Skip to content

[Feature] AsyncBatchedCollector: backend params and performance optimizations#3511

Merged
vmoens merged 4 commits intogh/vmoens/242/basefrom
gh/vmoens/242/head
Feb 21, 2026
Merged

[Feature] AsyncBatchedCollector: backend params and performance optimizations#3511
vmoens merged 4 commits intogh/vmoens/242/basefrom
gh/vmoens/242/head

Conversation

@vmoens
Copy link
Copy Markdown
Collaborator

@vmoens vmoens commented Feb 16, 2026 •
edited
Loading

Stack from ghstack (oldest at bottom):

  • Three-tier backend system: backend (global default), env_backend
    (env pool override), policy_backend (transport override), mirroring
    the device parameter pattern.
  • Lock-free SlotTransport: per-env slots with no shared lock, replacing
    ThreadingTransport as the default for in-process threading.
  • min_batch_size parameter for InferenceServer to accumulate requests.
  • Batch drain from result queue (get_nowait after first blocking get).
  • Remove redundant .copy() in ProcessorAsyncEnvPool._env_exec.

Co-authored-by: Cursor cursoragent@cursor.com

@vmoens
Update
50f9bf4 
[ghstack-poisoned]
@pytorch-bot
Copy link
Copy Markdown

pytorch-bot bot commented Feb 16, 2026 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/3511

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures

As of commit 2d201bf with merge base 266e4aa (image):

NEW FAILURES - The following jobs have failed:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

vmoens added a commit that referenced this pull request Feb 16, 2026
@vmoens @cursoragent
[Feature] AsyncBatchedCollector: backend params and performance optim…
c41776c 
…izations

- Three-tier backend system: `backend` (global default), `env_backend`
  (env pool override), `policy_backend` (transport override), mirroring
  the device parameter pattern.
- Lock-free SlotTransport: per-env slots with no shared lock, replacing
  ThreadingTransport as the default for in-process threading.
- min_batch_size parameter for InferenceServer to accumulate requests.
- Batch drain from result queue (get_nowait after first blocking get).
- Remove redundant .copy() in ProcessorAsyncEnvPool._env_exec.

Co-authored-by: Cursor <cursoragent@cursor.com>
ghstack-source-id: 58cc17b
Pull-Request: #3511
This was referenced Feb 12, 2026
Merged
Merged
Merged
@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 Feb 16, 2026
@vmoens vmoens mentioned this pull request Feb 16, 2026
Merged
@github-actions github-actions bot added the Feature New feature label Feb 16, 2026
This was referenced Feb 12, 2026
Merged
Merged
This was referenced Feb 12, 2026
Merged
Merged
@github-actions
Copy link
Copy Markdown
Contributor

