Skip to content

[Feature] Tabular data import/export (pandas, CSV, Parquet, JSON)#1679

Merged
vmoens merged 7 commits intomainfrom
from-data-format
Apr 29, 2026
Merged

[Feature] Tabular data import/export (pandas, CSV, Parquet, JSON)#1679
vmoens merged 7 commits intomainfrom
from-data-format

Conversation

@vmoens
Copy link
Copy Markdown
Collaborator

@vmoens vmoens commented Apr 24, 2026

Summary

  • Add from_pandas / to_pandas for converting between pandas DataFrames and TensorDict (numeric columns → tensors, string columns → NonTensorData)
  • Add from_csv / to_csv, from_parquet / to_parquet, from_json / to_json for direct file format support
  • File-format functions auto-select the best available backend (pandas or pyarrow); all dependencies are optional
  • separator parameter enables nested TensorDicts from dotted column names (e.g., column "obs.x"td["obs", "x"]) — useful for RL data
  • dtype parameter allows casting all numeric columns to a single dtype (e.g., torch.float32)
  • from_any now auto-detects pandas DataFrames
  • New optional dependency groups in pyproject.toml: pandas, parquet, tabular

Test plan

  • pytest test/test_tabular.py -v — 36 tests covering all formats, roundtrips, edge cases, separator nesting, dtype override, empty DataFrames, categorical columns, from_any integration, module-level wrappers
  • pytest test/test_tensordict.py -k "test_from_any or test_from_dict or test_from_struct" — verify no regressions in existing from_* functions
  • Verify import tensordict works without pandas/pyarrow installed (no hard dependency)

🤖 Generated with Claude Code

Add from_pandas/to_pandas, from_csv/to_csv, from_parquet/to_parquet,
and from_json/to_json for first-class tabular data support. Numeric
columns become tensors, string columns become NonTensorData. A
separator parameter enables nested TensorDicts from dotted column
names (e.g. "obs.x" -> td["obs","x"]). File formats auto-select
the best available backend (pandas or pyarrow). All new dependencies
are optional.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@meta-cla meta-cla 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 24, 2026
@github-actions github-actions Bot added Test Feature New feature labels Apr 24, 2026
@github-actions
Copy link
Copy Markdown
Contributor

github-actions Bot commented Apr 24, 2026

$\color{#D29922}\textsf{\Large&amp;#x26A0;\kern{0.2cm}\normalsize Warning}$ Result of GPU Benchmark Tests

