested 58.2529ms 0.4159ms 2.4041 KOps/s 2.9550 KOps/s $\textbf{\color{#d91a1a}-18.64\%}$
test_lock_stack_nested 0.4824ms 0.3030ms 3.3005 KOps/s 3.3253 KOps/s $\color{#d91a1a}-0.75\%$
test_unlock_nested 0.1019s 0.4595ms 2.1762 KOps/s 2.2504 KOps/s $\color{#d91a1a}-3.30\%$
test_unlock_stack_nested 0.5029ms 0.3111ms 3.2140 KOps/s 3.2450 KOps/s $\color{#d91a1a}-0.96\%$
test_flatten_speed 0.6082ms 91.8322μs 10.8894 KOps/s 10.9993 KOps/s $\color{#d91a1a}-1.00\%$
test_unflatten_speed 0.8223ms 0.4053ms 2.4670 KOps/s 2.4865 KOps/s $\color{#d91a1a}-0.78\%$
test_common_ops 4.5387ms 0.7487ms 1.3357 KOps/s 1.3871 KOps/s $\color{#d91a1a}-3.71\%$
test_creation 0.1153ms 1.8052μs 553.9597 KOps/s 554.8262 KOps/s $\color{#d91a1a}-0.16\%$
test_creation_empty 47.4980μs 11.4409μs 87.4056 KOps/s 87.3943 KOps/s $\color{#35bf28}+0.01\%$
test_creation_nested_1 48.5610μs 14.1545μs 70.6490 KOps/s 70.0839 KOps/s $\color{#35bf28}+0.81\%$
test_creation_nested_2 57.7380μs 17.7079μs 56.4721 KOps/s 58.1530 KOps/s $\color{#d91a1a}-2.89\%$
test_clone 0.1667ms 14.1572μs 70.6355 KOps/s 75.6108 KOps/s $\textbf{\color{#d91a1a}-6.58\%}$
test_getitem[int] 37.6200μs 11.8847μs 84.1420 KOps/s 89.8048 KOps/s $\textbf{\color{#d91a1a}-6.31\%}$
test_getitem[slice_int] 55.7250μs 22.9652μs 43.5441 KOps/s 46.8985 KOps/s $\textbf{\color{#d91a1a}-7.15\%}$
test_getitem[range] 0.2320ms 42.7814μs 23.3747 KOps/s 24.9482 KOps/s $\textbf{\color{#d91a1a}-6.31\%}$
test_getitem[tuple] 74.8400μs 18.9569μs 52.7512 KOps/s 55.5917 KOps/s $\textbf{\color{#d91a1a}-5.11\%}$
test_getitem[list] 0.3324ms 38.1470μs 26.2144 KOps/s 26.7849 KOps/s $\color{#d91a1a}-2.13\%$
test_setitem_dim[int] 86.5720μs 36.4853μs 27.4083 KOps/s 29.8857 KOps/s $\textbf{\color{#d91a1a}-8.29\%}$
test_setitem_dim[slice_int] 0.1116ms 62.7674μs 15.9318 KOps/s 16.8976 KOps/s $\textbf{\color{#d91a1a}-5.72\%}$
test_setitem_dim[range] 0.1407ms 81.7130μs 12.2380 KOps/s 13.0080 KOps/s $\textbf{\color{#d91a1a}-5.92\%}$
test_setitem_dim[tuple] 77.3650μs 51.7488μs 19.3241 KOps/s 20.7359 KOps/s $\textbf{\color{#d91a1a}-6.81\%}$
test_setitem 0.1325ms 22.0146μs 45.4244 KOps/s 49.5131 KOps/s $\textbf{\color{#d91a1a}-8.26\%}$
test_set 0.1777ms 21.4839μs 46.5466 KOps/s 50.2191 KOps/s $\textbf{\color{#d91a1a}-7.31\%}$
test_set_shared 2.2951ms 0.1513ms 6.6072 KOps/s 7.0634 KOps/s $\textbf{\color{#d91a1a}-6.46\%}$
test_update 0.1782ms 23.7635μs 42.0813 KOps/s 45.1465 KOps/s $\textbf{\color{#d91a1a}-6.79\%}$
test_update_nested 0.1903ms 32.3988μs 30.8653 KOps/s 33.2724 KOps/s $\textbf{\color{#d91a1a}-7.23\%}$
test_update__nested 0.1462ms 25.7958μs 38.7660 KOps/s 40.9142 KOps/s $\textbf{\color{#d91a1a}-5.25\%}$
test_set_nested 0.1736ms 23.7439μs 42.1161 KOps/s 46.0792 KOps/s $\textbf{\color{#d91a1a}-8.60\%}$