github-actions bot commented Feb 16, 2026 •
edited
Loading

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

Total Benchmarks: 173. Improved: $\large\color{#35bf28}13$. Worsened: $\large\color{#d91a1a}12$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_tensor_to_bytestream_speed[pickle] 81.4887μs 80.0912μs 12.4858 KOps/s 12.4521 KOps/s $\color{#35bf28}+0.27\%$
test_tensor_to_bytestream_speed[torch.save] 0.1391ms 0.1384ms 7.2254 KOps/s 7.1783 KOps/s $\color{#35bf28}+0.66\%$
test_tensor_to_bytestream_speed[untyped_storage] 0.1116s 0.1108s 9.0217 Ops/s 9.1212 Ops/s $\color{#d91a1a}-1.09\%$
test_tensor_to_bytestream_speed[numpy] 2.5963μs 2.5892μs 386.2164 KOps/s 391.3210 KOps/s $\color{#d91a1a}-1.30\%$
test_tensor_to_bytestream_speed[safetensors] 36.7086μs 36.4363μs 27.4452 KOps/s 27.5417 KOps/s $\color{#d91a1a}-0.35\%$
test_simple 0.5425s 0.5398s 1.8525 Ops/s 1.7736 Ops/s $\color{#35bf28}+4.44\%$
test_transformed 1.0735s 1.0722s 0.9327 Ops/s 0.9085 Ops/s $\color{#35bf28}+2.66\%$
test_serial 1.6558s 1.6533s 0.6049 Ops/s 0.5920 Ops/s $\color{#35bf28}+2.17\%$
test_parallel 1.1346s 1.0360s 0.9653 Ops/s 0.9256 Ops/s $\color{#35bf28}+4.29\%$
test_step_mdp_speed[True-True-True-True-True] 0.2040ms 41.8115μs 23.9169 KOps/s 23.7501 KOps/s $\color{#35bf28}+0.70\%$
test_step_mdp_speed[True-True-True-True-False] 46.7910μs 23.7276μs 42.1450 KOps/s 42.1206 KOps/s $\color{#35bf28}+0.06\%$
test_step_mdp_speed[True-True-True-False-True] 97.2120μs 23.8172μs 41.9865 KOps/s 42.3279 KOps/s $\color{#d91a1a}-0.81\%$
test_step_mdp_speed[True-True-True-False-False] 46.0900μs 12.7766μs 78.2681 KOps/s 76.6442 KOps/s $\color{#35bf28}+2.12\%$
test_step_mdp_speed[True-True-False-True-True] 72.9810μs 44.8064μs 22.3183 KOps/s 21.9285 KOps/s $\color{#35bf28}+1.78\%$
test_step_mdp_speed[True-True-False-True-False] 66.0310μs 26.2109μs 38.1520 KOps/s 38.1355 KOps/s $\color{#35bf28}+0.04\%$
test_step_mdp_speed[True-True-False-False-True] 55.0310μs 26.3126μs 38.0046 KOps/s 38.0718 KOps/s $\color{#d91a1a}-0.18\%$
test_step_mdp_speed[True-True-False-False-False] 52.8510μs 15.7355μs 63.5504 KOps/s 63.3848 KOps/s $\color{#35bf28}+0.26\%$
test_step_mdp_speed[True-False-True-True-True] 97.8720μs 47.4159μs 21.0900 KOps/s 20.6241 KOps/s $\color{#35bf28}+2.26\%$
test_step_mdp_speed[True-False-True-True-False] 55.3410μs 28.8751μs 34.6319 KOps/s 34.5649 KOps/s $\color{#35bf28}+0.19\%$
test_step_mdp_speed[True-False-True-False-True] 57.0510μs 25.9497μs 38.5361 KOps/s 38.0268 KOps/s $\color{#35bf28}+1.34\%$
test_step_mdp_speed[True-False-True-False-False] 55.6610μs 15.5646μs 64.2485 KOps/s 63.0624 KOps/s $\color{#35bf28}+1.88\%$
test_step_mdp_speed[True-False-False-True-True] 0.1037ms 50.3166μs 19.8742 KOps/s 19.9435 KOps/s $\color{#d91a1a}-0.35\%$
test_step_mdp_speed[True-False-False-True-False] 69.2910μs 30.8664μs 32.3977 KOps/s 31.7892 KOps/s $\color{#35bf28}+1.91\%$
test_step_mdp_speed[True-False-False-False-True] 64.6510μs 28.4550μs 35.1433 KOps/s 35.0585 KOps/s $\color{#35bf28}+0.24\%$
test_step_mdp_speed[True-False-False-False-False] 48.5200μs 18.3010μs 54.6419 KOps/s 54.3177 KOps/s $\color{#35bf28}+0.60\%$
test_step_mdp_speed[False-True-True-True-True] 81.9510μs 47.5646μs 21.0241 KOps/s 20.8614 KOps/s $\color{#35bf28}+0.78\%$
test_step_mdp_speed[False-True-True-True-False] 76.8410μs 28.4597μs 35.1375 KOps/s 34.5828 KOps/s $\color{#35bf28}+1.60\%$
test_step_mdp_speed[False-True-True-False-True] 2.5880ms 30.0594μs 33.2674 KOps/s 33.1331 KOps/s $\color{#35bf28}+0.41\%$
test_step_mdp_speed[False-True-True-False-False] 62.4410μs 17.3667μs 57.5814 KOps/s 57.6287 KOps/s $\color{#d91a1a}-0.08\%$
test_step_mdp_speed[False-True-False-True-True] 79.1610μs 49.7734μs 20.0910 KOps/s 19.8530 KOps/s $\color{#35bf28}+1.20\%$
test_step_mdp_speed[False-True-False-True-False] 62.1410μs 31.3626μs 31.8851 KOps/s 31.6547 KOps/s $\color{#35bf28}+0.73\%$
test_step_mdp_speed[False-True-False-False-True] 63.8510μs 32.1552μs 31.0992 KOps/s 30.3718 KOps/s $\color{#35bf28}+2.39\%$
test_step_mdp_speed[False-True-False-False-False] 45.2210μs 19.6924μs 50.7811 KOps/s 49.9521 KOps/s $\color{#35bf28}+1.66\%$
test_step_mdp_speed[False-False-True-True-True] 87.7220μs 52.2781μs 19.1285 KOps/s 18.5772 KOps/s $\color{#35bf28}+2.97\%$
test_step_mdp_speed[False-False-True-True-False] 91.9220μs 33.7636μs 29.6177 KOps/s 28.9903 KOps/s $\color{#35bf28}+2.16\%$
test_step_mdp_speed[False-False-True-False-True] 0.1161ms 31.1497μs 32.1030 KOps/s 30.7233 KOps/s $\color{#35bf28}+4.49\%$
test_step_mdp_speed[False-False-True-False-False] 83.2210μs 19.6691μs 50.8411 KOps/s 49.9596 KOps/s $\color{#35bf28}+1.76\%$
test_step_mdp_speed[False-False-False-True-True] 92.5310μs 54.1941μs 18.4522 KOps/s 17.7258 KOps/s $\color{#35bf28}+4.10\%$
test_step_mdp_speed[False-False-False-True-False] 78.5410μs 36.1425μs 27.6682 KOps/s 27.6091 KOps/s $\color{#35bf28}+0.21\%$
test_step_mdp_speed[False-False-False-False-True] 64.9910μs 33.7715μs 29.6108 KOps/s 28.6883 KOps/s $\color{#35bf28}+3.22\%$
test_step_mdp_speed[False-False-False-False-False] 49.1310μs 22.1240μs 45.1997 KOps/s 44.4110 KOps/s $\color{#35bf28}+1.78\%$
test_non_tensor_env_rollout_speed[1000-single-True] 0.7127s 0.7092s 1.4100 Ops/s 1.3536 Ops/s $\color{#35bf28}+4.17\%$
test_non_tensor_env_rollout_speed[1000-single-False] 0.6994s 0.6032s 1.6580 Ops/s 1.6521 Ops/s $\color{#35bf28}+0.36\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 1.6993s 1.6233s 0.6160 Ops/s 0.6128 Ops/s $\color{#35bf28}+0.53\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] 1.4780s 1.3990s 0.7148 Ops/s 0.7097 Ops/s $\color{#35bf28}+0.72\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-True] 1.9452s 1.8659s 0.5359 Ops/s 0.5323 Ops/s $\color{#35bf28}+0.67\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-False] 1.7220s 1.6432s 0.6086 Ops/s 0.6044 Ops/s $\color{#35bf28}+0.70\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-True] 4.6730s 4.5879s 0.2180 Ops/s 0.2166 Ops/s $\color{#35bf28}+0.64\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-False] 4.4721s 4.3976s 0.2274 Ops/s 0.2251 Ops/s $\color{#35bf28}+1.00\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] 1.9270s 1.8478s 0.5412 Ops/s 0.5340 Ops/s $\color{#35bf28}+1.35\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] 1.6702s 1.5673s 0.6380 Ops/s 0.6164 Ops/s $\color{#35bf28}+3.51\%$
test_values[generalized_advantage_estimate-True-True] 10.7229ms 10.5249ms 95.0124 Ops/s 96.2354 Ops/s $\color{#d91a1a}-1.27\%$
test_values[vec_generalized_advantage_estimate-True-True] 19.5083ms 17.6351ms 56.7050 Ops/s 78.2648 Ops/s $\textbf{\color{#d91a1a}-27.55\%}$
test_values[td0_return_estimate-False-False] 0.2308ms 0.1329ms 7.5234 KOps/s 7.8271 KOps/s $\color{#d91a1a}-3.88\%$
test_values[td1_return_estimate-False-False] 29.9733ms 28.5899ms 34.9774 Ops/s 35.1031 Ops/s $\color{#d91a1a}-0.36\%$
test_values[vec_td1_return_estimate-False-False] 18.2925ms 17.7238ms 56.4212 Ops/s 74.8881 Ops/s $\textbf{\color{#d91a1a}-24.66\%}$