Total Benchmarks: 261. Improved: $\large\color{#35bf28}27$. Worsened: $\large\color{#d91a1a}17$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_plain_set_nested 37.4710μs 14.4322μs 69.2895 KOps/s 68.7598 KOps/s $\color{#35bf28}+0.77\%$
test_plain_set_stack_nested 35.3710μs 14.7394μs 67.8455 KOps/s 68.0889 KOps/s $\color{#d91a1a}-0.36\%$
test_plain_set_nested_inplace 47.6620μs 16.1757μs 61.8211 KOps/s 61.1705 KOps/s $\color{#35bf28}+1.06\%$
test_plain_set_stack_nested_inplace 40.3510μs 16.1698μs 61.8436 KOps/s 62.4830 KOps/s $\color{#d91a1a}-1.02\%$
test_items 25.6510μs 5.4528μs 183.3935 KOps/s 182.8226 KOps/s $\color{#35bf28}+0.31\%$
test_items_nested 0.5470ms 0.4484ms 2.2303 KOps/s 2.1705 KOps/s $\color{#35bf28}+2.75\%$
test_items_nested_locked 0.7232ms 0.4493ms 2.2259 KOps/s 2.1671 KOps/s $\color{#35bf28}+2.71\%$
test_items_nested_leaf 0.1341ms 92.0862μs 10.8594 KOps/s 10.8759 KOps/s $\color{#d91a1a}-0.15\%$
test_items_stack_nested 0.5506ms 0.4499ms 2.2227 KOps/s 2.1858 KOps/s $\color{#35bf28}+1.69\%$
test_items_stack_nested_leaf 0.1379ms 92.1670μs 10.8499 KOps/s 10.7917 KOps/s $\color{#35bf28}+0.54\%$
test_items_stack_nested_locked 0.6533ms 0.4471ms 2.2368 KOps/s 2.1840 KOps/s $\color{#35bf28}+2.42\%$
test_keys 39.6520μs 4.1597μs 240.4038 KOps/s 242.8862 KOps/s $\color{#d91a1a}-1.02\%$
test_keys_nested 0.3826ms 0.1278ms 7.8233 KOps/s 7.8331 KOps/s $\color{#d91a1a}-0.12\%$
test_keys_nested_locked 2.4869ms 0.1366ms 7.3195 KOps/s 7.3459 KOps/s $\color{#d91a1a}-0.36\%$
test_keys_nested_leaf 0.1698ms 0.1182ms 8.4568 KOps/s 8.5032 KOps/s $\color{#d91a1a}-0.55\%$
test_keys_stack_nested 0.1817ms 0.1284ms 7.7870 KOps/s 7.8371 KOps/s $\color{#d91a1a}-0.64\%$
test_keys_stack_nested_leaf 0.1553ms 0.1187ms 8.4255 KOps/s 8.4840 KOps/s $\color{#d91a1a}-0.69\%$
test_keys_stack_nested_locked 0.4229ms 0.1357ms 7.3673 KOps/s 7.3818 KOps/s $\color{#d91a1a}-0.20\%$
test_values 5.7982μs 1.0058μs 994.2594 KOps/s 989.6901 KOps/s $\color{#35bf28}+0.46\%$
test_values_nested 0.1144ms 50.9248μs 19.6368 KOps/s 19.6494 KOps/s $\color{#d91a1a}-0.06\%$
test_values_nested_locked 88.4530μs 54.1175μs 18.4783 KOps/s 18.5056 KOps/s $\color{#d91a1a}-0.15\%$
test_values_nested_leaf 94.8330μs 58.5317μs 17.0847 KOps/s 17.2136 KOps/s $\color{#d91a1a}-0.75\%$
test_values_stack_nested 86.8630μs 51.1673μs 19.5438 KOps/s 19.5369 KOps/s $\color{#35bf28}+0.04\%$
test_values_stack_nested_leaf 0.6574ms 58.8021μs 17.0062 KOps/s 17.0918 KOps/s $\color{#d91a1a}-0.50\%$
test_values_stack_nested_locked 86.7030μs 54.6840μs 18.2869 KOps/s 18.3576 KOps/s $\color{#d91a1a}-0.39\%$
test_membership 5.3852μs 0.7893μs 1.2669 MOps/s 1.2508 MOps/s $\color{#35bf28}+1.29\%$
test_membership_nested 31.6520μs 2.7414μs 364.7816 KOps/s 364.7007 KOps/s $\color{#35bf28}+0.02\%$
test_membership_nested_leaf 16.8760μs 2.6507μs 377.2593 KOps/s 362.6045 KOps/s $\color{#35bf28}+4.04\%$
test_membership_stacked_nested 30.3510μs 2.7265μs 366.7689 KOps/s 365.1490 KOps/s $\color{#35bf28}+0.44\%$
test_membership_stacked_nested_leaf 33.7910μs 2.7366μs 365.4126 KOps/s 368.6162 KOps/s $\color{#d91a1a}-0.87\%$
test_membership_nested_last 28.6710μs 4.1292μs 242.1783 KOps/s 242.7558 KOps/s $\color{#d91a1a}-0.24\%$
test_membership_nested_leaf_last 23.2600μs 4.0992μs 243.9488 KOps/s 242.9301 KOps/s $\color{#35bf28}+0.42\%$
test_membership_stacked_nested_last 39.0620μs 4.1383μs 241.6476 KOps/s 242.8227 KOps/s $\color{#d91a1a}-0.48\%$
test_membership_stacked_nested_leaf_last 31.8310μs 4.1190μs 242.7771 KOps/s 243.1639 KOps/s $\color{#d91a1a}-0.16\%$
test_nested_getleaf 45.5210μs 20.5530μs 48.6547 KOps/s 48.6282 KOps/s $\color{#35bf28}+0.05\%$
test_nested_get 54.6010μs 19.5433μs 51.1683 KOps/s 51.2088 KOps/s $\color{#d91a1a}-0.08\%$
test_stacked_getleaf 56.0820μs 20.7998μs 48.0774 KOps/s 48.3925 KOps/s $\color{#d91a1a}-0.65\%$
test_stacked_get 50.8710μs 19.7321μs 50.6788 KOps/s 51.1469 KOps/s $\color{#d91a1a}-0.92\%$
test_nested_getitemleaf 55.5110μs 21.3611μs 46.8142 KOps/s 48.6485 KOps/s $\color{#d91a1a}-3.77\%$
test_nested_getitem 48.0710μs 20.0392μs 49.9021 KOps/s 49.6696 KOps/s $\color{#35bf28}+0.47\%$
test_stacked_getitemleaf 44.4320μs 21.2763μs 47.0007 KOps/s 47.1073 KOps/s $\color{#d91a1a}-0.23\%$
test_stacked_getitem 47.3410μs 20.0448μs 49.8883 KOps/s 49.2618 KOps/s $\color{#35bf28}+1.27\%$
test_lock_nested 5.2623ms 0.4621ms 2.1641 KOps/s 2.1837 KOps/s $\color{#d91a1a}-0.90\%$
test_lock_stack_nested 0.5454ms 0.4626ms 2.1615 KOps/s 2.1342 KOps/s $\color{#35bf28}+1.28\%$
test_unlock_nested 0.6143ms 0.3707ms 2.6977 KOps/s 2.6499 KOps/s $\color{#35bf28}+1.80\%$
test_unlock_stack_nested 0.4286ms 0.3735ms 2.6771 KOps/s 2.6193 KOps/s $\color{#35bf28}+2.21\%$
test_flatten_speed 0.1711ms 0.1159ms 8.6287 KOps/s 8.5521 KOps/s $\color{#35bf28}+0.90\%$
test_unflatten_speed 0.6463ms 0.5478ms 1.8254 KOps/s 1.7935 KOps/s $\color{#35bf28}+1.78\%$
test_common_ops 96.0281ms 0.8223ms 1.2162 KOps/s 237.5913 Ops/s $\textbf{\color{#35bf28}+411.87\%}$
test_creation 0.1367ms 2.9632μs 337.4731 KOps/s 342.8558 KOps/s $\color{#d91a1a}-1.57\%$
test_creation_empty 34.5210μs 6.5499μs 152.6737 KOps/s 153.5918 KOps/s $\color{#d91a1a}-0.60\%$
test_creation_nested_1 40.0020μs 10.9530μs 91.2991 KOps/s 91.3344 KOps/s $\color{#d91a1a}-0.04\%$
test_creation_nested_2 37.6310μs 12.6182μs 79.2506 KOps/s 79.8929 KOps/s $\color{#d91a1a}-0.80\%$
test_creation_many_keys[10] 49.8520μs 19.6219μs 50.9636 KOps/s 50.6941 KOps/s $\color{#35bf28}+0.53\%$
test_creation_many_keys[50] 0.1281ms 83.4384μs 11.9849 KOps/s 11.7609 KOps/s $\color{#35bf28}+1.90\%$
test_creation_many_keys[100] 0.2252ms 0.1644ms 6.0833 KOps/s 5.9253 KOps/s $\color{#35bf28}+2.67\%$
test_creation_nested_many_keys[10] 77.7230μs 42.3053μs 23.6377 KOps/s 23.6558 KOps/s $\color{#d91a1a}-0.08\%$
test_creation_nested_many_keys[50] 0.2080ms 0.1727ms 5.7899 KOps/s 5.7562 KOps/s $\color{#35bf28}+0.59\%$
test_clone 70.1820μs 12.7572μs 78.3873 KOps/s 78.2131 KOps/s $\color{#35bf28}+0.22\%$
test_getitem[int] 1.6000ms 14.5989μs 68.4983 KOps/s 61.6194 KOps/s $\textbf{\color{#35bf28}+11.16\%}$
test_getitem[slice_int] 0.1411ms 23.8792μs 41.8775 KOps/s 41.3712 KOps/s $\color{#35bf28}+1.22\%$
test_getitem[range] 0.1746ms 65.2742μs 15.3200 KOps/s 16.2224 KOps/s $\textbf{\color{#d91a1a}-5.56\%}$
test_getitem[tuple] 0.1416ms 22.7779μs 43.9022 KOps/s 42.3080 KOps/s $\color{#35bf28}+3.77\%$
test_getitem[list] 0.1934ms 56.6847μs 17.6415 KOps/s 17.6431 KOps/s $-0.01\%$
test_setitem_dim[int] 46.0510μs 25.0140μs 39.9776 KOps/s 39.5628 KOps/s $\color{#35bf28}+1.05\%$
test_setitem_dim[slice_int] 78.6520μs 41.0631μs 24.3528 KOps/s 23.8754 KOps/s $\color{#35bf28}+2.00\%$
test_setitem_dim[range] 0.1135ms 90.5204μs 11.0472 KOps/s 10.7994 KOps/s $\color{#35bf28}+2.30\%$
test_setitem_dim[tuple] 57.3320μs 37.5163μs 26.6550 KOps/s 25.5956 KOps/s $\color{#35bf28}+4.14\%$
test_setitem 59.2720μs 16.9820μs 58.8860 KOps/s 58.7491 KOps/s $\color{#35bf28}+0.23\%$
test_set 60.4020μs 17.2835μs 57.8587 KOps/s 60.4298 KOps/s $\color{#d91a1a}-4.25\%$
test_set_shared 0.4971ms 0.2045ms 4.8903 KOps/s 4.8148 KOps/s $\color{#35bf28}+1.57\%$
test_update 0.3499ms 21.0203μs 47.5731 KOps/s 46.8256 KOps/s $\color{#35bf28}+1.60\%$
test_update_nested 79.0130μs 32.4069μs 30.8576 KOps/s 30.3850 KOps/s $\color{#35bf28}+1.56\%$
test_update__nested 0.5183ms 32.7733μs 30.5126 KOps/s 29.0247 KOps/s $\textbf{\color{#35bf28}+5.13\%}$
test_set_nested 62.5120μs 18.1861μs 54.9870 KOps/s 54.9121 KOps/s $\color{#35bf28}+0.14\%$
test_set_nested_new 71.2620μs 23.3031μs 42.9127 KOps/s 42.4938 KOps/s $\color{#35bf28}+0.99\%$
test_select 93.9830μs 38.7853μs 25.7830 KOps/s 25.8056 KOps/s $\color{#d91a1a}-0.09\%$
test_select_nested 99.8830μs 70.0930μs 14.2668 KOps/s 14.3586 KOps/s $\color{#d91a1a}-0.64\%$
test_exclude_nested 0.1190ms 87.2289μs 11.4641 KOps/s 11.5194 KOps/s $\color{#d91a1a}-0.48\%$
test_empty[True] 0.4372ms 0.3863ms 2.5890 KOps/s 2.5764 KOps/s $\color{#35bf28}+0.49\%$
test_empty[False] 8.4502μs 1.2498μs 800.1264 KOps/s 809.8462 KOps/s $\color{#d91a1a}-1.20\%$
test_to 0.1053ms 73.8837μs 13.5348 KOps/s 13.3285 KOps/s $\color{#35bf28}+1.55\%$
test_to_nonblocking 0.1221ms 68.2060μs 14.6615 KOps/s 14.9925 KOps/s $\color{#d91a1a}-2.21\%$
test_unbind_speed 0.3719ms 0.3207ms 3.1184 KOps/s 3.1242 KOps/s $\color{#d91a1a}-0.18\%$
test_unbind_speed_stack0 0.4354ms 0.3174ms 3.1506 KOps/s 3.1386 KOps/s $\color{#35bf28}+0.38\%$
test_unbind_speed_stack1 0.1059s 1.0105ms 989.6554 Ops/s 1.2179 KOps/s $\textbf{\color{#d91a1a}-18.74\%}$
test_split 1.2137ms 1.0893ms 918.0444 Ops/s 820.8376 Ops/s $\textbf{\color{#35bf28}+11.84\%}$
test_chunk 0.1055s 1.1644ms 858.8465 Ops/s 958.1635 Ops/s $\textbf{\color{#d91a1a}-10.37\%}$
test_to_cpu_blocking 19.7894ms 19.1207ms 52.2994 Ops/s 52.0514 Ops/s $\color{#35bf28}+0.48\%$
test_to_cpu_global_sync 11.6620ms 11.4946ms 86.9973 Ops/s 78.3932 Ops/s $\textbf{\color{#35bf28}+10.98\%}$
test_to_cpu_event_sync 0.1188s 13.7036ms 72.9737 Ops/s 81.2651 Ops/s $\textbf{\color{#d91a1a}-10.20\%}$
test_to_cpu_default 12.8728ms 12.3742ms 80.8131 Ops/s 80.8596 Ops/s $\color{#d91a1a}-0.06\%$