test_set_nested_new 0.1679ms 27.4375μs 36.4464 KOps/s 38.5667 KOps/s $\textbf{\color{#d91a1a}-5.50\%}$
test_select 0.1538ms 42.4735μs 23.5441 KOps/s 24.8289 KOps/s $\textbf{\color{#d91a1a}-5.17\%}$
test_select_nested 0.1247ms 60.5053μs 16.5275 KOps/s 17.1432 KOps/s $\color{#d91a1a}-3.59\%$
test_exclude_nested 0.2244ms 0.1215ms 8.2338 KOps/s 8.5277 KOps/s $\color{#d91a1a}-3.45\%$
test_empty[True] 0.6262ms 0.3943ms 2.5359 KOps/s 2.4159 KOps/s $\color{#35bf28}+4.97\%$
test_empty[False] 7.3558μs 1.0727μs 932.2125 KOps/s 983.6447 KOps/s $\textbf{\color{#d91a1a}-5.23\%}$
test_unbind_speed 1.7495ms 0.2588ms 3.8638 KOps/s 4.0215 KOps/s $\color{#d91a1a}-3.92\%$
test_unbind_speed_stack0 0.4004ms 0.2464ms 4.0592 KOps/s 4.1255 KOps/s $\color{#d91a1a}-1.61\%$
test_unbind_speed_stack1 0.1469s 0.7012ms 1.4262 KOps/s 1.4759 KOps/s $\color{#d91a1a}-3.37\%$
test_split 1.7748ms 1.5231ms 656.5521 Ops/s 602.1056 Ops/s $\textbf{\color{#35bf28}+9.04\%}$
test_chunk 0.1451s 1.7799ms 561.8283 Ops/s 684.9268 Ops/s $\textbf{\color{#d91a1a}-17.97\%}$
test_creation[device0] 6.2002ms 0.1085ms 9.2141 KOps/s 9.8194 KOps/s $\textbf{\color{#d91a1a}-6.16\%}$
test_creation_from_tensor 0.3128ms 84.3574μs 11.8543 KOps/s 12.0007 KOps/s $\color{#d91a1a}-1.22\%$
test_add_one[memmap_tensor0] 0.1906ms 5.7311μs 174.4861 KOps/s 182.4240 KOps/s $\color{#d91a1a}-4.35\%$
test_contiguous[memmap_tensor0] 25.6380μs 0.6358μs 1.5729 MOps/s 1.5766 MOps/s $\color{#d91a1a}-0.24\%$
test_stack[memmap_tensor0] 38.8120μs 3.9714μs 251.8005 KOps/s 279.0343 KOps/s $\textbf{\color{#d91a1a}-9.76\%}$
test_memmaptd_index 1.2068ms 0.2441ms 4.0958 KOps/s 4.3798 KOps/s $\textbf{\color{#d91a1a}-6.48\%}$
test_memmaptd_index_astensor 0.5688ms 0.3068ms 3.2595 KOps/s 3.2679 KOps/s $\color{#d91a1a}-0.26\%$
test_memmaptd_index_op 1.2065ms 0.6593ms 1.5168 KOps/s 1.6408 KOps/s $\textbf{\color{#d91a1a}-7.56\%}$
test_serialize_model 0.1122s 0.1036s 9.6517 Ops/s 8.6982 Ops/s $\textbf{\color{#35bf28}+10.96\%}$
test_serialize_model_pickle 0.5890s 0.3868s 2.5854 Ops/s 2.6135 Ops/s $\color{#d91a1a}-1.08\%$
test_serialize_weights 0.1050s 99.1323ms 10.0875 Ops/s 10.0000 Ops/s $\color{#35bf28}+0.88\%$
test_serialize_weights_returnearly 0.1319s 0.1250s 7.9996 Ops/s 8.1036 Ops/s $\color{#d91a1a}-1.28\%$
test_serialize_weights_pickle 1.1066s 0.5686s 1.7588 Ops/s 1.5513 Ops/s $\textbf{\color{#35bf28}+13.38\%}$
test_serialize_weights_filesystem 91.8738ms 88.8080ms 11.2603 Ops/s 11.1692 Ops/s $\color{#35bf28}+0.82\%$
test_serialize_model_filesystem 0.1033s 94.0672ms 10.6307 Ops/s 10.7787 Ops/s $\color{#d91a1a}-1.37\%$
test_reshape_pytree 48.7910μs 21.2871μs 46.9767 KOps/s 49.1469 KOps/s $\color{#d91a1a}-4.42\%$
test_reshape_td 68.8490μs 32.7807μs 30.5057 KOps/s 32.0559 KOps/s $\color{#d91a1a}-4.84\%$