test_values[td_lambda_return_estimate-True-False] 42.6604ms 42.2242ms 23.6831 Ops/s 23.7610 Ops/s $\color{#d91a1a}-0.33\%$
test_values[vec_td_lambda_return_estimate-True-False] 18.5412ms 17.7248ms 56.4180 Ops/s 64.4234 Ops/s $\textbf{\color{#d91a1a}-12.43\%}$
test_gae_speed[generalized_advantage_estimate-False-1-512] 9.3434ms 9.2873ms 107.6735 Ops/s 108.6851 Ops/s $\color{#d91a1a}-0.93\%$
test_gae_speed[vec_generalized_advantage_estimate-True-1-512] 1.7433ms 1.5231ms 656.5398 Ops/s 650.0677 Ops/s $\color{#35bf28}+1.00\%$
test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 0.5094ms 0.4198ms 2.3820 KOps/s 2.3520 KOps/s $\color{#35bf28}+1.28\%$
test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 35.2686ms 34.9812ms 28.5868 Ops/s 33.7075 Ops/s $\textbf{\color{#d91a1a}-15.19\%}$
test_gae_speed[vec_generalized_advantage_estimate-False-32-512] 1.9514ms 1.7240ms 580.0514 Ops/s 575.3822 Ops/s $\color{#35bf28}+0.81\%$
test_dqn_speed[False-None] 1.4871ms 1.3941ms 717.2830 Ops/s 716.9126 Ops/s $\color{#35bf28}+0.05\%$
test_dqn_speed[False-backward] 1.9467ms 1.9055ms 524.7834 Ops/s 524.5762 Ops/s $\color{#35bf28}+0.04\%$
test_dqn_speed[True-None] 0.6464ms 0.5513ms 1.8138 KOps/s 1.8292 KOps/s $\color{#d91a1a}-0.84\%$
test_dqn_speed[True-backward] 1.0600ms 1.0082ms 991.8372 Ops/s 987.5591 Ops/s $\color{#35bf28}+0.43\%$
test_dqn_speed[reduce-overhead-None] 0.9270ms 0.5418ms 1.8458 KOps/s 1.7838 KOps/s $\color{#35bf28}+3.48\%$
test_ddpg_speed[False-None] 3.2041ms 2.8480ms 351.1178 Ops/s 352.5018 Ops/s $\color{#d91a1a}-0.39\%$
test_ddpg_speed[False-backward] 4.2128ms 4.0660ms 245.9409 Ops/s 245.9924 Ops/s $\color{#d91a1a}-0.02\%$
test_ddpg_speed[True-None] 1.4806ms 1.4045ms 711.9819 Ops/s 703.6782 Ops/s $\color{#35bf28}+1.18\%$
test_ddpg_speed[True-backward] 2.4628ms 2.3916ms 418.1219 Ops/s 355.6053 Ops/s $\textbf{\color{#35bf28}+17.58\%}$
test_ddpg_speed[reduce-overhead-None] 1.5432ms 1.3952ms 716.7507 Ops/s 720.1714 Ops/s $\color{#d91a1a}-0.47\%$
test_sac_speed[False-None] 8.5198ms 7.9458ms 125.8519 Ops/s 126.6381 Ops/s $\color{#d91a1a}-0.62\%$
test_sac_speed[False-backward] 11.6746ms 11.2071ms 89.2295 Ops/s 89.8875 Ops/s $\color{#d91a1a}-0.73\%$
test_sac_speed[True-None] 2.3060ms 2.1376ms 467.8149 Ops/s 464.6589 Ops/s $\color{#35bf28}+0.68\%$
test_sac_speed[True-backward] 4.1849ms 3.9643ms 252.2499 Ops/s 245.1878 Ops/s $\color{#35bf28}+2.88\%$
test_sac_speed[reduce-overhead-None] 2.4458ms 2.1143ms 472.9626 Ops/s 456.6087 Ops/s $\color{#35bf28}+3.58\%$
test_redq_speed[False-None] 15.0428ms 10.4577ms 95.6229 Ops/s 95.3684 Ops/s $\color{#35bf28}+0.27\%$
test_redq_speed[False-backward] 18.3060ms 17.7250ms 56.4176 Ops/s 56.1217 Ops/s $\color{#35bf28}+0.53\%$
test_redq_speed[True-None] 4.5483ms 4.4056ms 226.9857 Ops/s 237.9239 Ops/s $\color{#d91a1a}-4.60\%$
test_redq_speed[True-backward] 9.9195ms 9.6345ms 103.7937 Ops/s 107.2757 Ops/s $\color{#d91a1a}-3.25\%$
test_redq_speed[reduce-overhead-None] 4.4808ms 4.3306ms 230.9174 Ops/s 243.7059 Ops/s $\textbf{\color{#d91a1a}-5.25\%}$
test_redq_deprec_speed[False-None] 11.6098ms 11.0787ms 90.2635 Ops/s 89.8760 Ops/s $\color{#35bf28}+0.43\%$
test_redq_deprec_speed[False-backward] 16.6046ms 16.1301ms 61.9959 Ops/s 60.8019 Ops/s $\color{#35bf28}+1.96\%$
test_redq_deprec_speed[True-None] 4.0568ms 3.6360ms 275.0251 Ops/s 274.0763 Ops/s $\color{#35bf28}+0.35\%$
test_redq_deprec_speed[True-backward] 7.9247ms 7.7248ms 129.4524 Ops/s 115.8814 Ops/s $\textbf{\color{#35bf28}+11.71\%}$
test_redq_deprec_speed[reduce-overhead-None] 4.0282ms 3.5914ms 278.4460 Ops/s 278.5059 Ops/s $\color{#d91a1a}-0.02\%$
test_td3_speed[False-None] 8.1196ms 7.9944ms 125.0875 Ops/s 124.8329 Ops/s $\color{#35bf28}+0.20\%$
test_td3_speed[False-backward] 11.3241ms 10.8788ms 91.9223 Ops/s 92.2925 Ops/s $\color{#d91a1a}-0.40\%$
test_td3_speed[True-None] 1.8566ms 1.8093ms 552.7065 Ops/s 551.3814 Ops/s $\color{#35bf28}+0.24\%$
test_td3_speed[True-backward] 3.9934ms 3.6222ms 276.0784 Ops/s 254.8513 Ops/s $\textbf{\color{#35bf28}+8.33\%}$
test_td3_speed[reduce-overhead-None] 1.8287ms 1.7848ms 560.2875 Ops/s 561.6521 Ops/s $\color{#d91a1a}-0.24\%$
test_cql_speed[False-None] 29.4902ms 26.0911ms 38.3272 Ops/s 38.7013 Ops/s $\color{#d91a1a}-0.97\%$
test_cql_speed[False-backward] 39.4139ms 35.9790ms 27.7940 Ops/s 28.4227 Ops/s $\color{#d91a1a}-2.21\%$
test_cql_speed[True-None] 12.7015ms 12.1362ms 82.3982 Ops/s 80.4669 Ops/s $\color{#35bf28}+2.40\%$
test_cql_speed[True-backward] 18.5491ms 17.8288ms 56.0891 Ops/s 56.9261 Ops/s $\color{#d91a1a}-1.47\%$
test_cql_speed[reduce-overhead-None] 15.1833ms 12.3517ms 80.9604 Ops/s 80.8646 Ops/s $\color{#35bf28}+0.12\%$
test_a2c_speed[False-None] 5.5082ms 5.3575ms 186.6549 Ops/s 183.8003 Ops/s $\color{#35bf28}+1.55\%$
test_a2c_speed[False-backward] 12.3257ms 11.8938ms 84.0773 Ops/s 83.8273 Ops/s $\color{#35bf28}+0.30\%$
test_a2c_speed[True-None] 4.0843ms 3.7084ms 269.6569 Ops/s 264.4769 Ops/s $\color{#35bf28}+1.96\%$
test_a2c_speed[True-backward] 8.8186ms 8.4721ms 118.0344 Ops/s 116.0167 Ops/s $\color{#35bf28}+1.74\%$
test_a2c_speed[reduce-overhead-None] 3.9794ms 3.7011ms 270.1902 Ops/s 270.7680 Ops/s $\color{#d91a1a}-0.21\%$
test_ppo_speed[False-None] 6.2667ms 5.9166ms 169.0147 Ops/s 169.1419 Ops/s $\color{#d91a1a}-0.08\%$
test_ppo_speed[False-backward] 12.6894ms 12.3532ms 80.9509 Ops/s 79.9254 Ops/s $\color{#35bf28}+1.28\%$
test_ppo_speed[True-None] 3.7661ms 3.6118ms 276.8725 Ops/s 274.1521 Ops/s $\color{#35bf28}+0.99\%$
test_ppo_speed[True-backward] 9.0317ms 8.3461ms 119.8161 Ops/s 117.4307 Ops/s $\color{#35bf28}+2.03\%$
test_ppo_speed[reduce-overhead-None] 4.0335ms 3.6041ms 277.4598 Ops/s 270.6630 Ops/s $\color{#35bf28}+2.51\%$
test_reinforce_speed[False-None] 5.0344ms 4.5904ms 217.8440 Ops/s 218.1519 Ops/s $\color{#d91a1a}-0.14\%$
test_reinforce_speed[False-backward] 7.6499ms 7.4038ms 135.0660 Ops/s 136.0527 Ops/s $\color{#d91a1a}-0.73\%$
test_reinforce_speed[True-None] 3.0582ms 2.8620ms 349.4083 Ops/s 330.1753 Ops/s $\textbf{\color{#35bf28}+5.83\%}$
test_reinforce_speed[True-backward] 7.9805ms 7.7048ms 129.7898 Ops/s 123.9434 Ops/s $\color{#35bf28}+4.72\%$
test_reinforce_speed[reduce-overhead-None] 3.3683ms 2.8564ms 350.0879 Ops/s 349.0029 Ops/s $\color{#35bf28}+0.31\%$
test_iql_speed[False-None] 25.4731ms 20.3115ms 49.2331 Ops/s 49.0965 Ops/s $\color{#35bf28}+0.28\%$
test_iql_speed[False-backward] 35.2636ms 30.7617ms 32.5079 Ops/s 32.4706 Ops/s $\color{#35bf28}+0.11\%$
test_iql_speed[True-None] 8.5855ms 8.3472ms 119.8003 Ops/s 117.4860 Ops/s $\color{#35bf28}+1.97\%$
test_iql_speed[True-backward] 17.0646ms 16.5746ms 60.3333 Ops/s 59.6273 Ops/s $\color{#35bf28}+1.18\%$
test_iql_speed[reduce-overhead-None] 8.6893ms 8.4309ms 118.6113 Ops/s 116.8362 Ops/s $\color{#35bf28}+1.52\%$