test_consolidate[False-None] 4.1699ms 3.9559ms 252.7843 Ops/s 216.7808 Ops/s $\textbf{\color{#35bf28}+16.61\%}$
test_consolidate[default-None] 2.3822ms 1.9527ms 512.1064 Ops/s 501.4082 Ops/s $\color{#35bf28}+2.13\%$
test_consolidate[reduce-overhead-None] 1.9508ms 1.8805ms 531.7710 Ops/s 516.3504 Ops/s $\color{#35bf28}+2.99\%$
test_consolidate_njt[False-None] 8.5235ms 8.2963ms 120.5358 Ops/s 119.0167 Ops/s $\color{#35bf28}+1.28\%$
test_to[False-False-None] 2.2714ms 2.1177ms 472.2105 Ops/s 466.7594 Ops/s $\color{#35bf28}+1.17\%$
test_to[True-False-None] 1.9798ms 1.8394ms 543.6573 Ops/s 547.6386 Ops/s $\color{#d91a1a}-0.73\%$
test_to[within-False-None] 6.2686ms 5.9365ms 168.4489 Ops/s 168.4932 Ops/s $\color{#d91a1a}-0.03\%$
test_to[True-default-None] 9.1538ms 8.9999ms 111.1120 Ops/s 109.7025 Ops/s $\color{#35bf28}+1.28\%$
test_to_njt[False-False-None] 8.4812ms 8.4000ms 119.0472 Ops/s 119.7092 Ops/s $\color{#d91a1a}-0.55\%$
test_to_njt[True-False-None] 7.1916ms 6.8767ms 145.4179 Ops/s 148.8550 Ops/s $\color{#d91a1a}-2.31\%$
test_to_njt[within-False-None] 15.3956ms 15.1313ms 66.0880 Ops/s 65.8167 Ops/s $\color{#35bf28}+0.41\%$
test_creation[device0] 0.2890ms 0.1174ms 8.5178 KOps/s 8.6298 KOps/s $\color{#d91a1a}-1.30\%$
test_creation_from_tensor 0.3955ms 0.1146ms 8.7274 KOps/s 8.9329 KOps/s $\color{#d91a1a}-2.30\%$
test_add_one[memmap_tensor0] 0.1943ms 6.4031μs 156.1742 KOps/s 154.6713 KOps/s $\color{#35bf28}+0.97\%$
test_contiguous[memmap_tensor0] 26.4910μs 0.5889μs 1.6980 MOps/s 2.1004 MOps/s $\textbf{\color{#d91a1a}-19.16\%}$
test_stack[memmap_tensor0] 37.0420μs 4.4048μs 227.0254 KOps/s 224.5209 KOps/s $\color{#35bf28}+1.12\%$
test_memmaptd_index 1.0309ms 0.2606ms 3.8380 KOps/s 3.8929 KOps/s $\color{#d91a1a}-1.41\%$
test_memmaptd_index_astensor 0.5078ms 0.3583ms 2.7906 KOps/s 2.8297 KOps/s $\color{#d91a1a}-1.38\%$
test_memmaptd_index_op 0.9151ms 0.6089ms 1.6423 KOps/s 1.6565 KOps/s $\color{#d91a1a}-0.85\%$
test_serialize_model 0.1348s 0.1342s 7.4537 Ops/s 7.4068 Ops/s $\color{#35bf28}+0.63\%$
test_serialize_model_pickle 1.3612s 1.2126s 0.8246 Ops/s 0.8260 Ops/s $\color{#d91a1a}-0.16\%$
test_serialize_weights 0.1366s 0.1330s 7.5188 Ops/s 6.1386 Ops/s $\textbf{\color{#35bf28}+22.48\%}$
test_serialize_weights_returnearly 68.6627ms 61.2348ms 16.3306 Ops/s 15.6133 Ops/s $\color{#35bf28}+4.59\%$
test_serialize_weights_pickle 1.3500s 1.2106s 0.8260 Ops/s 0.8212 Ops/s $\color{#35bf28}+0.59\%$
test_reshape_pytree 0.2137ms 30.6915μs 32.5823 KOps/s 32.5610 KOps/s $\color{#35bf28}+0.07\%$
test_reshape_td 72.3930μs 43.9870μs 22.7340 KOps/s 23.4681 KOps/s $\color{#d91a1a}-3.13\%$
test_view_pytree 0.2182ms 30.1502μs 33.1672 KOps/s 32.8946 KOps/s $\color{#35bf28}+0.83\%$
test_view_td 85.3530μs 52.1549μs 19.1736 KOps/s 19.7741 KOps/s $\color{#d91a1a}-3.04\%$
test_unbind_pytree 0.2323ms 34.8124μs 28.7254 KOps/s 29.1438 KOps/s $\color{#d91a1a}-1.44\%$
test_unbind_td 0.1777ms 48.3150μs 20.6975 KOps/s 20.9556 KOps/s $\color{#d91a1a}-1.23\%$
test_split_pytree 0.2377ms 40.0064μs 24.9960 KOps/s 24.2714 KOps/s $\color{#35bf28}+2.99\%$
test_split_td 0.1579ms 60.7877μs 16.4507 KOps/s 16.2060 KOps/s $\color{#35bf28}+1.51\%$
test_add_pytree 0.2241ms 41.0458μs 24.3631 KOps/s 25.1976 KOps/s $\color{#d91a1a}-3.31\%$
test_add_td 0.1045ms 55.6876μs 17.9573 KOps/s 17.9886 KOps/s $\color{#d91a1a}-0.17\%$
test_compile_add_one_nested[tensordict-compile] 0.2638ms 0.1649ms 6.0660 KOps/s 6.0300 KOps/s $\color{#35bf28}+0.60\%$
test_compile_add_one_nested[tensordict-eager] 0.5415ms 0.1944ms 5.1451 KOps/s 5.2123 KOps/s $\color{#d91a1a}-1.29\%$
test_compile_add_one_nested[pytree-compile] 0.2860ms 0.1246ms 8.0266 KOps/s 7.7415 KOps/s $\color{#35bf28}+3.68\%$
test_compile_add_one_nested[pytree-eager] 0.4343ms 0.1739ms 5.7488 KOps/s 5.7570 KOps/s $\color{#d91a1a}-0.14\%$
test_compile_copy_nested[tensordict-compile] 0.1282ms 16.5181μs 60.5397 KOps/s 61.9284 KOps/s $\color{#d91a1a}-2.24\%$
test_compile_copy_nested[tensordict-eager] 0.4915ms 50.5076μs 19.7990 KOps/s 19.6390 KOps/s $\color{#35bf28}+0.81\%$
test_compile_copy_nested[pytree-compile] 0.4659ms 15.5453μs 64.3279 KOps/s 64.4437 KOps/s $\color{#d91a1a}-0.18\%$
test_compile_copy_nested[pytree-eager] 0.5029ms 63.3080μs 15.7958 KOps/s 15.7625 KOps/s $\color{#35bf28}+0.21\%$
test_compile_add_one_flat[tensordict-compile] 0.6423ms 0.1956ms 5.1133 KOps/s 5.0029 KOps/s $\color{#35bf28}+2.21\%$
test_compile_add_one_flat[tensordict-eager] 0.3587ms 0.2727ms 3.6675 KOps/s 3.6524 KOps/s $\color{#35bf28}+0.41\%$
test_compile_add_one_flat[tensorclass-compile] 0.2020ms 0.1316ms 7.5986 KOps/s 7.4666 KOps/s $\color{#35bf28}+1.77\%$
test_compile_add_one_flat[tensorclass-eager] 0.5082ms 72.6740μs 13.7601 KOps/s 13.2280 KOps/s $\color{#35bf28}+4.02\%$
test_compile_add_one_flat[pytree-compile] 0.6164ms 0.1762ms 5.6768 KOps/s 5.5660 KOps/s $\color{#35bf28}+1.99\%$
test_compile_add_one_flat[pytree-eager] 0.9526ms 0.5109ms 1.9573 KOps/s 1.8374 KOps/s $\textbf{\color{#35bf28}+6.53\%}$
test_compile_add_self_flat[tensordict-eager] 0.7357ms 0.3229ms 3.0965 KOps/s 3.0790 KOps/s $\color{#35bf28}+0.57\%$
test_compile_add_self_flat[tensordict-compile] 0.6227ms 0.1975ms 5.0625 KOps/s 4.8176 KOps/s $\textbf{\color{#35bf28}+5.08\%}$
test_compile_add_self_flat[tensorclass-eager] 0.1340ms 88.7272μs 11.2705 KOps/s 11.3553 KOps/s $\color{#d91a1a}-0.75\%$
test_compile_add_self_flat[tensorclass-compile] 0.3847ms 0.1343ms 7.4463 KOps/s 7.0864 KOps/s $\textbf{\color{#35bf28}+5.08\%}$
test_compile_add_self_flat[pytree-eager] 0.8709ms 0.4262ms 2.3462 KOps/s 2.3346 KOps/s $\color{#35bf28}+0.50\%$
test_compile_add_self_flat[pytree-compile] 0.6174ms 0.1771ms 5.6457 KOps/s 5.3757 KOps/s $\textbf{\color{#35bf28}+5.02\%}$
test_compile_copy_flat[tensordict-compile] 0.4627ms 20.2507μs 49.3811 KOps/s 50.9402 KOps/s $\color{#d91a1a}-3.06\%$
test_compile_copy_flat[tensordict-eager] 78.0830μs 39.9514μs 25.0304 KOps/s 24.9166 KOps/s $\color{#35bf28}+0.46\%$
test_compile_copy_flat[pytree-compile] 0.7242ms 16.2270μs 61.6255 KOps/s 61.4617 KOps/s $\color{#35bf28}+0.27\%$
test_compile_copy_flat[pytree-eager] 0.5004ms 50.9150μs 19.6406 KOps/s 19.5652 KOps/s $\color{#35bf28}+0.38\%$
test_compile_assign_and_add[tensordict-compile] 2.0084ms 0.1863ms 5.3673 KOps/s 5.1088 KOps/s $\textbf{\color{#35bf28}+5.06\%}$
test_compile_assign_and_add[tensordict-eager] 3.6971ms 3.2992ms 303.1044 Ops/s 303.0308 Ops/s $\color{#35bf28}+0.02\%$
test_compile_assign_and_add[pytree-compile] 1.9891ms 0.1717ms 5.8231 KOps/s 5.6657 KOps/s $\color{#35bf28}+2.78\%$
test_compile_assign_and_add[pytree-eager] 3.1539ms 2.7710ms 360.8822 Ops/s 345.5969 Ops/s $\color{#35bf28}+4.42\%$
test_compile_indexing[tensor-tensordict-compile] 0.5680ms 0.1248ms 8.0143 KOps/s 7.8072 KOps/s $\color{#35bf28}+2.65\%$
test_compile_indexing[tensor-tensordict-eager] 0.3071ms 73.8245μs 13.5456 KOps/s 13.9588 KOps/s $\color{#d91a1a}-2.96\%$
test_compile_indexing[tensor-tensorclass-compile] 0.2207ms 0.1118ms 8.9449 KOps/s 8.8887 KOps/s $\color{#35bf28}+0.63\%$
test_compile_indexing[tensor-tensorclass-eager] 0.5366ms 44.1323μs 22.6592 KOps/s 21.4414 KOps/s $\textbf{\color{#35bf28}+5.68\%}$
test_compile_indexing[tensor-pytree-compile] 0.5533ms 0.1126ms 8.8819 KOps/s 8.7986 KOps/s $\color{#35bf28}+0.95\%$
test_compile_indexing[tensor-pytree-eager] 0.4770ms 44.0359μs 22.7088 KOps/s 21.3683 KOps/s $\textbf{\color{#35bf28}+6.27\%}$
test_compile_indexing[slice-tensordict-compile] 0.5075ms 69.3368μs 14.4224 KOps/s 14.3711 KOps/s $\color{#35bf28}+0.36\%$
test_compile_indexing[slice-tensordict-eager] 0.2073ms 27.9212μs 35.8150 KOps/s 37.0603 KOps/s $\color{#d91a1a}-3.36\%$
test_compile_indexing[slice-tensorclass-compile] 0.5036ms 56.5167μs 17.6939 KOps/s 17.7648 KOps/s $\color{#d91a1a}-0.40\%$
test_compile_indexing[slice-tensorclass-eager] 0.2363ms 21.1452μs 47.2921 KOps/s 46.0916 KOps/s $\color{#35bf28}+2.60\%$
test_compile_indexing[slice-pytree-compile] 0.4910ms 57.4531μs 17.4055 KOps/s 17.1712 KOps/s $\color{#35bf28}+1.36\%$
test_compile_indexing[slice-pytree-eager] 0.4496ms 20.9745μs 47.6769 KOps/s 47.5148 KOps/s $\color{#35bf28}+0.34\%$
test_compile_indexing[int-tensordict-compile] 0.4980ms 69.6694μs 14.3535 KOps/s 14.0119 KOps/s $\color{#35bf28}+2.44\%$
test_compile_indexing[int-tensordict-eager] 0.4888ms 27.0074μs 37.0269 KOps/s 34.2645 KOps/s $\textbf{\color{#35bf28}+8.06\%}$
test_compile_indexing[int-tensorclass-compile] 0.4894ms 56.4979μs 17.6998 KOps/s 16.3134 KOps/s $\textbf{\color{#35bf28}+8.50\%}$
test_compile_indexing[int-tensorclass-eager] 0.2327ms 20.9399μs 47.7557 KOps/s 45.1225 KOps/s $\textbf{\color{#35bf28}+5.84\%}$
test_compile_indexing[int-pytree-compile] 0.4835ms 56.5558μs 17.6817 KOps/s 16.3199 KOps/s $\textbf{\color{#35bf28}+8.34\%}$
test_compile_indexing[int-pytree-eager] 0.4476ms 21.0406μs 47.5273 KOps/s 44.8059 KOps/s $\textbf{\color{#35bf28}+6.07\%}$
test_compile_replace[single-eager] 0.4849ms 46.6607μs 21.4313 KOps/s 20.0568 KOps/s $\textbf{\color{#35bf28}+6.85\%}$
test_compile_replace[single-compile] 0.5538ms 0.1197ms 8.3576 KOps/s 8.3769 KOps/s $\color{#d91a1a}-0.23\%$