test_view_pytree 54.3010μs 21.0891μs 47.4178 KOps/s 48.9576 KOps/s $\color{#d91a1a}-3.15\%$
test_view_td 0.1224s 63.0740μs 15.8544 KOps/s 16.0418 KOps/s $\color{#d91a1a}-1.17\%$
test_unbind_pytree 66.2540μs 24.4584μs 40.8857 KOps/s 40.8435 KOps/s $\color{#35bf28}+0.10\%$
test_unbind_td 0.5306ms 37.3979μs 26.7395 KOps/s 27.7590 KOps/s $\color{#d91a1a}-3.67\%$
test_split_pytree 54.1110μs 24.4954μs 40.8239 KOps/s 43.4516 KOps/s $\textbf{\color{#d91a1a}-6.05\%}$
test_split_td 0.1218ms 41.0355μs 24.3691 KOps/s 25.5260 KOps/s $\color{#d91a1a}-4.53\%$
test_add_pytree 79.6590μs 30.1942μs 33.1190 KOps/s 33.5007 KOps/s $\color{#d91a1a}-1.14\%$
test_add_td 0.1171ms 57.1346μs 17.5025 KOps/s 17.9561 KOps/s $\color{#d91a1a}-2.53\%$
test_distributed 0.2594ms 99.7724μs 10.0228 KOps/s 9.8067 KOps/s $\color{#35bf28}+2.20\%$
test_tdmodule 33.4430μs 17.7709μs 56.2719 KOps/s 56.3376 KOps/s $\color{#d91a1a}-0.12\%$
test_tdmodule_dispatch 70.5020μs 36.0207μs 27.7618 KOps/s 27.6773 KOps/s $\color{#35bf28}+0.31\%$
test_tdseq 45.3450μs 20.7877μs 48.1054 KOps/s 48.7830 KOps/s $\color{#d91a1a}-1.39\%$
test_tdseq_dispatch 65.9730μs 40.6388μs 24.6070 KOps/s 24.8227 KOps/s $\color{#d91a1a}-0.87\%$
test_instantiation_functorch 1.4182ms 1.3031ms 767.3957 Ops/s 778.5141 Ops/s $\color{#d91a1a}-1.43\%$
test_instantiation_td 1.6620ms 1.0093ms 990.8087 Ops/s 1.0106 KOps/s $\color{#d91a1a}-1.96\%$
test_exec_functorch 0.2871ms 0.1572ms 6.3598 KOps/s 6.3004 KOps/s $\color{#35bf28}+0.94\%$
test_exec_functional_call 0.2804ms 0.1470ms 6.8030 KOps/s 6.7665 KOps/s $\color{#35bf28}+0.54\%$
test_exec_td 0.2269ms 0.1416ms 7.0643 KOps/s 7.1038 KOps/s $\color{#d91a1a}-0.56\%$
test_exec_td_decorator 0.7598ms 0.1965ms 5.0878 KOps/s 5.1653 KOps/s $\color{#d91a1a}-1.50\%$
test_vmap_mlp_speed[True-True] 1.2726ms 0.4879ms 2.0497 KOps/s 2.1187 KOps/s $\color{#d91a1a}-3.26\%$
test_vmap_mlp_speed[True-False] 0.7031ms 0.4785ms 2.0901 KOps/s 2.1313 KOps/s $\color{#d91a1a}-1.93\%$
test_vmap_mlp_speed[False-True] 0.7205ms 0.3879ms 2.5778 KOps/s 2.6379 KOps/s $\color{#d91a1a}-2.28\%$
test_vmap_mlp_speed[False-False] 0.5993ms 0.3834ms 2.6082 KOps/s 2.6365 KOps/s $\color{#d91a1a}-1.07\%$
test_vmap_mlp_speed_decorator[True-True] 1.2410ms 0.5008ms 1.9966 KOps/s 2.0237 KOps/s $\color{#d91a1a}-1.34\%$
test_vmap_mlp_speed_decorator[True-False] 0.7336ms 0.4975ms 2.0099 KOps/s 2.0341 KOps/s $\color{#d91a1a}-1.19\%$
test_vmap_mlp_speed_decorator[False-True] 1.1350ms 0.4085ms 2.4477 KOps/s 2.4989 KOps/s $\color{#d91a1a}-2.05\%$
test_vmap_mlp_speed_decorator[False-False] 0.7120ms 0.4036ms 2.4777 KOps/s 2.5068 KOps/s $\color{#d91a1a}-1.16\%$
test_to_module_speed[True] 2.2566ms 1.4340ms 697.3707 Ops/s 726.2776 Ops/s $\color{#d91a1a}-3.98\%$
test_to_module_speed[False] 1.5380ms 1.3924ms 718.1925 Ops/s 727.6538 Ops/s $\color{#d91a1a}-1.30\%$