test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 6.1739ms 6.0545ms 165.1674 Ops/s 165.4229 Ops/s $\color{#d91a1a}-0.15\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 2.7387ms 0.3442ms 2.9052 KOps/s 3.4956 KOps/s $\textbf{\color{#d91a1a}-16.89\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.6075ms 0.3151ms 3.1739 KOps/s 3.7525 KOps/s $\textbf{\color{#d91a1a}-15.42\%}$
test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.0149ms 5.7621ms 173.5488 Ops/s 173.4713 Ops/s $\color{#35bf28}+0.04\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 0.8726ms 0.3714ms 2.6924 KOps/s 3.5745 KOps/s $\textbf{\color{#d91a1a}-24.68\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.6057ms 0.3599ms 2.7785 KOps/s 3.7979 KOps/s $\textbf{\color{#d91a1a}-26.84\%}$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] 1.7399ms 1.4415ms 693.7309 Ops/s 796.7032 Ops/s $\textbf{\color{#d91a1a}-12.92\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 1.6326ms 1.3655ms 732.3529 Ops/s 844.7166 Ops/s $\textbf{\color{#d91a1a}-13.30\%}$
test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 9.9630ms 6.0246ms 165.9848 Ops/s 169.5761 Ops/s $\color{#d91a1a}-2.12\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 1.6367ms 0.4504ms 2.2205 KOps/s 2.0186 KOps/s $\textbf{\color{#35bf28}+10.00\%}$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.8998ms 0.4517ms 2.2136 KOps/s 2.0330 KOps/s $\textbf{\color{#35bf28}+8.88\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 5.8469ms 5.7423ms 174.1455 Ops/s 171.2942 Ops/s $\color{#35bf28}+1.66\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 2.0195ms 0.3158ms 3.1663 KOps/s 3.4760 KOps/s $\textbf{\color{#d91a1a}-8.91\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.4570ms 0.2670ms 3.7453 KOps/s 3.1936 KOps/s $\textbf{\color{#35bf28}+17.27\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 5.9886ms 5.7536ms 173.8034 Ops/s 173.0401 Ops/s $\color{#35bf28}+0.44\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 1.8195ms 0.2832ms 3.5316 KOps/s 2.8860 KOps/s $\textbf{\color{#35bf28}+22.37\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.4690ms 0.2623ms 3.8118 KOps/s 2.8850 KOps/s $\textbf{\color{#35bf28}+32.12\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 6.1497ms 5.9229ms 168.8364 Ops/s 168.9743 Ops/s $\color{#d91a1a}-0.08\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 1.1619ms 0.4745ms 2.1074 KOps/s 2.0773 KOps/s $\color{#35bf28}+1.45\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6931ms 0.4555ms 2.1955 KOps/s 1.9783 KOps/s $\textbf{\color{#35bf28}+10.98\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 6.3848ms 4.9537ms 201.8694 Ops/s 199.6974 Ops/s $\color{#35bf28}+1.09\%$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 10.3110ms 2.1168ms 472.4045 Ops/s 489.9363 Ops/s $\color{#d91a1a}-3.58\%$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] 0.9939ms 0.8701ms 1.1493 KOps/s 1.0723 KOps/s $\textbf{\color{#35bf28}+7.18\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 0.5303s 15.5511ms 64.3039 Ops/s 58.9567 Ops/s $\textbf{\color{#35bf28}+9.07\%}$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 12.6084ms 1.9513ms 512.4693 Ops/s 511.6013 Ops/s $\color{#35bf28}+0.17\%$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 7.1174ms 1.1775ms 849.2515 Ops/s 843.8040 Ops/s $\color{#35bf28}+0.65\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 6.5893ms 5.1133ms 195.5669 Ops/s 193.3979 Ops/s $\color{#35bf28}+1.12\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 6.0784ms 1.9324ms 517.4976 Ops/s 517.6228 Ops/s $\color{#d91a1a}-0.02\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 1.3830ms 1.0455ms 956.4756 Ops/s 960.4800 Ops/s $\color{#d91a1a}-0.42\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] 37.7015ms 35.9430ms 27.8219 Ops/s 27.7796 Ops/s $\color{#35bf28}+0.15\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] 19.5400ms 17.9946ms 55.5724 Ops/s 55.1316 Ops/s $\color{#35bf28}+0.80\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] 39.4308ms 37.1551ms 26.9142 Ops/s 26.9956 Ops/s $\color{#d91a1a}-0.30\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] 19.8651ms 18.4100ms 54.3184 Ops/s 54.6840 Ops/s $\color{#d91a1a}-0.67\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] 40.6812ms 38.7405ms 25.8128 Ops/s 25.8112 Ops/s $+0.01\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] 21.4355ms 19.9220ms 50.1957 Ops/s 50.7159 Ops/s $\color{#d91a1a}-1.03\%$
test_storage_write_lazystack[50-img_shape0-small] 0.8202ms 0.2173ms 4.6014 KOps/s 4.4014 KOps/s $\color{#35bf28}+4.54\%$
test_storage_write_lazystack[100-img_shape1-atari] 1.6632ms 1.3821ms 723.5252 Ops/s 709.6916 Ops/s $\color{#35bf28}+1.95\%$
test_storage_write_lazystack[100-img_shape2-large_img] 2.6954ms 2.3118ms 432.5649 Ops/s 430.2753 Ops/s $\color{#35bf28}+0.53\%$
test_storage_write_lazystack[200-img_shape3-large_batch] 3.0764ms 2.9265ms 341.7023 Ops/s 340.7706 Ops/s $\color{#35bf28}+0.27\%$
test_storage_write_contiguous[50-img_shape0-small] 0.5347ms 0.1326ms 7.5434 KOps/s 7.5768 KOps/s $\color{#d91a1a}-0.44\%$
test_storage_write_contiguous[100-img_shape1-atari] 0.3532ms 0.1864ms 5.3648 KOps/s 5.0553 KOps/s $\textbf{\color{#35bf28}+6.12\%}$
test_storage_write_contiguous[100-img_shape2-large_img] 1.9319ms 1.7809ms 561.5289 Ops/s 564.7873 Ops/s $\color{#d91a1a}-0.58\%$
test_storage_write_contiguous[200-img_shape3-large_batch] 1.6945ms 1.3396ms 746.4942 Ops/s 766.0448 Ops/s $\color{#d91a1a}-2.55\%$
test_collector_stack_then_write[50-img_shape0-small] 1.2454ms 1.1187ms 893.9209 Ops/s 892.7533 Ops/s $\color{#35bf28}+0.13\%$
test_collector_stack_then_write[100-img_shape1-atari] 7.5002ms 3.6329ms 275.2619 Ops/s 281.9876 Ops/s $\color{#d91a1a}-2.39\%$
test_collector_stack_then_write[100-img_shape2-large_img] 11.5480ms 5.7321ms 174.4562 Ops/s 174.0969 Ops/s $\color{#35bf28}+0.21\%$
test_collector_stack_then_write[200-img_shape3-large_batch] 15.0441ms 7.1312ms 140.2296 Ops/s 143.5403 Ops/s $\color{#d91a1a}-2.31\%$
test_collector_lazystack_then_write[50-img_shape0-small] 0.4417ms 0.2737ms 3.6536 KOps/s 3.5840 KOps/s $\color{#35bf28}+1.94\%$
test_collector_lazystack_then_write[100-img_shape1-atari] 1.6689ms 1.5184ms 658.5666 Ops/s 647.9313 Ops/s $\color{#35bf28}+1.64\%$
test_collector_lazystack_then_write[100-img_shape2-large_img] 2.6432ms 2.4195ms 413.3059 Ops/s 405.9709 Ops/s $\color{#35bf28}+1.81\%$
test_collector_lazystack_then_write[200-img_shape3-large_batch] 3.3469ms 3.1272ms 319.7777 Ops/s 316.9655 Ops/s $\color{#35bf28}+0.89\%$
test_collector_without_rb[100-img_shape0-atari] 33.8054ms 32.7763ms 30.5099 Ops/s 30.7263 Ops/s $\color{#d91a1a}-0.70\%$
test_collector_without_rb[200-img_shape1-large_batch] 65.9918ms 64.4532ms 15.5151 Ops/s 15.5397 Ops/s $\color{#d91a1a}-0.16\%$
test_collector_with_rb[100-img_shape0-atari] 38.6054ms 37.3138ms 26.7997 Ops/s 27.0101 Ops/s $\color{#d91a1a}-0.78\%$
test_collector_with_rb[200-img_shape1-large_batch] 96.1414ms 74.3630ms 13.4475 Ops/s 13.6749 Ops/s $\color{#d91a1a}-1.66\%$