test_compile_replace[multi-eager] 1.0327ms 0.5576ms 1.7934 KOps/s 1.8176 KOps/s $\color{#d91a1a}-1.34\%$
test_compile_replace[multi-compile] 0.6009ms 0.1261ms 7.9317 KOps/s 7.9785 KOps/s $\color{#d91a1a}-0.59\%$
test_compile_tc_getattr_20[eager] 0.5914ms 0.1649ms 6.0633 KOps/s 6.0474 KOps/s $\color{#35bf28}+0.26\%$
test_compile_tc_getattr_20[compile] 0.1941ms 0.1351ms 7.4028 KOps/s 7.4197 KOps/s $\color{#d91a1a}-0.23\%$
test_compile_clone_shallow[20-eager] 0.4556ms 18.6250μs 53.6913 KOps/s 54.0307 KOps/s $\color{#d91a1a}-0.63\%$
test_compile_clone_shallow[20-compile] 0.4586ms 17.2477μs 57.9786 KOps/s 59.1482 KOps/s $\color{#d91a1a}-1.98\%$
test_compile_clone_shallow[40-eager] 0.4620ms 33.2021μs 30.1185 KOps/s 30.5711 KOps/s $\color{#d91a1a}-1.48\%$
test_compile_clone_shallow[40-compile] 0.4581ms 18.7546μs 53.3203 KOps/s 54.6172 KOps/s $\color{#d91a1a}-2.37\%$
test_compile_clone_shallow[80-eager] 0.4949ms 61.2538μs 16.3255 KOps/s 16.5724 KOps/s $\color{#d91a1a}-1.49\%$
test_compile_clone_shallow[80-compile] 0.1096ms 20.9705μs 47.6859 KOps/s 48.5006 KOps/s $\color{#d91a1a}-1.68\%$
test_compile_update_inplace[eager] 0.4900ms 58.0294μs 17.2326 KOps/s 17.2471 KOps/s $\color{#d91a1a}-0.08\%$
test_compile_update_inplace[compile] 0.5875ms 0.1463ms 6.8333 KOps/s 6.7761 KOps/s $\color{#35bf28}+0.84\%$
test_mod_add[eager] 0.6491ms 0.2161ms 4.6283 KOps/s 20.0249 KOps/s $\textbf{\color{#d91a1a}-76.89\%}$
test_mod_add[compile] 0.5469ms 0.1166ms 8.5791 KOps/s 8.5721 KOps/s $\color{#35bf28}+0.08\%$
test_mod_add[compile-overhead] 0.2434ms 0.1641ms 6.0921 KOps/s 6.0666 KOps/s $\color{#35bf28}+0.42\%$
test_mod_wrap[eager] 0.9100ms 0.4659ms 2.1462 KOps/s 3.5433 KOps/s $\textbf{\color{#d91a1a}-39.43\%}$
test_mod_wrap[compile] 0.4658ms 0.3724ms 2.6853 KOps/s 2.6374 KOps/s $\color{#35bf28}+1.81\%$
test_mod_wrap[compile-overhead] 9.2526ms 5.0177ms 199.2932 Ops/s 199.7564 Ops/s $\color{#d91a1a}-0.23\%$
test_mod_wrap_and_backward[eager] 1.8177ms 1.6817ms 594.6483 Ops/s 661.7099 Ops/s $\textbf{\color{#d91a1a}-10.13\%}$
test_mod_wrap_and_backward[compile] 2.4335ms 1.4926ms 669.9816 Ops/s 636.6783 Ops/s $\textbf{\color{#35bf28}+5.23\%}$
test_mod_wrap_and_backward[compile-overhead] 1.2361ms 0.8955ms 1.1166 KOps/s 979.1355 Ops/s $\textbf{\color{#35bf28}+14.04\%}$
test_seq_add[eager] 0.5836ms 0.4940ms 2.0242 KOps/s 6.3876 KOps/s $\textbf{\color{#d91a1a}-68.31\%}$
test_seq_add[compile] 0.1725ms 0.1274ms 7.8517 KOps/s 7.4654 KOps/s $\textbf{\color{#35bf28}+5.18\%}$
test_seq_add[compile-overhead] 0.2080ms 0.1691ms 5.9120 KOps/s 5.8385 KOps/s $\color{#35bf28}+1.26\%$
test_seq_wrap[eager] 1.4676ms 1.3736ms 727.9949 Ops/s 1.9254 KOps/s $\textbf{\color{#d91a1a}-62.19\%}$
test_seq_wrap[compile] 0.4712ms 0.3797ms 2.6336 KOps/s 2.6380 KOps/s $\color{#d91a1a}-0.17\%$
test_seq_wrap[compile-overhead] 0.3540ms 0.2817ms 3.5494 KOps/s 3.5150 KOps/s $\color{#35bf28}+0.98\%$
test_func_call_runtime[False-eager] 0.9545ms 0.8531ms 1.1722 KOps/s 1.2324 KOps/s $\color{#d91a1a}-4.89\%$
test_func_call_runtime[False-compile] 1.0688ms 0.9063ms 1.1034 KOps/s 1.1027 KOps/s $\color{#35bf28}+0.06\%$
test_func_call_runtime[False-compile-overhead] 0.5303ms 0.4721ms 2.1184 KOps/s 2.1016 KOps/s $\color{#35bf28}+0.80\%$
test_func_call_runtime[True-eager] 1.1254ms 1.0471ms 955.0058 Ops/s 955.2535 Ops/s $\color{#d91a1a}-0.03\%$
test_func_call_runtime[True-compile] 1.0189ms 0.9165ms 1.0911 KOps/s 1.0715 KOps/s $\color{#35bf28}+1.83\%$
test_func_call_runtime[True-compile-overhead] 0.5276ms 0.4839ms 2.0664 KOps/s 2.0228 KOps/s $\color{#35bf28}+2.16\%$
test_func_call_cm_runtime[False-eager] 0.9416ms 0.8213ms 1.2175 KOps/s 1.1682 KOps/s $\color{#35bf28}+4.22\%$
test_func_call_cm_runtime[False-compile] 1.0000ms 0.8820ms 1.1338 KOps/s 1.1103 KOps/s $\color{#35bf28}+2.11\%$
test_func_call_cm_runtime[False-compile-overhead] 2.7961ms 0.4484ms 2.2302 KOps/s 2.2046 KOps/s $\color{#35bf28}+1.16\%$
test_func_call_cm_runtime[True-eager] 1.3035ms 1.1872ms 842.3178 Ops/s 838.9576 Ops/s $\color{#35bf28}+0.40\%$
test_func_call_cm_runtime[True-compile] 1.0456ms 0.9197ms 1.0873 KOps/s 1.0784 KOps/s $\color{#35bf28}+0.83\%$
test_func_call_cm_runtime[True-compile-overhead] 0.5398ms 0.4937ms 2.0256 KOps/s 1.9911 KOps/s $\color{#35bf28}+1.73\%$
test_vmap_func_call_cm_runtime[eager] 2.8314ms 2.3306ms 429.0659 Ops/s 427.4644 Ops/s $\color{#35bf28}+0.37\%$
test_vmap_func_call_cm_runtime[compile] 0.9934ms 0.9369ms 1.0674 KOps/s 1.0519 KOps/s $\color{#35bf28}+1.47\%$
test_vmap_func_call_cm_runtime[compile-overhead] 0.5411ms 0.4976ms 2.0098 KOps/s 1.9614 KOps/s $\color{#35bf28}+2.47\%$
test_distributed 0.5650ms 0.1520ms 6.5806 KOps/s 5.8870 KOps/s $\textbf{\color{#35bf28}+11.78\%}$
test_tdmodule 0.8047ms 0.1844ms 5.4217 KOps/s 35.9956 KOps/s $\textbf{\color{#d91a1a}-84.94\%}$
test_tdmodule_dispatch 0.3890ms 0.3537ms 2.8276 KOps/s 21.4647 KOps/s $\textbf{\color{#d91a1a}-86.83\%}$
test_tdseq 45.3320μs 25.2671μs 39.5772 KOps/s 37.2051 KOps/s $\textbf{\color{#35bf28}+6.38\%}$
test_tdseq_dispatch 0.2467ms 0.2065ms 4.8426 KOps/s 20.9701 KOps/s $\textbf{\color{#d91a1a}-76.91\%}$
test_instantiation_functorch 2.0798ms 1.9737ms 506.6505 Ops/s 502.6079 Ops/s $\color{#35bf28}+0.80\%$
test_exec_functorch 0.2254ms 0.1704ms 5.8692 KOps/s 5.7641 KOps/s $\color{#35bf28}+1.82\%$
test_exec_functional_call 0.2179ms 0.1535ms 6.5144 KOps/s 6.4586 KOps/s $\color{#35bf28}+0.86\%$
test_exec_td_decorator 0.4473ms 0.2257ms 4.4305 KOps/s 4.3751 KOps/s $\color{#35bf28}+1.27\%$
test_vmap_mlp_speed_decorator[True-True] 1.1938ms 0.9866ms 1.0136 KOps/s 1.2394 KOps/s $\textbf{\color{#d91a1a}-18.22\%}$
test_vmap_mlp_speed_decorator[True-False] 1.7595ms 0.9853ms 1.0150 KOps/s 1.2455 KOps/s $\textbf{\color{#d91a1a}-18.51\%}$
test_vmap_mlp_speed_decorator[False-True] 0.8855ms 0.6955ms 1.4379 KOps/s 1.4388 KOps/s $\color{#d91a1a}-0.07\%$
test_vmap_mlp_speed_decorator[False-False] 0.8743ms 0.6934ms 1.4421 KOps/s 1.4388 KOps/s $\color{#35bf28}+0.23\%$
test_vmap_transformer_speed_decorator[True-True] 21.3026ms 20.6316ms 48.4694 Ops/s 49.1331 Ops/s $\color{#d91a1a}-1.35\%$
test_vmap_transformer_speed_decorator[True-False] 21.4587ms 20.5803ms 48.5901 Ops/s 49.3310 Ops/s $\color{#d91a1a}-1.50\%$
test_vmap_transformer_speed_decorator[False-True] 20.8634ms 20.2404ms 49.4062 Ops/s 49.7249 Ops/s $\color{#d91a1a}-0.64\%$
test_vmap_transformer_speed_decorator[False-False] 20.4606ms 20.2185ms 49.4597 Ops/s 49.7964 Ops/s $\color{#d91a1a}-0.68\%$
test_to_module_speed[True] 1.5114ms 1.4230ms 702.7274 Ops/s 701.9709 Ops/s $\color{#35bf28}+0.11\%$
test_to_module_speed[False] 1.4923ms 1.3923ms 718.2289 Ops/s 714.3434 Ops/s $\color{#35bf28}+0.54\%$
test_tc_init 69.7020μs 42.4033μs 23.5831 KOps/s 22.3423 KOps/s $\textbf{\color{#35bf28}+5.55\%}$
test_tc_init_tensor_only 41.8120μs 9.3774μs 106.6396 KOps/s 107.1550 KOps/s $\color{#d91a1a}-0.48\%$
test_tc_init_nested 0.1364ms 84.3276μs 11.8585 KOps/s 11.7081 KOps/s $\color{#35bf28}+1.28\%$
test_tc_init_many_fields 39.4520μs 15.5517μs 64.3016 KOps/s 63.6534 KOps/s $\color{#35bf28}+1.02\%$
test_tc_first_layer_tensor 23.7210μs 1.7106μs 584.5781 KOps/s 580.6594 KOps/s $\color{#35bf28}+0.67\%$
test_tc_first_layer_tensor_only 1.8941μs 0.3812μs 2.6232 MOps/s 2.5743 MOps/s $\color{#35bf28}+1.90\%$
test_tc_first_layer_tensor_set 33.4710μs 3.6740μs 272.1792 KOps/s 263.2402 KOps/s $\color{#35bf28}+3.40\%$
test_tc_first_layer_tensor_only_set 25.2410μs 3.0655μs 326.2148 KOps/s 323.8972 KOps/s $\color{#35bf28}+0.72\%$
test_tc_first_layer_nontensor 32.3610μs 5.7842μs 172.8861 KOps/s 171.2173 KOps/s $\color{#35bf28}+0.97\%$
test_tc_second_layer_tensor 29.6410μs 4.0795μs 245.1253 KOps/s 248.5821 KOps/s $\color{#d91a1a}-1.39\%$
test_tc_second_layer_nontensor 37.8810μs 8.1906μs 122.0912 KOps/s 121.2382 KOps/s $\color{#35bf28}+0.70\%$
test_unbind 0.2807s 13.6619ms 73.1963 Ops/s 65.1682 Ops/s $\textbf{\color{#35bf28}+12.32\%}$
test_full_like 4.8223ms 4.2805ms 233.6179 Ops/s 225.5270 Ops/s $\color{#35bf28}+3.59\%$
test_zeros_like 5.0616ms 4.3670ms 228.9887 Ops/s 229.1786 Ops/s $\color{#d91a1a}-0.08\%$
test_ones_like 4.7968ms 4.3694ms 228.8622 Ops/s 228.5355 Ops/s $\color{#35bf28}+0.14\%$
test_clone 6.8035ms 6.5662ms 152.2949 Ops/s 152.7845 Ops/s $\color{#d91a1a}-0.32\%$
test_squeeze 0.1636ms 13.3369μs 74.9800 KOps/s 74.1681 KOps/s $\color{#35bf28}+1.09\%$
test_unsqueeze 0.1874ms 0.1103ms 9.0648 KOps/s 9.2716 KOps/s $\color{#d91a1a}-2.23\%$
test_split 0.2807ms 0.1785ms 5.6027 KOps/s 5.6879 KOps/s $\color{#d91a1a}-1.50\%$
test_permute 0.2592ms 0.2089ms 4.7880 KOps/s 4.9866 KOps/s $\color{#d91a1a}-3.98\%$
test_stack 51.9851ms 51.3933ms 19.4578 Ops/s 19.4160 Ops/s $\color{#35bf28}+0.22\%$
test_cat 51.9964ms 51.4403ms 19.4400 Ops/s 19.4280 Ops/s $\color{#35bf28}+0.06\%$
test_sequential_tensordict 0.7942ms 0.5548ms 1.8023 KOps/s 4.7360 KOps/s $\textbf{\color{#d91a1a}-61.94\%}$
test_sequential_graph_module 0.5249ms 0.1183ms 8.4516 KOps/s 8.7494 KOps/s $\color{#d91a1a}-3.40\%$
test_nested_tensordict 0.8844ms 0.7884ms 1.2684 KOps/s 3.4945 KOps/s $\textbf{\color{#d91a1a}-63.70\%}$
test_nested_graph_module 0.5154ms 0.1266ms 7.8980 KOps/s 8.0522 KOps/s $\color{#d91a1a}-1.92\%$