@vmoens vmoens merged commit cf7cc95 into main Apr 18, 2024
44 of 48 checks passed
@vmoens vmoens deleted the faster-flatten branch April 18, 2024 09:25
@matteobettini
Copy link
Contributor

matteobettini commented Apr 18, 2024 •
edited
Loading

This PR has broken the following script

from tensordict.nn import TensorDictModule
from torch import nn

from torchrl.modules import VDNMixer, MultiAgentMLP, QValueModule, SafeSequential
from torchrl.objectives import QMixerLoss

net = MultiAgentMLP(
    n_agent_inputs=3,
    n_agent_outputs=4,
    n_agents=2,
    centralised=False,
    share_params=True,
    device="cpu",
    depth=2,
    num_cells=256,
    activation_class=nn.Tanh,
)
module = TensorDictModule(
    net, in_keys=[("agents", "observation")], out_keys=[("agents", "action_value")]
)
value_module = QValueModule(
    action_value_key=("agents", "action_value"),
    out_keys=[
        ("agents", "action"),
        ("agents", "action_value"),
        ("agents", "chosen_action_value"),
    ],
    action_space="categorical",
)
qnet = SafeSequential(module, value_module)

mixer = TensorDictModule(
    module=VDNMixer(
        n_agents=2,
        device="cpu",
    ),
    in_keys=[("agents", "chosen_action_value")],
    out_keys=["chosen_action_value"],
)
loss_module = QMixerLoss(qnet, mixer, delay_value=True, action_space="categorical")
loss_module.state_dict()
Traceback (most recent call last):
  File "/Users/Matteo/PycharmProjects/torchrl/torchrl/prova.py", line 44, in <module>
    loss_module.state_dict()
  File "/opt/homebrew/Caskroom/miniforge/base/envs/torchrl/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1895, in state_dict
    module.state_dict(destination=destination, prefix=prefix + name + '.', keep_vars=keep_vars)
  File "/Users/Matteo/PycharmProjects/tensordict/tensordict/nn/params.py", line 996, in state_dict
    return self._param_td.state_dict(
  File "/Users/Matteo/PycharmProjects/tensordict/tensordict/base.py", line 1903, in state_dict
    source = source.flatten_keys(".")
  File "/Users/Matteo/PycharmProjects/tensordict/tensordict/utils.py", line 1060, in newfun
    return fun(_self, *args, **kwargs)
  File "/Users/Matteo/PycharmProjects/tensordict/tensordict/base.py", line 6215, in flatten_keys
    return self._flatten_keys_outplace(separator=separator, is_leaf=is_leaf)
  File "/Users/Matteo/PycharmProjects/tensordict/tensordict/base.py", line 6220, in _flatten_keys_outplace
    all_leaves, all_vals = zip(
ValueError: not enough values to unpack (expected 2, got 0)

tested to work before the PR

@vmoens
Copy link
Contributor Author

vmoens commented Apr 18, 2024

Got it
Next time don't hesitate to post a smaller example, it makes it easier to replicate and indentify the issue

@vmoens
Copy link
Contributor Author

vmoens commented Apr 18, 2024

(additionally i would post it as an issue and not a comment, there's a high probability I miss comments on closed PRs)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. Performance
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@vmoens @matteobettini @facebook-github-bot