@github-actions
Copy link
Copy Markdown
Contributor

github-actions bot commented Feb 16, 2026 •
edited
Loading

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

Total Benchmarks: 172. Improved: $\large\color{#35bf28}9$. Worsened: $\large\color{#d91a1a}11$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_tensor_to_bytestream_speed[pickle] 86.4853μs 84.3182μs 11.8598 KOps/s 12.6680 KOps/s $\textbf{\color{#d91a1a}-6.38\%}$
test_tensor_to_bytestream_speed[torch.save] 0.1442ms 0.1432ms 6.9857 KOps/s 7.2688 KOps/s $\color{#d91a1a}-3.90\%$
test_tensor_to_bytestream_speed[untyped_storage] 0.1048s 0.1039s 9.6235 Ops/s 9.5688 Ops/s $\color{#35bf28}+0.57\%$
test_tensor_to_bytestream_speed[numpy] 2.5105μs 2.5043μs 399.3063 KOps/s 400.1154 KOps/s $\color{#d91a1a}-0.20\%$
test_tensor_to_bytestream_speed[safetensors] 37.0755μs 36.8040μs 27.1710 KOps/s 26.1142 KOps/s $\color{#35bf28}+4.05\%$
test_simple 0.7758s 0.7745s 1.2912 Ops/s 1.2565 Ops/s $\color{#35bf28}+2.76\%$
test_transformed 1.3604s 1.3575s 0.7366 Ops/s 0.7246 Ops/s $\color{#35bf28}+1.66\%$
test_serial 2.2470s 2.2449s 0.4454 Ops/s 0.4418 Ops/s $\color{#35bf28}+0.82\%$
test_parallel 1.8852s 1.8047s 0.5541 Ops/s 0.5528 Ops/s $\color{#35bf28}+0.23\%$
test_step_mdp_speed[True-True-True-True-True] 0.2605ms 40.4083μs 24.7474 KOps/s 25.0295 KOps/s $\color{#d91a1a}-1.13\%$
test_step_mdp_speed[True-True-True-True-False] 50.6910μs 22.5102μs 44.4243 KOps/s 43.6220 KOps/s $\color{#35bf28}+1.84\%$
test_step_mdp_speed[True-True-True-False-True] 89.8620μs 22.6556μs 44.1391 KOps/s 43.8794 KOps/s $\color{#35bf28}+0.59\%$
test_step_mdp_speed[True-True-True-False-False] 77.3820μs 12.3371μs 81.0562 KOps/s 80.7492 KOps/s $\color{#35bf28}+0.38\%$
test_step_mdp_speed[True-True-False-True-True] 79.7520μs 43.9212μs 22.7680 KOps/s 23.2910 KOps/s $\color{#d91a1a}-2.25\%$
test_step_mdp_speed[True-True-False-True-False] 56.2110μs 25.2505μs 39.6032 KOps/s 40.7054 KOps/s $\color{#d91a1a}-2.71\%$
test_step_mdp_speed[True-True-False-False-True] 57.4310μs 25.3983μs 39.3728 KOps/s 39.6041 KOps/s $\color{#d91a1a}-0.58\%$
test_step_mdp_speed[True-True-False-False-False] 52.7910μs 15.2476μs 65.5841 KOps/s 67.6866 KOps/s $\color{#d91a1a}-3.11\%$
test_step_mdp_speed[True-False-True-True-True] 87.1120μs 46.2981μs 21.5992 KOps/s 21.7425 KOps/s $\color{#d91a1a}-0.66\%$
test_step_mdp_speed[True-False-True-True-False] 90.6320μs 27.0205μs 37.0090 KOps/s 36.3296 KOps/s $\color{#35bf28}+1.87\%$
test_step_mdp_speed[True-False-True-False-True] 59.1710μs 25.7621μs 38.8167 KOps/s 39.6337 KOps/s $\color{#d91a1a}-2.06\%$
test_step_mdp_speed[True-False-True-False-False] 43.2610μs 15.0169μs 66.5916 KOps/s 66.8580 KOps/s $\color{#d91a1a}-0.40\%$
test_step_mdp_speed[True-False-False-True-True] 85.2020μs 48.3388μs 20.6873 KOps/s 20.9307 KOps/s $\color{#d91a1a}-1.16\%$
test_step_mdp_speed[True-False-False-True-False] 64.4320μs 30.1730μs 33.1423 KOps/s 33.5222 KOps/s $\color{#d91a1a}-1.13\%$
test_step_mdp_speed[True-False-False-False-True] 58.7610μs 28.0874μs 35.6031 KOps/s 35.8817 KOps/s $\color{#d91a1a}-0.78\%$
test_step_mdp_speed[True-False-False-False-False] 46.6110μs 17.4723μs 57.2333 KOps/s 57.7316 KOps/s $\color{#d91a1a}-0.86\%$
test_step_mdp_speed[False-True-True-True-True] 85.2320μs 45.1286μs 22.1589 KOps/s 21.9837 KOps/s $\color{#35bf28}+0.80\%$
test_step_mdp_speed[False-True-True-True-False] 86.0320μs 27.1314μs 36.8577 KOps/s 35.8211 KOps/s $\color{#35bf28}+2.89\%$
test_step_mdp_speed[False-True-True-False-True] 2.6334ms 29.5213μs 33.8738 KOps/s 34.2017 KOps/s $\color{#d91a1a}-0.96\%$
test_step_mdp_speed[False-True-True-False-False] 46.5310μs 16.8213μs 59.4484 KOps/s 59.8743 KOps/s $\color{#d91a1a}-0.71\%$
test_step_mdp_speed[False-True-False-True-True] 88.4420μs 48.4539μs 20.6382 KOps/s 20.8224 KOps/s $\color{#d91a1a}-0.89\%$
test_step_mdp_speed[False-True-False-True-False] 68.5610μs 29.9594μs 33.3785 KOps/s 33.1982 KOps/s $\color{#35bf28}+0.54\%$
test_step_mdp_speed[False-True-False-False-True] 72.3220μs 30.9669μs 32.2925 KOps/s 32.1602 KOps/s $\color{#35bf28}+0.41\%$
test_step_mdp_speed[False-True-False-False-False] 57.2520μs 19.0269μs 52.5572 KOps/s 52.6404 KOps/s $\color{#d91a1a}-0.16\%$
test_step_mdp_speed[False-False-True-True-True] 0.1090ms 51.6961μs 19.3438 KOps/s 19.9986 KOps/s $\color{#d91a1a}-3.27\%$
test_step_mdp_speed[False-False-True-True-False] 74.3520μs 33.5504μs 29.8059 KOps/s 30.7357 KOps/s $\color{#d91a1a}-3.03\%$
test_step_mdp_speed[False-False-True-False-True] 72.2220μs 31.9441μs 31.3047 KOps/s 32.2747 KOps/s $\color{#d91a1a}-3.01\%$
test_step_mdp_speed[False-False-True-False-False] 51.6510μs 19.3975μs 51.5531 KOps/s 52.9922 KOps/s $\color{#d91a1a}-2.72\%$
test_step_mdp_speed[False-False-False-True-True] 94.1420μs 52.9639μs 18.8808 KOps/s 18.9715 KOps/s $\color{#d91a1a}-0.48\%$
test_step_mdp_speed[False-False-False-True-False] 70.1520μs 35.0606μs 28.5220 KOps/s 28.9376 KOps/s $\color{#d91a1a}-1.44\%$
test_step_mdp_speed[False-False-False-False-True] 69.4510μs 34.1145μs 29.3131 KOps/s 30.3000 KOps/s $\color{#d91a1a}-3.26\%$
test_step_mdp_speed[False-False-False-False-False] 50.3010μs 21.2565μs 47.0444 KOps/s 47.3314 KOps/s $\color{#d91a1a}-0.61\%$
test_non_tensor_env_rollout_speed[1000-single-True] 0.8282s 0.7219s 1.3853 Ops/s 1.3852 Ops/s $+0.01\%$
test_non_tensor_env_rollout_speed[1000-single-False] 0.6899s 0.5926s 1.6875 Ops/s 1.6830 Ops/s $\color{#35bf28}+0.27\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 1.6700s 1.5915s 0.6283 Ops/s 0.6298 Ops/s $\color{#d91a1a}-0.23\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] 1.4596s 1.3698s 0.7301 Ops/s 0.7296 Ops/s $\color{#35bf28}+0.06\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-True] 1.9109s 1.8269s 0.5474 Ops/s 0.5472 Ops/s $\color{#35bf28}+0.04\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-False] 1.6820s 1.6011s 0.6246 Ops/s 0.6187 Ops/s $\color{#35bf28}+0.95\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-True] 4.6869s 4.5693s 0.2189 Ops/s 0.2186 Ops/s $\color{#35bf28}+0.12\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-False] 4.5208s 4.3508s 0.2298 Ops/s 0.2274 Ops/s $\color{#35bf28}+1.09\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] 2.0947s 1.8791s 0.5322 Ops/s 0.5463 Ops/s $\color{#d91a1a}-2.59\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] 1.6257s 1.5381s 0.6502 Ops/s 0.6460 Ops/s $\color{#35bf28}+0.64\%$
test_values[generalized_advantage_estimate-True-True] 20.5906ms 20.2726ms 49.3278 Ops/s 49.0950 Ops/s $\color{#35bf28}+0.47\%$
test_values[vec_generalized_advantage_estimate-True-True] 0.1336s 3.5897ms 278.5765 Ops/s 270.1173 Ops/s $\color{#35bf28}+3.13\%$
test_values[td0_return_estimate-False-False] 0.1055ms 82.6874μs 12.0937 KOps/s 12.0063 KOps/s $\color{#35bf28}+0.73\%$
test_values[td1_return_estimate-False-False] 48.1128ms 47.5291ms 21.0398 Ops/s 20.9031 Ops/s $\color{#35bf28}+0.65\%$