@github-actions
Copy link
Copy Markdown
Contributor

github-actions Bot commented Apr 24, 2026

$\color{#D29922}\textsf{\Large&amp;#x26A0;\kern{0.2cm}\normalsize Warning}$ Result of CPU Benchmark Tests

Total Benchmarks: 261. Improved: $\large\color{#35bf28}34$. Worsened: $\large\color{#d91a1a}23$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_plain_set_nested 34.3310μs 14.8770μs 67.2179 KOps/s 67.6161 KOps/s $\color{#d91a1a}-0.59\%$
test_plain_set_stack_nested 55.4220μs 15.1715μs 65.9132 KOps/s 66.3310 KOps/s $\color{#d91a1a}-0.63\%$
test_plain_set_nested_inplace 38.7310μs 16.6772μs 59.9622 KOps/s 59.8046 KOps/s $\color{#35bf28}+0.26\%$
test_plain_set_stack_nested_inplace 53.4310μs 16.4693μs 60.7190 KOps/s 60.4289 KOps/s $\color{#35bf28}+0.48\%$
test_items 31.2900μs 6.0334μs 165.7453 KOps/s 168.6230 KOps/s $\color{#d91a1a}-1.71\%$
test_items_nested 0.5417ms 0.4727ms 2.1155 KOps/s 2.1477 KOps/s $\color{#d91a1a}-1.50\%$
test_items_nested_locked 0.5563ms 0.4767ms 2.0977 KOps/s 2.1240 KOps/s $\color{#d91a1a}-1.24\%$
test_items_nested_leaf 0.1439ms 97.8602μs 10.2187 KOps/s 10.2757 KOps/s $\color{#d91a1a}-0.55\%$
test_items_stack_nested 0.5662ms 0.4741ms 2.1092 KOps/s 2.1637 KOps/s $\color{#d91a1a}-2.52\%$
test_items_stack_nested_leaf 0.1377ms 97.3675μs 10.2704 KOps/s 10.3851 KOps/s $\color{#d91a1a}-1.11\%$
test_items_stack_nested_locked 0.5606ms 0.4708ms 2.1239 KOps/s 2.1085 KOps/s $\color{#35bf28}+0.73\%$
test_keys 29.4400μs 4.2276μs 236.5389 KOps/s 236.5276 KOps/s $+0.00\%$
test_keys_nested 0.1738ms 0.1306ms 7.6541 KOps/s 7.6406 KOps/s $\color{#35bf28}+0.18\%$
test_keys_nested_locked 2.5106ms 0.1392ms 7.1861 KOps/s 7.1424 KOps/s $\color{#35bf28}+0.61\%$
test_keys_nested_leaf 0.1676ms 0.1211ms 8.2555 KOps/s 8.3000 KOps/s $\color{#d91a1a}-0.54\%$
test_keys_stack_nested 0.1682ms 0.1310ms 7.6359 KOps/s 7.6619 KOps/s $\color{#d91a1a}-0.34\%$
test_keys_stack_nested_leaf 0.1686ms 0.1215ms 8.2324 KOps/s 8.2664 KOps/s $\color{#d91a1a}-0.41\%$
test_keys_stack_nested_locked 0.2146ms 0.1401ms 7.1382 KOps/s 7.1852 KOps/s $\color{#d91a1a}-0.65\%$
test_values 5.6482μs 1.0221μs 978.3937 KOps/s 975.1464 KOps/s $\color{#35bf28}+0.33\%$
test_values_nested 82.8320μs 52.4534μs 19.0646 KOps/s 18.9550 KOps/s $\color{#35bf28}+0.58\%$
test_values_nested_locked 87.5320μs 55.8414μs 17.9079 KOps/s 17.9457 KOps/s $\color{#d91a1a}-0.21\%$
test_values_nested_leaf 94.1120μs 59.9859μs 16.6706 KOps/s 16.5212 KOps/s $\color{#35bf28}+0.90\%$
test_values_stack_nested 0.1294ms 52.2834μs 19.1265 KOps/s 18.9554 KOps/s $\color{#35bf28}+0.90\%$
test_values_stack_nested_leaf 95.6620μs 60.3974μs 16.5570 KOps/s 16.5693 KOps/s $\color{#d91a1a}-0.07\%$
test_values_stack_nested_locked 82.9620μs 55.9149μs 17.8843 KOps/s 17.7688 KOps/s $\color{#35bf28}+0.65\%$
test_membership 4.5118μs 0.8328μs 1.2008 MOps/s 1.1732 MOps/s $\color{#35bf28}+2.36\%$
test_membership_nested 22.5900μs 2.8759μs 347.7176 KOps/s 347.8562 KOps/s $\color{#d91a1a}-0.04\%$
test_membership_nested_leaf 32.8410μs 2.8771μs 347.5748 KOps/s 343.9731 KOps/s $\color{#35bf28}+1.05\%$
test_membership_stacked_nested 29.0210μs 2.8761μs 347.6946 KOps/s 345.1299 KOps/s $\color{#35bf28}+0.74\%$
test_membership_stacked_nested_leaf 27.9210μs 2.8970μs 345.1819 KOps/s 348.2886 KOps/s $\color{#d91a1a}-0.89\%$
test_membership_nested_last 37.9310μs 4.3023μs 232.4338 KOps/s 229.6422 KOps/s $\color{#35bf28}+1.22\%$
test_membership_nested_leaf_last 32.1910μs 4.3250μs 231.2151 KOps/s 229.3939 KOps/s $\color{#35bf28}+0.79\%$
test_membership_stacked_nested_last 35.6010μs 4.2993μs 232.5983 KOps/s 227.4833 KOps/s $\color{#35bf28}+2.25\%$
test_membership_stacked_nested_leaf_last 39.0910μs 4.2909μs 233.0516 KOps/s 231.2151 KOps/s $\color{#35bf28}+0.79\%$
test_nested_getleaf 93.3820μs 21.3701μs 46.7942 KOps/s 45.8193 KOps/s $\color{#35bf28}+2.13\%$
test_nested_get 53.6110μs 20.3995μs 49.0209 KOps/s 49.1394 KOps/s $\color{#d91a1a}-0.24\%$
test_stacked_getleaf 67.2520μs 21.5409μs 46.4234 KOps/s 46.9049 KOps/s $\color{#d91a1a}-1.03\%$
test_stacked_get 53.5220μs 20.4100μs 48.9957 KOps/s 48.9963 KOps/s $-0.00\%$
test_nested_getitemleaf 60.8310μs 22.1089μs 45.2306 KOps/s 45.1822 KOps/s $\color{#35bf28}+0.11\%$
test_nested_getitem 80.0420μs 20.9134μs 47.8163 KOps/s 47.5746 KOps/s $\color{#35bf28}+0.51\%$
test_stacked_getitemleaf 50.0110μs 22.4489μs 44.5455 KOps/s 45.4752 KOps/s $\color{#d91a1a}-2.04\%$
test_stacked_getitem 53.0610μs 20.8936μs 47.8615 KOps/s 48.3638 KOps/s $\color{#d91a1a}-1.04\%$
test_lock_nested 5.1805ms 0.4820ms 2.0748 KOps/s 2.0741 KOps/s $\color{#35bf28}+0.04\%$
test_lock_stack_nested 0.5736ms 0.4828ms 2.0711 KOps/s 2.0353 KOps/s $\color{#35bf28}+1.76\%$
test_unlock_nested 0.5577ms 0.3888ms 2.5720 KOps/s 2.5300 KOps/s $\color{#35bf28}+1.66\%$
test_unlock_stack_nested 0.4954ms 0.3921ms 2.5506 KOps/s 2.4941 KOps/s $\color{#35bf28}+2.27\%$
test_flatten_speed 0.1675ms 0.1228ms 8.1412 KOps/s 8.1797 KOps/s $\color{#d91a1a}-0.47\%$
test_unflatten_speed 0.7726ms 0.5690ms 1.7574 KOps/s 1.7185 KOps/s $\color{#35bf28}+2.26\%$
test_common_ops 99.1333ms 1.0220ms 978.5088 Ops/s 210.4040 Ops/s $\textbf{\color{#35bf28}+365.06\%}$
test_creation 98.6620μs 3.1293μs 319.5587 KOps/s 329.8549 KOps/s $\color{#d91a1a}-3.12\%$
test_creation_empty 31.3710μs 7.0029μs 142.7970 KOps/s 143.0541 KOps/s $\color{#d91a1a}-0.18\%$
test_creation_nested_1 61.5710μs 11.5328μs 86.7089 KOps/s 87.3349 KOps/s $\color{#d91a1a}-0.72\%$
test_creation_nested_2 45.0910μs 13.4004μs 74.6247 KOps/s 74.4675 KOps/s $\color{#35bf28}+0.21\%$
test_creation_many_keys[10] 55.2810μs 21.1674μs 47.2425 KOps/s 46.8451 KOps/s $\color{#35bf28}+0.85\%$
test_creation_many_keys[50] 0.1214ms 90.4476μs 11.0561 KOps/s 10.9659 KOps/s $\color{#35bf28}+0.82\%$
test_creation_many_keys[100] 0.2288ms 0.1771ms 5.6468 KOps/s 5.5740 KOps/s $\color{#35bf28}+1.31\%$
test_creation_nested_many_keys[10] 81.8820μs 45.2251μs 22.1116 KOps/s 22.1538 KOps/s $\color{#d91a1a}-0.19\%$
test_creation_nested_many_keys[50] 0.2206ms 0.1850ms 5.4063 KOps/s 5.3946 KOps/s $\color{#35bf28}+0.22\%$
test_clone 61.3820μs 13.3298μs 75.0200 KOps/s 72.9230 KOps/s $\color{#35bf28}+2.88\%$
test_getitem[int] 1.5277ms 15.2586μs 65.5368 KOps/s 55.0869 KOps/s $\textbf{\color{#35bf28}+18.97\%}$
test_getitem[slice_int] 0.1477ms 24.7472μs 40.4086 KOps/s 38.2991 KOps/s $\textbf{\color{#35bf28}+5.51\%}$
test_getitem[range] 0.1708ms 62.9920μs 15.8750 KOps/s 15.3443 KOps/s $\color{#35bf28}+3.46\%$
test_getitem[tuple] 0.1465ms 24.4879μs 40.8365 KOps/s 39.8603 KOps/s $\color{#35bf28}+2.45\%$
test_getitem[list] 0.2224ms 58.4482μs 17.1092 KOps/s 16.6384 KOps/s $\color{#35bf28}+2.83\%$
test_setitem_dim[int] 45.9910μs 26.5668μs 37.6410 KOps/s 36.8731 KOps/s $\color{#35bf28}+2.08\%$
test_setitem_dim[slice_int] 65.7120μs 42.6461μs 23.4488 KOps/s 22.6021 KOps/s $\color{#35bf28}+3.75\%$
test_setitem_dim[range] 0.1228ms 94.6326μs 10.5672 KOps/s 10.3884 KOps/s $\color{#35bf28}+1.72\%$
test_setitem_dim[tuple] 72.1320μs 39.0563μs 25.6041 KOps/s 24.4383 KOps/s $\color{#35bf28}+4.77\%$