test_values[vec_td1_return_estimate-False-False] 1.3020ms 1.0833ms 923.0709 Ops/s 921.2984 Ops/s $\color{#35bf28}+0.19\%$
test_values[td_lambda_return_estimate-True-False] 78.4021ms 77.8382ms 12.8472 Ops/s 12.7829 Ops/s $\color{#35bf28}+0.50\%$
test_values[vec_td_lambda_return_estimate-True-False] 1.3276ms 1.0802ms 925.7386 Ops/s 923.4219 Ops/s $\color{#35bf28}+0.25\%$
test_gae_speed[generalized_advantage_estimate-False-1-512] 20.5993ms 20.3312ms 49.1855 Ops/s 49.2807 Ops/s $\color{#d91a1a}-0.19\%$
test_gae_speed[vec_generalized_advantage_estimate-True-1-512] 1.0308ms 0.7684ms 1.3014 KOps/s 1.3194 KOps/s $\color{#d91a1a}-1.37\%$
test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 0.7292ms 0.6748ms 1.4819 KOps/s 1.4875 KOps/s $\color{#d91a1a}-0.37\%$
test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 1.5523ms 1.4875ms 672.2850 Ops/s 671.9196 Ops/s $\color{#35bf28}+0.05\%$
test_gae_speed[vec_generalized_advantage_estimate-False-32-512] 0.7422ms 0.6907ms 1.4479 KOps/s 1.4456 KOps/s $\color{#35bf28}+0.16\%$
test_dqn_speed[False-None] 1.5945ms 1.5119ms 661.3978 Ops/s 666.6180 Ops/s $\color{#d91a1a}-0.78\%$
test_dqn_speed[False-backward] 2.2062ms 2.1406ms 467.1655 Ops/s 460.3535 Ops/s $\color{#35bf28}+1.48\%$
test_dqn_speed[True-None] 0.6443ms 0.5715ms 1.7496 KOps/s 1.8117 KOps/s $\color{#d91a1a}-3.42\%$
test_dqn_speed[True-backward] 1.2365ms 1.1871ms 842.3948 Ops/s 925.6328 Ops/s $\textbf{\color{#d91a1a}-8.99\%}$
test_dqn_speed[reduce-overhead-None] 0.6806ms 0.5655ms 1.7683 KOps/s 1.7171 KOps/s $\color{#35bf28}+2.98\%$
test_ddpg_speed[False-None] 3.1944ms 2.8315ms 353.1686 Ops/s 347.1468 Ops/s $\color{#35bf28}+1.73\%$
test_ddpg_speed[False-backward] 4.6008ms 4.2123ms 237.4011 Ops/s 243.1628 Ops/s $\color{#d91a1a}-2.37\%$
test_ddpg_speed[True-None] 1.3675ms 1.2750ms 784.3157 Ops/s 778.2771 Ops/s $\color{#35bf28}+0.78\%$
test_ddpg_speed[True-backward] 2.5399ms 2.4659ms 405.5391 Ops/s 426.9548 Ops/s $\textbf{\color{#d91a1a}-5.02\%}$
test_ddpg_speed[reduce-overhead-None] 1.3825ms 1.2997ms 769.3867 Ops/s 759.4937 Ops/s $\color{#35bf28}+1.30\%$
test_sac_speed[False-None] 8.7860ms 8.2126ms 121.7635 Ops/s 121.1040 Ops/s $\color{#35bf28}+0.54\%$
test_sac_speed[False-backward] 11.9442ms 11.5003ms 86.9544 Ops/s 87.9898 Ops/s $\color{#d91a1a}-1.18\%$
test_sac_speed[True-None] 1.8606ms 1.7603ms 568.0996 Ops/s 568.3600 Ops/s $\color{#d91a1a}-0.05\%$
test_sac_speed[True-backward] 3.5677ms 3.4925ms 286.3285 Ops/s 285.5088 Ops/s $\color{#35bf28}+0.29\%$
test_sac_speed[reduce-overhead-None] 19.2667ms 10.9511ms 91.3146 Ops/s 82.7557 Ops/s $\textbf{\color{#35bf28}+10.34\%}$
test_redq_deprec_speed[False-None] 10.0982ms 9.2071ms 108.6118 Ops/s 108.9016 Ops/s $\color{#d91a1a}-0.27\%$
test_redq_deprec_speed[False-backward] 13.2265ms 12.6468ms 79.0716 Ops/s 79.5146 Ops/s $\color{#d91a1a}-0.56\%$
test_redq_deprec_speed[True-None] 2.7729ms 2.4845ms 402.4933 Ops/s 402.7102 Ops/s $\color{#d91a1a}-0.05\%$
test_redq_deprec_speed[True-backward] 4.4177ms 4.2215ms 236.8810 Ops/s 236.9568 Ops/s $\color{#d91a1a}-0.03\%$
test_redq_deprec_speed[reduce-overhead-None] 15.9015ms 9.7175ms 102.9074 Ops/s 103.0663 Ops/s $\color{#d91a1a}-0.15\%$
test_td3_speed[False-None] 8.4445ms 8.1029ms 123.4124 Ops/s 124.1218 Ops/s $\color{#d91a1a}-0.57\%$
test_td3_speed[False-backward] 11.4524ms 10.7168ms 93.3117 Ops/s 93.5678 Ops/s $\color{#d91a1a}-0.27\%$
test_td3_speed[True-None] 1.5958ms 1.5736ms 635.4808 Ops/s 636.0620 Ops/s $\color{#d91a1a}-0.09\%$
test_td3_speed[True-backward] 3.2221ms 3.1542ms 317.0339 Ops/s 313.1120 Ops/s $\color{#35bf28}+1.25\%$
test_td3_speed[reduce-overhead-None] 46.1053ms 23.6915ms 42.2092 Ops/s 41.3820 Ops/s $\color{#35bf28}+2.00\%$
test_cql_speed[False-None] 17.2756ms 16.9834ms 58.8810 Ops/s 58.6626 Ops/s $\color{#35bf28}+0.37\%$
test_cql_speed[False-backward] 23.0615ms 22.6084ms 44.2314 Ops/s 44.1509 Ops/s $\color{#35bf28}+0.18\%$
test_cql_speed[True-None] 3.2605ms 3.1523ms 317.2287 Ops/s 313.9966 Ops/s $\color{#35bf28}+1.03\%$
test_cql_speed[True-backward] 5.3890ms 5.2006ms 192.2858 Ops/s 185.0527 Ops/s $\color{#35bf28}+3.91\%$
test_cql_speed[reduce-overhead-None] 19.0078ms 11.8886ms 84.1143 Ops/s 84.4314 Ops/s $\color{#d91a1a}-0.38\%$
test_a2c_speed[False-None] 4.0201ms 3.1928ms 313.2004 Ops/s 312.7209 Ops/s $\color{#35bf28}+0.15\%$
test_a2c_speed[False-backward] 6.5045ms 6.3632ms 157.1528 Ops/s 156.6608 Ops/s $\color{#35bf28}+0.31\%$
test_a2c_speed[True-None] 1.7651ms 1.3026ms 767.7146 Ops/s 773.0855 Ops/s $\color{#d91a1a}-0.69\%$
test_a2c_speed[True-backward] 3.1078ms 3.0351ms 329.4755 Ops/s 328.3458 Ops/s $\color{#35bf28}+0.34\%$
test_a2c_speed[reduce-overhead-None] 1.0320ms 0.9536ms 1.0487 KOps/s 1.0478 KOps/s $\color{#35bf28}+0.09\%$
test_ppo_speed[False-None] 4.0465ms 3.8102ms 262.4561 Ops/s 263.3869 Ops/s $\color{#d91a1a}-0.35\%$
test_ppo_speed[False-backward] 7.5949ms 7.1509ms 139.8434 Ops/s 138.5100 Ops/s $\color{#35bf28}+0.96\%$
test_ppo_speed[True-None] 1.5327ms 1.3883ms 720.2924 Ops/s 730.9135 Ops/s $\color{#d91a1a}-1.45\%$
test_ppo_speed[True-backward] 3.3728ms 3.1948ms 313.0083 Ops/s 311.1927 Ops/s $\color{#35bf28}+0.58\%$
test_ppo_speed[reduce-overhead-None] 1.0703ms 1.0179ms 982.4040 Ops/s 957.1338 Ops/s $\color{#35bf28}+2.64\%$
test_reinforce_speed[False-None] 2.4220ms 2.2544ms 443.5841 Ops/s 440.4614 Ops/s $\color{#35bf28}+0.71\%$
test_reinforce_speed[False-backward] 3.6757ms 3.3906ms 294.9359 Ops/s 292.4441 Ops/s $\color{#35bf28}+0.85\%$
test_reinforce_speed[True-None] 1.3039ms 1.2269ms 815.0544 Ops/s 811.9252 Ops/s $\color{#35bf28}+0.39\%$
test_reinforce_speed[True-backward] 3.1137ms 3.0068ms 332.5760 Ops/s 332.1172 Ops/s $\color{#35bf28}+0.14\%$
test_reinforce_speed[reduce-overhead-None] 17.2783ms 9.5487ms 104.7258 Ops/s 108.7011 Ops/s $\color{#d91a1a}-3.66\%$
test_iql_speed[False-None] 10.1142ms 9.3334ms 107.1426 Ops/s 107.0234 Ops/s $\color{#35bf28}+0.11\%$
test_iql_speed[False-backward] 14.1801ms 13.3737ms 74.7734 Ops/s 74.7342 Ops/s $\color{#35bf28}+0.05\%$
test_iql_speed[True-None] 2.3143ms 2.1144ms 472.9488 Ops/s 471.3198 Ops/s $\color{#35bf28}+0.35\%$
test_iql_speed[True-backward] 4.9999ms 4.7769ms 209.3415 Ops/s 212.6807 Ops/s $\color{#d91a1a}-1.57\%$
test_iql_speed[reduce-overhead-None] 17.7117ms 10.4191ms 95.9776 Ops/s 95.9974 Ops/s $\color{#d91a1a}-0.02\%$
test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 5.8903ms 5.7351ms 174.3652 Ops/s 174.0445 Ops/s $\color{#35bf28}+0.18\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 1.6738ms 0.2886ms 3.4646 KOps/s 2.8889 KOps/s $\textbf{\color{#35bf28}+19.93\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.5136ms 0.2634ms 3.7961 KOps/s 3.0151 KOps/s $\textbf{\color{#35bf28}+25.91\%}$