test_setitem 68.9310μs 17.6916μs 56.5240 KOps/s 53.3774 KOps/s $\textbf{\color{#35bf28}+5.90\%}$
test_set 79.5920μs 17.1060μs 58.4590 KOps/s 56.2257 KOps/s $\color{#35bf28}+3.97\%$
test_set_shared 0.4868ms 0.2079ms 4.8105 KOps/s 4.8603 KOps/s $\color{#d91a1a}-1.02\%$
test_update 0.1928ms 22.7380μs 43.9793 KOps/s 43.6593 KOps/s $\color{#35bf28}+0.73\%$
test_update_nested 90.5320μs 35.4628μs 28.1986 KOps/s 28.3194 KOps/s $\color{#d91a1a}-0.43\%$
test_update__nested 0.5256ms 35.7314μs 27.9866 KOps/s 28.5343 KOps/s $\color{#d91a1a}-1.92\%$
test_set_nested 73.2120μs 20.0208μs 49.9481 KOps/s 50.4177 KOps/s $\color{#d91a1a}-0.93\%$
test_set_nested_new 74.5820μs 25.8310μs 38.7131 KOps/s 39.4410 KOps/s $\color{#d91a1a}-1.85\%$
test_select 93.5620μs 42.7696μs 23.3811 KOps/s 23.7866 KOps/s $\color{#d91a1a}-1.71\%$
test_select_nested 0.1062ms 73.9955μs 13.5143 KOps/s 13.4864 KOps/s $\color{#35bf28}+0.21\%$
test_exclude_nested 0.1295ms 92.6484μs 10.7935 KOps/s 10.8646 KOps/s $\color{#d91a1a}-0.65\%$
test_empty[True] 0.5370ms 0.3977ms 2.5144 KOps/s 2.5104 KOps/s $\color{#35bf28}+0.16\%$
test_empty[False] 7.6575μs 1.3210μs 756.9956 KOps/s 760.3614 KOps/s $\color{#d91a1a}-0.44\%$
test_to 0.1127ms 80.0484μs 12.4924 KOps/s 13.1545 KOps/s $\textbf{\color{#d91a1a}-5.03\%}$
test_to_nonblocking 1.2548ms 71.9649μs 13.8957 KOps/s 14.3200 KOps/s $\color{#d91a1a}-2.96\%$
test_unbind_speed 0.4151ms 0.3344ms 2.9905 KOps/s 2.9444 KOps/s $\color{#35bf28}+1.56\%$
test_unbind_speed_stack0 0.4375ms 0.3322ms 3.0101 KOps/s 2.9582 KOps/s $\color{#35bf28}+1.76\%$
test_unbind_speed_stack1 0.1058s 0.9193ms 1.0878 KOps/s 1.1706 KOps/s $\textbf{\color{#d91a1a}-7.07\%}$
test_split 1.2669ms 1.1411ms 876.3573 Ops/s 787.8774 Ops/s $\textbf{\color{#35bf28}+11.23\%}$
test_chunk 0.1058s 1.2158ms 822.5098 Ops/s 910.3761 Ops/s $\textbf{\color{#d91a1a}-9.65\%}$
test_to_cpu_blocking 20.7819ms 20.1921ms 49.5244 Ops/s 34.1653 Ops/s $\textbf{\color{#35bf28}+44.96\%}$
test_to_cpu_global_sync 11.9402ms 11.8173ms 84.6215 Ops/s 73.8688 Ops/s $\textbf{\color{#35bf28}+14.56\%}$
test_to_cpu_event_sync 13.2534ms 12.8156ms 78.0298 Ops/s 75.7696 Ops/s $\color{#35bf28}+2.98\%$
test_to_cpu_default 0.1184s 14.2155ms 70.3459 Ops/s 75.7187 Ops/s $\textbf{\color{#d91a1a}-7.10\%}$
test_consolidate[False-None] 4.5808ms 4.1253ms 242.4055 Ops/s 236.6137 Ops/s $\color{#35bf28}+2.45\%$
test_consolidate[default-None] 2.1322ms 2.0471ms 488.4878 Ops/s 458.7407 Ops/s $\textbf{\color{#35bf28}+6.48\%}$
test_consolidate[reduce-overhead-None] 2.0647ms 1.9604ms 510.1039 Ops/s 485.2324 Ops/s $\textbf{\color{#35bf28}+5.13\%}$
test_consolidate_njt[False-None] 8.7344ms 8.4798ms 117.9270 Ops/s 114.2608 Ops/s $\color{#35bf28}+3.21\%$
test_to[False-False-None] 2.2510ms 2.1720ms 460.3977 Ops/s 440.8689 Ops/s $\color{#35bf28}+4.43\%$
test_to[True-False-None] 2.1697ms 1.9006ms 526.1630 Ops/s 507.2579 Ops/s $\color{#35bf28}+3.73\%$
test_to[within-False-None] 6.2095ms 6.1073ms 163.7386 Ops/s 157.6770 Ops/s $\color{#35bf28}+3.84\%$
test_to[True-default-None] 9.7474ms 9.2564ms 108.0332 Ops/s 105.8602 Ops/s $\color{#35bf28}+2.05\%$
test_to_njt[False-False-None] 8.7376ms 8.6033ms 116.2339 Ops/s 113.7033 Ops/s $\color{#35bf28}+2.23\%$
test_to_njt[True-False-None] 7.1359ms 7.0363ms 142.1197 Ops/s 140.6600 Ops/s $\color{#35bf28}+1.04\%$
test_to_njt[within-False-None] 15.7970ms 15.5868ms 64.1569 Ops/s 62.3782 Ops/s $\color{#35bf28}+2.85\%$
test_creation[device0] 0.4516ms 0.1152ms 8.6790 KOps/s 8.6505 KOps/s $\color{#35bf28}+0.33\%$
test_creation_from_tensor 0.4183ms 0.1131ms 8.8409 KOps/s 8.8733 KOps/s $\color{#d91a1a}-0.37\%$
test_add_one[memmap_tensor0] 0.3338ms 6.5506μs 152.6584 KOps/s 143.9172 KOps/s $\textbf{\color{#35bf28}+6.07\%}$
test_contiguous[memmap_tensor0] 23.9600μs 0.6451μs 1.5501 MOps/s 2.2611 MOps/s $\textbf{\color{#d91a1a}-31.44\%}$
test_stack[memmap_tensor0] 23.9310μs 4.7373μs 211.0900 KOps/s 210.7710 KOps/s $\color{#35bf28}+0.15\%$
test_memmaptd_index 1.1570ms 0.2673ms 3.7406 KOps/s 3.5936 KOps/s $\color{#35bf28}+4.09\%$
test_memmaptd_index_astensor 0.5516ms 0.3730ms 2.6807 KOps/s 2.6378 KOps/s $\color{#35bf28}+1.63\%$
test_memmaptd_index_op 0.7970ms 0.6206ms 1.6114 KOps/s 1.5467 KOps/s $\color{#35bf28}+4.18\%$
test_serialize_model 0.1389s 0.1354s 7.3876 Ops/s 7.3865 Ops/s $\color{#35bf28}+0.01\%$
test_serialize_model_pickle 1.3484s 1.1917s 0.8391 Ops/s 0.8262 Ops/s $\color{#35bf28}+1.56\%$
test_serialize_weights 0.1378s 0.1355s 7.3826 Ops/s 3.5108 Ops/s $\textbf{\color{#35bf28}+110.28\%}$
test_serialize_weights_returnearly 0.4705s 89.1660ms 11.2150 Ops/s 15.8722 Ops/s $\textbf{\color{#d91a1a}-29.34\%}$
test_serialize_weights_pickle 1.3647s 1.1987s 0.8342 Ops/s 0.8230 Ops/s $\color{#35bf28}+1.36\%$
test_reshape_pytree 0.2041ms 32.7757μs 30.5104 KOps/s 30.6680 KOps/s $\color{#d91a1a}-0.51\%$
test_reshape_td 97.2720μs 47.9337μs 20.8621 KOps/s 21.8244 KOps/s $\color{#d91a1a}-4.41\%$
test_view_pytree 0.2196ms 32.1603μs 31.0942 KOps/s 30.9512 KOps/s $\color{#35bf28}+0.46\%$
test_view_td 0.1027ms 54.7594μs 18.2617 KOps/s 18.2985 KOps/s $\color{#d91a1a}-0.20\%$
test_unbind_pytree 0.2649ms 36.0398μs 27.7471 KOps/s 27.1496 KOps/s $\color{#35bf28}+2.20\%$
test_unbind_td 0.1410ms 50.1921μs 19.9235 KOps/s 19.6410 KOps/s $\color{#35bf28}+1.44\%$
test_split_pytree 0.2455ms 42.3140μs 23.6328 KOps/s 23.6161 KOps/s $\color{#35bf28}+0.07\%$
test_split_td 0.2179ms 65.3832μs 15.2945 KOps/s 15.4735 KOps/s $\color{#d91a1a}-1.16\%$
test_add_pytree 0.2249ms 43.3262μs 23.0807 KOps/s 23.8030 KOps/s $\color{#d91a1a}-3.03\%$
test_add_td 0.1026ms 59.7244μs 16.7436 KOps/s 17.6489 KOps/s $\textbf{\color{#d91a1a}-5.13\%}$
test_compile_add_one_nested[tensordict-compile] 0.2767ms 0.1694ms 5.9039 KOps/s 5.7527 KOps/s $\color{#35bf28}+2.63\%$
test_compile_add_one_nested[tensordict-eager] 0.2883ms 0.2012ms 4.9695 KOps/s 4.9503 KOps/s $\color{#35bf28}+0.39\%$
test_compile_add_one_nested[pytree-compile] 0.2222ms 0.1289ms 7.7568 KOps/s 7.6698 KOps/s $\color{#35bf28}+1.13\%$
test_compile_add_one_nested[pytree-eager] 0.4313ms 0.1803ms 5.5475 KOps/s 5.4238 KOps/s $\color{#35bf28}+2.28\%$
test_compile_copy_nested[tensordict-compile] 75.0220μs 17.4903μs 57.1747 KOps/s 60.8367 KOps/s $\textbf{\color{#d91a1a}-6.02\%}$
test_compile_copy_nested[tensordict-eager] 0.4056ms 52.9854μs 18.8731 KOps/s 17.9797 KOps/s $\color{#35bf28}+4.97\%$
test_compile_copy_nested[pytree-compile] 0.1292ms 15.8395μs 63.1332 KOps/s 63.1692 KOps/s $\color{#d91a1a}-0.06\%$
test_compile_copy_nested[pytree-eager] 0.3631ms 67.5536μs 14.8031 KOps/s 14.8943 KOps/s $\color{#d91a1a}-0.61\%$
test_compile_add_one_flat[tensordict-compile] 0.2914ms 0.2003ms 4.9932 KOps/s 4.7923 KOps/s $\color{#35bf28}+4.19\%$
test_compile_add_one_flat[tensordict-eager] 0.3221ms 0.2802ms 3.5695 KOps/s 3.5116 KOps/s $\color{#35bf28}+1.65\%$
test_compile_add_one_flat[tensorclass-compile] 0.2121ms 0.1357ms 7.3672 KOps/s 7.1499 KOps/s $\color{#35bf28}+3.04\%$
test_compile_add_one_flat[tensorclass-eager] 0.1171ms 74.0076μs 13.5121 KOps/s 12.9625 KOps/s $\color{#35bf28}+4.24\%$
test_compile_add_one_flat[pytree-compile] 0.2429ms 0.1785ms 5.6026 KOps/s 5.3958 KOps/s $\color{#35bf28}+3.83\%$
test_compile_add_one_flat[pytree-eager] 0.8172ms 0.5282ms 1.8934 KOps/s 1.8207 KOps/s $\color{#35bf28}+3.99\%$
test_compile_add_self_flat[tensordict-eager] 0.4133ms 0.3302ms 3.0285 KOps/s 2.9740 KOps/s $\color{#35bf28}+1.83\%$
test_compile_add_self_flat[tensordict-compile] 0.2516ms 0.1991ms 5.0214 KOps/s 4.7613 KOps/s $\textbf{\color{#35bf28}+5.46\%}$
test_compile_add_self_flat[tensorclass-eager] 0.1481ms 90.5533μs 11.0432 KOps/s 10.8535 KOps/s $\color{#35bf28}+1.75\%$
test_compile_add_self_flat[tensorclass-compile] 0.1830ms 0.1372ms 7.2886 KOps/s 6.9979 KOps/s $\color{#35bf28}+4.15\%$
test_compile_add_self_flat[pytree-eager] 0.6551ms 0.4433ms 2.2560 KOps/s 2.2193 KOps/s $\color{#35bf28}+1.65\%$
test_compile_add_self_flat[pytree-compile] 0.2253ms 0.1800ms 5.5557 KOps/s 5.3588 KOps/s $\color{#35bf28}+3.68\%$
test_compile_copy_flat[tensordict-compile] 0.1333ms 19.7673μs 50.5887 KOps/s 48.4724 KOps/s $\color{#35bf28}+4.37\%$
test_compile_copy_flat[tensordict-eager] 74.2920μs 41.3403μs 24.1895 KOps/s 21.6935 KOps/s $\textbf{\color{#35bf28}+11.51\%}$
test_compile_copy_flat[pytree-compile] 0.2057ms 16.7858μs 59.5741 KOps/s 57.3875 KOps/s $\color{#35bf28}+3.81\%$
test_compile_copy_flat[pytree-eager] 0.3584ms 52.5781μs 19.0193 KOps/s 18.9990 KOps/s $\color{#35bf28}+0.11\%$
test_compile_assign_and_add[tensordict-compile] 2.0858ms 0.1902ms 5.2573 KOps/s 4.8321 KOps/s $\textbf{\color{#35bf28}+8.80\%}$
test_compile_assign_and_add[tensordict-eager] 3.5089ms 3.3452ms 298.9362 Ops/s 289.9476 Ops/s $\color{#35bf28}+3.10\%$
test_compile_assign_and_add[pytree-compile] 2.0816ms 0.1765ms 5.6662 KOps/s 5.4180 KOps/s $\color{#35bf28}+4.58\%$
test_compile_assign_and_add[pytree-eager] 2.9473ms 2.8150ms 355.2428 Ops/s 344.7518 Ops/s $\color{#35bf28}+3.04\%$
test_compile_indexing[tensor-tensordict-compile] 0.2422ms 0.1268ms 7.8864 KOps/s 7.4772 KOps/s $\textbf{\color{#35bf28}+5.47\%}$
test_compile_indexing[tensor-tensordict-eager] 0.2928ms 73.6295μs 13.5815 KOps/s 13.0479 KOps/s $\color{#35bf28}+4.09\%$
test_compile_indexing[tensor-tensorclass-compile] 0.2156ms 0.1132ms 8.8322 KOps/s 8.4831 KOps/s $\color{#35bf28}+4.12\%$
test_compile_indexing[tensor-tensorclass-eager] 0.2466ms 43.9392μs 22.7587 KOps/s 21.0540 KOps/s $\textbf{\color{#35bf28}+8.10\%}$
test_compile_indexing[tensor-pytree-compile] 0.1520ms 0.1140ms 8.7704 KOps/s 8.0711 KOps/s $\textbf{\color{#35bf28}+8.66\%}$
test_compile_indexing[tensor-pytree-eager] 0.2406ms 44.0280μs 22.7128 KOps/s 20.5053 KOps/s $\textbf{\color{#35bf28}+10.77\%}$
test_compile_indexing[slice-tensordict-compile] 0.2018ms 70.1159μs 14.2621 KOps/s 13.4208 KOps/s $\textbf{\color{#35bf28}+6.27\%}$
test_compile_indexing[slice-tensordict-eager] 0.2068ms 27.3663μs 36.5413 KOps/s 34.9114 KOps/s $\color{#35bf28}+4.67\%$
test_compile_indexing[slice-tensorclass-compile] 0.1594ms 57.4788μs 17.3977 KOps/s 16.6900 KOps/s $\color{#35bf28}+4.24\%$
test_compile_indexing[slice-tensorclass-eager] 0.2438ms 22.1167μs 45.2147 KOps/s 44.5604 KOps/s $\color{#35bf28}+1.47\%$
test_compile_indexing[slice-pytree-compile] 91.0320μs 58.6900μs 17.0387 KOps/s 15.8997 KOps/s $\textbf{\color{#35bf28}+7.16\%}$
test_compile_indexing[slice-pytree-eager] 0.2716ms 21.7058μs 46.0705 KOps/s 44.2311 KOps/s $\color{#35bf28}+4.16\%$
test_compile_indexing[int-tensordict-compile] 0.1098ms 69.8928μs 14.3076 KOps/s 13.1161 KOps/s $\textbf{\color{#35bf28}+9.08\%}$
test_compile_indexing[int-tensordict-eager] 0.2015ms 27.6534μs 36.1619 KOps/s 35.1518 KOps/s $\color{#35bf28}+2.87\%$
test_compile_indexing[int-tensorclass-compile] 0.1247ms 58.0927μs 17.2139 KOps/s 16.3831 KOps/s $\textbf{\color{#35bf28}+5.07\%}$
test_compile_indexing[int-tensorclass-eager] 0.2639ms 21.8067μs 45.8575 KOps/s 44.5438 KOps/s $\color{#35bf28}+2.95\%$
test_compile_indexing[int-pytree-compile] 0.2382ms 58.9446μs 16.9651 KOps/s 16.4438 KOps/s $\color{#35bf28}+3.17\%$
test_compile_indexing[int-pytree-eager] 0.2509ms 21.7797μs 45.9143 KOps/s 44.8330 KOps/s $\color{#35bf28}+2.41\%$
test_compile_replace[single-eager] 90.1720μs 47.5865μs 21.0144 KOps/s 20.0267 KOps/s $\color{#35bf28}+4.93\%$
test_compile_replace[single-compile] 0.1646ms 0.1217ms 8.2170 KOps/s 7.4377 KOps/s $\textbf{\color{#35bf28}+10.48\%}$
test_compile_replace[multi-eager] 0.6081ms 0.5610ms 1.7825 KOps/s 1.6618 KOps/s $\textbf{\color{#35bf28}+7.27\%}$
test_compile_replace[multi-compile] 0.2705ms 0.1285ms 7.7802 KOps/s 7.5096 KOps/s $\color{#35bf28}+3.60\%$
test_compile_tc_getattr_20[eager] 0.2198ms 0.1730ms 5.7811 KOps/s 5.8031 KOps/s $\color{#d91a1a}-0.38\%$
test_compile_tc_getattr_20[compile] 0.3208ms 0.1370ms 7.2983 KOps/s 6.9506 KOps/s $\textbf{\color{#35bf28}+5.00\%}$
test_compile_clone_shallow[20-eager] 51.6610μs 19.2019μs 52.0781 KOps/s 47.2106 KOps/s $\textbf{\color{#35bf28}+10.31\%}$
test_compile_clone_shallow[20-compile] 57.2820μs 17.5378μs 57.0197 KOps/s 53.6859 KOps/s $\textbf{\color{#35bf28}+6.21\%}$
test_compile_clone_shallow[40-eager] 61.8810μs 33.8828μs 29.5135 KOps/s 27.0746 KOps/s $\textbf{\color{#35bf28}+9.01\%}$
test_compile_clone_shallow[40-compile] 78.8820μs 18.8329μs 53.0986 KOps/s 50.6080 KOps/s $\color{#35bf28}+4.92\%$
test_compile_clone_shallow[80-eager] 99.7130μs 62.9103μs 15.8957 KOps/s 14.2778 KOps/s $\textbf{\color{#35bf28}+11.33\%}$
test_compile_clone_shallow[80-compile] 0.1113ms 20.8382μs 47.9888 KOps/s 44.9974 KOps/s $\textbf{\color{#35bf28}+6.65\%}$
test_compile_update_inplace[eager] 99.4320μs 60.4039μs 16.5552 KOps/s 16.3709 KOps/s $\color{#35bf28}+1.13\%$
test_compile_update_inplace[compile] 0.2022ms 0.1525ms 6.5569 KOps/s 6.0719 KOps/s $\textbf{\color{#35bf28}+7.99\%}$
test_mod_add[eager] 0.2684ms 0.2246ms 4.4524 KOps/s 19.0970 KOps/s $\textbf{\color{#d91a1a}-76.69\%}$
test_mod_add[compile] 0.4915ms 0.1182ms 8.4603 KOps/s 7.8440 KOps/s $\textbf{\color{#35bf28}+7.86\%}$
test_mod_add[compile-overhead] 0.3115ms 0.1673ms 5.9786 KOps/s 5.8094 KOps/s $\color{#35bf28}+2.91\%$
test_mod_wrap[eager] 0.5523ms 0.4743ms 2.1083 KOps/s 3.2209 KOps/s $\textbf{\color{#d91a1a}-34.54\%}$
test_mod_wrap[compile] 0.4285ms 0.3681ms 2.7165 KOps/s 2.5295 KOps/s $\textbf{\color{#35bf28}+7.39\%}$
test_mod_wrap[compile-overhead] 9.2980ms 5.0450ms 198.2144 Ops/s 201.0698 Ops/s $\color{#d91a1a}-1.42\%$
test_mod_wrap_and_backward[eager] 2.1206ms 1.7229ms 580.4177 Ops/s 607.9304 Ops/s $\color{#d91a1a}-4.53\%$
test_mod_wrap_and_backward[compile] 2.5974ms 1.5675ms 637.9427 Ops/s 617.2110 Ops/s $\color{#35bf28}+3.36\%$
test_mod_wrap_and_backward[compile-overhead] 1.3926ms 1.0118ms 988.3257 Ops/s 941.9607 Ops/s $\color{#35bf28}+4.92\%$
test_seq_add[eager] 0.5858ms 0.5060ms 1.9762 KOps/s 6.3422 KOps/s $\textbf{\color{#d91a1a}-68.84\%}$
test_seq_add[compile] 0.2168ms 0.1343ms 7.4484 KOps/s 7.3734 KOps/s $\color{#35bf28}+1.02\%$
test_seq_add[compile-overhead] 0.2230ms 0.1740ms 5.7470 KOps/s 5.5378 KOps/s $\color{#35bf28}+3.78\%$