test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 5.8018ms 5.5716ms 179.4828 Ops/s 178.3086 Ops/s $\color{#35bf28}+0.66\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 0.5729ms 0.2762ms 3.6203 KOps/s 3.1468 KOps/s $\textbf{\color{#35bf28}+15.05\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.5411ms 0.2604ms 3.8401 KOps/s 3.2477 KOps/s $\textbf{\color{#35bf28}+18.24\%}$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] 1.6210ms 1.2400ms 806.4714 Ops/s 800.3424 Ops/s $\color{#35bf28}+0.77\%$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 1.3820ms 1.1569ms 864.4040 Ops/s 849.2294 Ops/s $\color{#35bf28}+1.79\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 5.9814ms 5.8521ms 170.8782 Ops/s 173.1097 Ops/s $\color{#d91a1a}-1.29\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 2.2505ms 0.4626ms 2.1616 KOps/s 1.9998 KOps/s $\textbf{\color{#35bf28}+8.09\%}$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6766ms 0.4758ms 2.1015 KOps/s 2.0742 KOps/s $\color{#35bf28}+1.32\%$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 5.8068ms 5.6856ms 175.8815 Ops/s 177.3140 Ops/s $\color{#d91a1a}-0.81\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 0.8130ms 0.3235ms 3.0912 KOps/s 3.5716 KOps/s $\textbf{\color{#d91a1a}-13.45\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.4646ms 0.2652ms 3.7703 KOps/s 3.8370 KOps/s $\color{#d91a1a}-1.74\%$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 5.9055ms 5.6531ms 176.8935 Ops/s 177.3956 Ops/s $\color{#d91a1a}-0.28\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 1.8865ms 0.3475ms 2.8774 KOps/s 3.5817 KOps/s $\textbf{\color{#d91a1a}-19.66\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.6841ms 0.3515ms 2.8453 KOps/s 3.8321 KOps/s $\textbf{\color{#d91a1a}-25.75\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 5.8912ms 5.7928ms 172.6284 Ops/s 173.7971 Ops/s $\color{#d91a1a}-0.67\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 1.8814ms 0.4946ms 2.0218 KOps/s 2.3231 KOps/s $\textbf{\color{#d91a1a}-12.97\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.7359ms 0.4773ms 2.0953 KOps/s 2.4201 KOps/s $\textbf{\color{#d91a1a}-13.42\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 0.6055s 17.0219ms 58.7478 Ops/s 52.4601 Ops/s $\textbf{\color{#35bf28}+11.99\%}$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 4.4752ms 1.9828ms 504.3375 Ops/s 540.9307 Ops/s $\textbf{\color{#d91a1a}-6.76\%}$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] 10.7319ms 1.3855ms 721.7605 Ops/s 788.5837 Ops/s $\textbf{\color{#d91a1a}-8.47\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 8.0216ms 4.9784ms 200.8664 Ops/s 201.4418 Ops/s $\color{#d91a1a}-0.29\%$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 9.2846ms 1.8924ms 528.4217 Ops/s 559.8437 Ops/s $\textbf{\color{#d91a1a}-5.61\%}$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 1.0814ms 0.9202ms 1.0867 KOps/s 785.1668 Ops/s $\textbf{\color{#35bf28}+38.40\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 7.9664ms 5.1203ms 195.3002 Ops/s 193.0806 Ops/s $\color{#35bf28}+1.15\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 4.1841ms 1.9914ms 502.1647 Ops/s 480.3435 Ops/s $\color{#35bf28}+4.54\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 3.7624ms 1.1510ms 868.8248 Ops/s 853.6703 Ops/s $\color{#35bf28}+1.78\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] 38.8668ms 35.6752ms 28.0307 Ops/s 27.6821 Ops/s $\color{#35bf28}+1.26\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] 19.1765ms 17.8056ms 56.1620 Ops/s 55.0616 Ops/s $\color{#35bf28}+2.00\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] 39.6366ms 36.6107ms 27.3144 Ops/s 26.7079 Ops/s $\color{#35bf28}+2.27\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] 19.5884ms 18.0861ms 55.2911 Ops/s 53.8886 Ops/s $\color{#35bf28}+2.60\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] 39.9117ms 38.3637ms 26.0663 Ops/s 25.8648 Ops/s $\color{#35bf28}+0.78\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] 20.8645ms 19.5813ms 51.0691 Ops/s 50.3507 Ops/s $\color{#35bf28}+1.43\%$
test_storage_write_lazystack[50-img_shape0-small] 0.8777ms 0.2155ms 4.6401 KOps/s 4.4581 KOps/s $\color{#35bf28}+4.08\%$
test_storage_write_lazystack[100-img_shape1-atari] 1.6176ms 1.4346ms 697.0601 Ops/s 717.5726 Ops/s $\color{#d91a1a}-2.86\%$
test_storage_write_lazystack[100-img_shape2-large_img] 2.7865ms 2.3812ms 419.9602 Ops/s 427.2028 Ops/s $\color{#d91a1a}-1.70\%$
test_storage_write_lazystack[200-img_shape3-large_batch] 3.2630ms 2.9834ms 335.1881 Ops/s 337.5695 Ops/s $\color{#d91a1a}-0.71\%$
test_storage_write_contiguous[50-img_shape0-small] 0.2471ms 0.1587ms 6.3021 KOps/s 6.2245 KOps/s $\color{#35bf28}+1.25\%$
test_storage_write_contiguous[100-img_shape1-atari] 0.5740ms 0.2201ms 4.5434 KOps/s 3.7067 KOps/s $\textbf{\color{#35bf28}+22.57\%}$
test_storage_write_contiguous[100-img_shape2-large_img] 1.9258ms 1.7650ms 566.5664 Ops/s 573.1812 Ops/s $\color{#d91a1a}-1.15\%$
test_storage_write_contiguous[200-img_shape3-large_batch] 1.5525ms 1.3632ms 733.5902 Ops/s 717.2477 Ops/s $\color{#35bf28}+2.28\%$
test_collector_stack_then_write[50-img_shape0-small] 1.2693ms 1.1256ms 888.3942 Ops/s 889.7392 Ops/s $\color{#d91a1a}-0.15\%$
test_collector_stack_then_write[100-img_shape1-atari] 4.6971ms 3.5793ms 279.3866 Ops/s 276.8953 Ops/s $\color{#35bf28}+0.90\%$
test_collector_stack_then_write[100-img_shape2-large_img] 6.2440ms 5.8144ms 171.9869 Ops/s 171.2313 Ops/s $\color{#35bf28}+0.44\%$
test_collector_stack_then_write[200-img_shape3-large_batch] 7.4388ms 6.9666ms 143.5419 Ops/s 143.8070 Ops/s $\color{#d91a1a}-0.18\%$
test_collector_lazystack_then_write[50-img_shape0-small] 0.4345ms 0.2709ms 3.6914 KOps/s 3.6851 KOps/s $\color{#35bf28}+0.17\%$
test_collector_lazystack_then_write[100-img_shape1-atari] 1.6899ms 1.5302ms 653.4943 Ops/s 645.3392 Ops/s $\color{#35bf28}+1.26\%$
test_collector_lazystack_then_write[100-img_shape2-large_img] 2.6490ms 2.4933ms 401.0807 Ops/s 415.2368 Ops/s $\color{#d91a1a}-3.41\%$
test_collector_lazystack_then_write[200-img_shape3-large_batch] 3.5910ms 3.1675ms 315.7018 Ops/s 316.8657 Ops/s $\color{#d91a1a}-0.37\%$
test_collector_without_rb[100-img_shape0-atari] 32.6684ms 32.2726ms 30.9860 Ops/s 31.3243 Ops/s $\color{#d91a1a}-1.08\%$
test_collector_without_rb[200-img_shape1-large_batch] 63.7183ms 63.2599ms 15.8078 Ops/s 15.8606 Ops/s $\color{#d91a1a}-0.33\%$
test_collector_with_rb[100-img_shape0-atari] 37.4057ms 36.7150ms 27.2369 Ops/s 27.3827 Ops/s $\color{#d91a1a}-0.53\%$
test_collector_with_rb[200-img_shape1-large_batch] 71.7634ms 71.1853ms 14.0478 Ops/s 14.0278 Ops/s $\color{#35bf28}+0.14\%$
test_collector_without_rb_cuda[100-img_shape0-atari] 54.6722ms 54.4680ms 18.3594 Ops/s 18.2879 Ops/s $\color{#35bf28}+0.39\%$
test_collector_without_rb_cuda[200-img_shape1-large_batch] 0.1088s 0.1085s 9.2185 Ops/s 9.1248 Ops/s $\color{#35bf28}+1.03\%$
test_collector_with_rb_cuda[100-img_shape0-atari] 56.7828ms 56.5164ms 17.6940 Ops/s 17.7418 Ops/s $\color{#d91a1a}-0.27\%$
test_collector_with_rb_cuda[200-img_shape1-large_batch] 0.1132s 0.1124s 8.8986 Ops/s 8.8959 Ops/s $\color{#35bf28}+0.03\%$

@vmoens
Update
88ff619 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Feb 16, 2026
@vmoens @cursoragent
[Feature] AsyncBatchedCollector: backend params and performance optim…
c82d890 
…izations

- Three-tier backend system: `backend` (global default), `env_backend`
  (env pool override), `policy_backend` (transport override), mirroring
  the device parameter pattern.
- Lock-free SlotTransport: per-env slots with no shared lock, replacing
  ThreadingTransport as the default for in-process threading.
- min_batch_size parameter for InferenceServer to accumulate requests.
- Batch drain from result queue (get_nowait after first blocking get).
- Remove redundant .copy() in ProcessorAsyncEnvPool._env_exec.

Co-authored-by: Cursor <cursoragent@cursor.com>
ghstack-source-id: 5b0282d
Pull-Request: #3511
Co-authored-by: Cursor <cursoragent@cursor.com>
@vmoens
Update
8c2309c 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Feb 17, 2026
@vmoens @cursoragent
[Feature] AsyncBatchedCollector: backend params and performance optim…
a761379 
…izations

- Three-tier backend system: `backend` (global default), `env_backend`
  (env pool override), `policy_backend` (transport override), mirroring
  the device parameter pattern.
- Lock-free SlotTransport: per-env slots with no shared lock, replacing
  ThreadingTransport as the default for in-process threading.
- min_batch_size parameter for InferenceServer to accumulate requests.
- Batch drain from result queue (get_nowait after first blocking get).
- Remove redundant .copy() in ProcessorAsyncEnvPool._env_exec.

Co-authored-by: Cursor <cursoragent@cursor.com>
ghstack-source-id: d7fc567
Pull-Request: #3511
Co-authored-by: Cursor <cursoragent@cursor.com>
@vmoens
Update
2d201bf 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Feb 20, 2026
@vmoens @cursoragent
[Feature] AsyncBatchedCollector: backend params and performance optim…
b41273c 
…izations

- Three-tier backend system: `backend` (global default), `env_backend`
  (env pool override), `policy_backend` (transport override), mirroring
  the device parameter pattern.
- Lock-free SlotTransport: per-env slots with no shared lock, replacing
  ThreadingTransport as the default for in-process threading.
- min_batch_size parameter for InferenceServer to accumulate requests.
- Batch drain from result queue (get_nowait after first blocking get).
- Remove redundant .copy() in ProcessorAsyncEnvPool._env_exec.

Co-authored-by: Cursor <cursoragent@cursor.com>
ghstack-source-id: 3e3cd93
Pull-Request: #3511
Co-authored-by: Cursor <cursoragent@cursor.com>
@github-actions github-actions bot added the Documentation Improvements or additions to documentation label Feb 20, 2026
vmoens added a commit that referenced this pull request Feb 21, 2026
@vmoens @cursoragent
[Feature] AsyncBatchedCollector: backend params and performance optim…
8325c4d 
…izations

- Three-tier backend system: `backend` (global default), `env_backend`
  (env pool override), `policy_backend` (transport override), mirroring
  the device parameter pattern.
- Lock-free SlotTransport: per-env slots with no shared lock, replacing
  ThreadingTransport as the default for in-process threading.
- min_batch_size parameter for InferenceServer to accumulate requests.
- Batch drain from result queue (get_nowait after first blocking get).
- Remove redundant .copy() in ProcessorAsyncEnvPool._env_exec.

Co-authored-by: Cursor <cursoragent@cursor.com>
ghstack-source-id: 3e3cd93
Pull-Request: #3511
Co-authored-by: Cursor <cursoragent@cursor.com>
vmoens added a commit that referenced this pull request Feb 21, 2026
@vmoens @cursoragent
[Feature] AsyncBatchedCollector: backend params and performance optim…
d192a07 
…izations (#3511)

- Three-tier backend system: `backend` (global default), `env_backend`
  (env pool override), `policy_backend` (transport override), mirroring
  the device parameter pattern.
- Lock-free SlotTransport: per-env slots with no shared lock, replacing
  ThreadingTransport as the default for in-process threading.
- min_batch_size parameter for InferenceServer to accumulate requests.
- Batch drain from result queue (get_nowait after first blocking get).
- Remove redundant .copy() in ProcessorAsyncEnvPool._env_exec.

Co-authored-by: Cursor <cursoragent@cursor.com>
ghstack-source-id: 3e3cd93
Pull-Request: #3511
Co-authored-by: Cursor <cursoragent@cursor.com>
@vmoens vmoens merged commit 2d201bf into gh/vmoens/242/base Feb 21, 2026
114 of 116 checks passed
@vmoens vmoens deleted the gh/vmoens/242/head branch February 21, 2026 21:12
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

Benchmarks rl/benchmark changes CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. Collectors Documentation Improvements or additions to documentation Examples Feature New feature Modules

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@vmoens