test_seq_wrap[eager] 1.5071ms 1.4154ms 706.5330 Ops/s 1.8636 KOps/s $\textbf{\color{#d91a1a}-62.09\%}$
test_seq_wrap[compile] 0.4549ms 0.4021ms 2.4872 KOps/s 2.4916 KOps/s $\color{#d91a1a}-0.18\%$
test_seq_wrap[compile-overhead] 0.3593ms 0.2881ms 3.4708 KOps/s 3.3420 KOps/s $\color{#35bf28}+3.85\%$
test_func_call_runtime[False-eager] 0.9510ms 0.8535ms 1.1717 KOps/s 1.1585 KOps/s $\color{#35bf28}+1.13\%$
test_func_call_runtime[False-compile] 1.0326ms 0.9390ms 1.0649 KOps/s 1.0480 KOps/s $\color{#35bf28}+1.62\%$
test_func_call_runtime[False-compile-overhead] 0.5865ms 0.4917ms 2.0338 KOps/s 1.9915 KOps/s $\color{#35bf28}+2.12\%$
test_func_call_runtime[True-eager] 1.2317ms 1.0936ms 914.4377 Ops/s 899.3226 Ops/s $\color{#35bf28}+1.68\%$
test_func_call_runtime[True-compile] 1.0167ms 0.9382ms 1.0659 KOps/s 1.0347 KOps/s $\color{#35bf28}+3.02\%$
test_func_call_runtime[True-compile-overhead] 0.5996ms 0.5053ms 1.9790 KOps/s 1.9312 KOps/s $\color{#35bf28}+2.47\%$
test_func_call_cm_runtime[False-eager] 0.9671ms 0.8892ms 1.1246 KOps/s 1.0968 KOps/s $\color{#35bf28}+2.54\%$
test_func_call_cm_runtime[False-compile] 1.1124ms 0.9032ms 1.1072 KOps/s 1.0718 KOps/s $\color{#35bf28}+3.30\%$
test_func_call_cm_runtime[False-compile-overhead] 0.5417ms 0.4693ms 2.1310 KOps/s 2.0810 KOps/s $\color{#35bf28}+2.41\%$
test_func_call_cm_runtime[True-eager] 1.3247ms 1.2255ms 815.9926 Ops/s 797.0986 Ops/s $\color{#35bf28}+2.37\%$
test_func_call_cm_runtime[True-compile] 1.0427ms 0.9447ms 1.0586 KOps/s 1.0212 KOps/s $\color{#35bf28}+3.66\%$
test_func_call_cm_runtime[True-compile-overhead] 0.5602ms 0.5148ms 1.9426 KOps/s 1.9004 KOps/s $\color{#35bf28}+2.22\%$
test_vmap_func_call_cm_runtime[eager] 2.9016ms 2.3906ms 418.3036 Ops/s 411.7903 Ops/s $\color{#35bf28}+1.58\%$
test_vmap_func_call_cm_runtime[compile] 1.0775ms 0.9901ms 1.0100 KOps/s 984.2923 Ops/s $\color{#35bf28}+2.62\%$
test_vmap_func_call_cm_runtime[compile-overhead] 0.6395ms 0.5188ms 1.9277 KOps/s 1.8707 KOps/s $\color{#35bf28}+3.05\%$
test_distributed 0.4692ms 0.1563ms 6.3988 KOps/s 6.3305 KOps/s $\color{#35bf28}+1.08\%$
test_tdmodule 0.6998ms 0.1921ms 5.2043 KOps/s 33.8087 KOps/s $\textbf{\color{#d91a1a}-84.61\%}$
test_tdmodule_dispatch 0.4011ms 0.3640ms 2.7474 KOps/s 21.5191 KOps/s $\textbf{\color{#d91a1a}-87.23\%}$
test_tdseq 47.7810μs 27.0956μs 36.9064 KOps/s 35.3018 KOps/s $\color{#35bf28}+4.55\%$
test_tdseq_dispatch 0.2494ms 0.2144ms 4.6634 KOps/s 19.8232 KOps/s $\textbf{\color{#d91a1a}-76.48\%}$
test_instantiation_functorch 2.1635ms 2.0656ms 484.1288 Ops/s 475.3664 Ops/s $\color{#35bf28}+1.84\%$
test_exec_functorch 0.2326ms 0.1804ms 5.5431 KOps/s 5.4744 KOps/s $\color{#35bf28}+1.25\%$
test_exec_functional_call 0.2178ms 0.1685ms 5.9340 KOps/s 6.0941 KOps/s $\color{#d91a1a}-2.63\%$
test_exec_td_decorator 0.4657ms 0.2396ms 4.1733 KOps/s 4.0653 KOps/s $\color{#35bf28}+2.66\%$
test_vmap_mlp_speed_decorator[True-True] 1.1971ms 1.0099ms 990.1935 Ops/s 1.1523 KOps/s $\textbf{\color{#d91a1a}-14.07\%}$
test_vmap_mlp_speed_decorator[True-False] 1.2552ms 1.0081ms 991.9708 Ops/s 1.1533 KOps/s $\textbf{\color{#d91a1a}-13.99\%}$
test_vmap_mlp_speed_decorator[False-True] 0.9202ms 0.7207ms 1.3876 KOps/s 1.3514 KOps/s $\color{#35bf28}+2.68\%$
test_vmap_mlp_speed_decorator[False-False] 0.9274ms 0.7168ms 1.3950 KOps/s 1.3394 KOps/s $\color{#35bf28}+4.15\%$
test_vmap_transformer_speed_decorator[True-True] 21.6048ms 20.9968ms 47.6264 Ops/s 46.5857 Ops/s $\color{#35bf28}+2.23\%$
test_vmap_transformer_speed_decorator[True-False] 21.8023ms 20.9982ms 47.6231 Ops/s 47.3013 Ops/s $\color{#35bf28}+0.68\%$
test_vmap_transformer_speed_decorator[False-True] 21.5560ms 20.7038ms 48.3004 Ops/s 47.8642 Ops/s $\color{#35bf28}+0.91\%$
test_vmap_transformer_speed_decorator[False-False] 21.5790ms 20.7103ms 48.2851 Ops/s 47.8980 Ops/s $\color{#35bf28}+0.81\%$
test_to_module_speed[True] 2.0692ms 1.4804ms 675.4729 Ops/s 676.9963 Ops/s $\color{#d91a1a}-0.23\%$
test_to_module_speed[False] 1.5636ms 1.4495ms 689.8928 Ops/s 691.8965 Ops/s $\color{#d91a1a}-0.29\%$
test_tc_init 66.2420μs 44.1233μs 22.6637 KOps/s 21.9498 KOps/s $\color{#35bf28}+3.25\%$
test_tc_init_tensor_only 43.1810μs 9.8211μs 101.8215 KOps/s 101.8171 KOps/s $+0.00\%$
test_tc_init_nested 0.1382ms 87.4680μs 11.4328 KOps/s 11.0730 KOps/s $\color{#35bf28}+3.25\%$
test_tc_init_many_fields 47.0910μs 16.7389μs 59.7409 KOps/s 60.3839 KOps/s $\color{#d91a1a}-1.06\%$
test_tc_first_layer_tensor 27.9400μs 1.8420μs 542.8899 KOps/s 553.3850 KOps/s $\color{#d91a1a}-1.90\%$
test_tc_first_layer_tensor_only 2.2545μs 0.3992μs 2.5051 MOps/s 2.5043 MOps/s $\color{#35bf28}+0.03\%$
test_tc_first_layer_tensor_set 25.1710μs 3.9023μs 256.2597 KOps/s 254.5454 KOps/s $\color{#35bf28}+0.67\%$
test_tc_first_layer_tensor_only_set 30.5310μs 3.2478μs 307.9018 KOps/s 302.1141 KOps/s $\color{#35bf28}+1.92\%$
test_tc_first_layer_nontensor 33.2810μs 6.4431μs 155.2048 KOps/s 161.1048 KOps/s $\color{#d91a1a}-3.66\%$
test_tc_second_layer_tensor 24.8310μs 4.4261μs 225.9302 KOps/s 222.0899 KOps/s $\color{#35bf28}+1.73\%$
test_tc_second_layer_nontensor 38.7310μs 8.9172μs 112.1426 KOps/s 113.9098 KOps/s $\color{#d91a1a}-1.55\%$
test_unbind 0.2714s 16.5364ms 60.4726 Ops/s 68.8456 Ops/s $\textbf{\color{#d91a1a}-12.16\%}$
test_full_like 4.6651ms 4.3883ms 227.8764 Ops/s 60.0357 Ops/s $\textbf{\color{#35bf28}+279.57\%}$
test_zeros_like 16.8138ms 16.6900ms 59.9162 Ops/s 228.5028 Ops/s $\textbf{\color{#d91a1a}-73.78\%}$
test_ones_like 17.9106ms 16.7378ms 59.7449 Ops/s 227.8668 Ops/s $\textbf{\color{#d91a1a}-73.78\%}$
test_clone 18.4002ms 17.7571ms 56.3154 Ops/s 153.6706 Ops/s $\textbf{\color{#d91a1a}-63.35\%}$
test_squeeze 0.1157ms 13.8996μs 71.9448 KOps/s 70.8623 KOps/s $\color{#35bf28}+1.53\%$
test_unsqueeze 0.1641ms 0.1154ms 8.6619 KOps/s 8.7616 KOps/s $\color{#d91a1a}-1.14\%$
test_split 0.3967ms 0.1877ms 5.3269 KOps/s 5.3876 KOps/s $\color{#d91a1a}-1.13\%$
test_permute 0.3023ms 0.2140ms 4.6719 KOps/s 4.8434 KOps/s $\color{#d91a1a}-3.54\%$
test_stack 51.8963ms 51.5581ms 19.3956 Ops/s 19.3609 Ops/s $\color{#35bf28}+0.18\%$
test_cat 52.3520ms 51.5477ms 19.3995 Ops/s 19.3972 Ops/s $\color{#35bf28}+0.01\%$
test_sequential_tensordict 0.6627ms 0.5663ms 1.7657 KOps/s 4.5135 KOps/s $\textbf{\color{#d91a1a}-60.88\%}$
test_sequential_graph_module 0.1695ms 0.1180ms 8.4749 KOps/s 8.2327 KOps/s $\color{#35bf28}+2.94\%$
test_nested_tensordict 4.4564ms 0.8143ms 1.2280 KOps/s 3.5010 KOps/s $\textbf{\color{#d91a1a}-64.92\%}$
test_nested_graph_module 0.1922ms 0.1256ms 7.9629 KOps/s 7.7894 KOps/s $\color{#35bf28}+2.23\%$

@github-actions github-actions Bot added the CI label Apr 29, 2026
@vmoens vmoens merged commit 8c46f0c into main Apr 29, 2026
72 of 73 checks passed
@vmoens vmoens deleted the from-data-format branch April 29, 2026 12:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CI CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. Feature New feature tensorclass Test

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant