-
Notifications
You must be signed in to change notification settings - Fork 361
[Feature] TensorDictPrimer with single default_value callable #2732
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
vmoens
added a commit
that referenced
this pull request
Jan 30, 2025
ghstack-source-id: 172825a4bf036c332c9012e45d070fc0fe348a0d Pull Request resolved: #2732
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/2732
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 1 Cancelled Job, 2 Unrelated FailuresAs of commit d0ad84e with merge base 8c9dc05 ( NEW FAILURE - The following job has failed:
CANCELLED JOB - The following job was cancelled. Please retry:
BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Example script: from torchrl.envs import TensorDictPrimer, PendulumEnv, StepCounter, TrajCounter
import torch
from tensordict import assert_close
# let's build a primer that outputs the same Pendulum initial state 4 times in a row
env = PendulumEnv()
def val_iterator(N = 4, env=env):
i = 0
while True:
i += 1
r = env.reset()
for _ in range(N):
torch.manual_seed(i)
yield r
iterator = val_iterator()
def val_generator(iterator=iterator):
return next(iterator)
print('observation_spec', env.observation_spec)
primer = TensorDictPrimer(primers=env.observation_spec, default_value=val_generator, single_default_value=True)
env = env.append_transform(primer)
env = env.append_transform(StepCounter(max_steps=50))
env = env.append_transform(TrajCounter())
r = env.rollout(1000, break_when_any_done=False)
print(r[0]["th"], r[0]["thdot"])
print(r[50]["th"], r[50]["thdot"])
# To compute the GAE with empirical adv, we fist need to collapse the trajs
r["traj_count"] = r["traj_count"]//4
r["next", "traj_count"] = r["next", "traj_count"]//4
assert_close(r[:50], r[50:100])
assert_close(r[:50], r[100:150])
# Reshape r
print(r.shape)
r = r.reshape(-1, 4, 50)
assert r[0]["traj_count"].unique().numel() == 1 |
vmoens
added a commit
that referenced
this pull request
Feb 3, 2025
ghstack-source-id: b9f7df7bf2abd312dc8de56cac757c4b2975c62c Pull Request resolved: #2732
|
Name | Max | Mean | Ops | Ops on Repo HEAD
|
Change |
---|---|---|---|---|---|
test_simple | 0.6091s | 0.5242s | 1.9078 Ops/s | 1.9953 Ops/s | |
test_transformed | 1.1301s | 1.0452s | 0.9568 Ops/s | 0.9681 Ops/s | |
test_serial | 1.6408s | 1.5436s | 0.6479 Ops/s | 0.6512 Ops/s | |
test_parallel | 1.3916s | 1.3032s | 0.7673 Ops/s | 0.7602 Ops/s | |
test_step_mdp_speed[True-True-True-True-True] | 0.1772ms | 30.3643μs | 32.9334 KOps/s | 32.6400 KOps/s | |
test_step_mdp_speed[True-True-True-True-False] | 42.0200μs | 18.2386μs | 54.8289 KOps/s | 55.3440 KOps/s | |
test_step_mdp_speed[True-True-True-False-True] | 52.8100μs | 17.5393μs | 57.0150 KOps/s | 56.6458 KOps/s | |
test_step_mdp_speed[True-True-True-False-False] | 39.6050μs | 10.3882μs | 96.2630 KOps/s | 98.0335 KOps/s | |
test_step_mdp_speed[True-True-False-True-True] | 75.9940μs | 33.2766μs | 30.0511 KOps/s | 30.4378 KOps/s | |
test_step_mdp_speed[True-True-False-True-False] | 44.9450μs | 20.2635μs | 49.3497 KOps/s | 50.0526 KOps/s | |
test_step_mdp_speed[True-True-False-False-True] | 81.4810μs | 19.6370μs | 50.9243 KOps/s | 51.2091 KOps/s | |
test_step_mdp_speed[True-True-False-False-False] | 45.0950μs | 12.3038μs | 81.2757 KOps/s | 82.2219 KOps/s | |
test_step_mdp_speed[True-False-True-True-True] | 94.5390μs | 34.6725μs | 28.8413 KOps/s | 28.8837 KOps/s | |
test_step_mdp_speed[True-False-True-True-False] | 59.1620μs | 22.1739μs | 45.0982 KOps/s | 45.8259 KOps/s | |
test_step_mdp_speed[True-False-True-False-True] | 50.8370μs | 19.4629μs | 51.3797 KOps/s | 51.3829 KOps/s | |
test_step_mdp_speed[True-False-True-False-False] | 46.5280μs | 12.2654μs | 81.5301 KOps/s | 82.2614 KOps/s | |
test_step_mdp_speed[True-False-False-True-True] | 74.3320μs | 36.4387μs | 27.4434 KOps/s | 27.5165 KOps/s | |
test_step_mdp_speed[True-False-False-True-False] | 77.6010μs | 23.7465μs | 42.1115 KOps/s | 42.2972 KOps/s | |
test_step_mdp_speed[True-False-False-False-True] | 42.0700μs | 21.2267μs | 47.1105 KOps/s | 47.1683 KOps/s | |
test_step_mdp_speed[True-False-False-False-False] | 37.7710μs | 13.9835μs | 71.5127 KOps/s | 72.2021 KOps/s | |
test_step_mdp_speed[False-True-True-True-True] | 70.6340μs | 34.7140μs | 28.8068 KOps/s | 28.1863 KOps/s | |
test_step_mdp_speed[False-True-True-True-False] | 59.5640μs | 22.1637μs | 45.1189 KOps/s | 45.6589 KOps/s | |
test_step_mdp_speed[False-True-True-False-True] | 58.1200μs | 22.2846μs | 44.8740 KOps/s | 44.8082 KOps/s | |
test_step_mdp_speed[False-True-True-False-False] | 36.5490μs | 13.5895μs | 73.5860 KOps/s | 73.6049 KOps/s | |
test_step_mdp_speed[False-True-False-True-True] | 0.1007ms | 36.4335μs | 27.4472 KOps/s | 27.0703 KOps/s | |
test_step_mdp_speed[False-True-False-True-False] | 72.6770μs | 23.6606μs | 42.2643 KOps/s | 42.0693 KOps/s | |
test_step_mdp_speed[False-True-False-False-True] | 2.5610ms | 23.8901μs | 41.8583 KOps/s | 42.0576 KOps/s | |
test_step_mdp_speed[False-True-False-False-False] | 45.7970μs | 15.3431μs | 65.1760 KOps/s | 65.5372 KOps/s | |
test_step_mdp_speed[False-False-True-True-True] | 72.7180μs | 38.0451μs | 26.2846 KOps/s | 26.1322 KOps/s | |
test_step_mdp_speed[False-False-True-True-False] | 68.6000μs | 25.8228μs | 38.7255 KOps/s | 38.9689 KOps/s | |
test_step_mdp_speed[False-False-True-False-True] | 60.8050μs | 24.0185μs | 41.6345 KOps/s | 41.7499 KOps/s | |
test_step_mdp_speed[False-False-True-False-False] | 62.0210μs | 15.1981μs | 65.7975 KOps/s | 65.3246 KOps/s | |
test_step_mdp_speed[False-False-False-True-True] | 80.0930μs | 39.5115μs | 25.3091 KOps/s | 24.9022 KOps/s | |
test_step_mdp_speed[False-False-False-True-False] | 52.9910μs | 27.1673μs | 36.8089 KOps/s | 36.9817 KOps/s | |
test_step_mdp_speed[False-False-False-False-True] | 59.9840μs | 25.6127μs | 39.0431 KOps/s | 39.2083 KOps/s | |
test_step_mdp_speed[False-False-False-False-False] | 52.1590μs | 16.8212μs | 59.4488 KOps/s | 58.6264 KOps/s | |
test_values[generalized_advantage_estimate-True-True] | 10.9890ms | 9.7375ms | 102.6963 Ops/s | 97.8178 Ops/s | |
test_values[vec_generalized_advantage_estimate-True-True] | 25.4592ms | 24.1131ms | 41.4713 Ops/s | 41.2173 Ops/s | |
test_values[td0_return_estimate-False-False] | 0.2522ms | 0.1767ms | 5.6607 KOps/s | 5.6167 KOps/s | |
test_values[td1_return_estimate-False-False] | 25.7207ms | 24.0957ms | 41.5012 Ops/s | 40.6521 Ops/s | |
test_values[vec_td1_return_estimate-False-False] | 27.9720ms | 24.2519ms | 41.2339 Ops/s | 40.7073 Ops/s | |
test_values[td_lambda_return_estimate-True-False] | 37.5117ms | 35.1980ms | 28.4107 Ops/s | 28.4375 Ops/s | |
test_values[vec_td_lambda_return_estimate-True-False] | 26.5685ms | 24.2116ms | 41.3026 Ops/s | 41.2574 Ops/s | |
test_gae_speed[generalized_advantage_estimate-False-1-512] | 11.7673ms | 8.5152ms | 117.4375 Ops/s | 115.1205 Ops/s | |
test_gae_speed[vec_generalized_advantage_estimate-True-1-512] | 2.2076ms | 1.9390ms | 515.7303 Ops/s | 533.1293 Ops/s | |
test_gae_speed[vec_generalized_advantage_estimate-False-1-512] | 0.4959ms | 0.3715ms | 2.6915 KOps/s | 2.6708 KOps/s | |
test_gae_speed[vec_generalized_advantage_estimate-True-32-512] | 45.1792ms | 43.2924ms | 23.0987 Ops/s | 23.8673 Ops/s | |
test_gae_speed[vec_generalized_advantage_estimate-False-32-512] | 5.2308ms | 3.4832ms | 287.0885 Ops/s | 291.2284 Ops/s | |
test_dqn_speed[False-None] | 2.3976ms | 1.4322ms | 698.2309 Ops/s | 705.9476 Ops/s | |
test_dqn_speed[False-backward] | 2.0444ms | 1.9193ms | 521.0098 Ops/s | 522.9396 Ops/s | |
test_dqn_speed[True-None] | 0.7423ms | 0.4820ms | 2.0749 KOps/s | 1.9729 KOps/s | |
test_dqn_speed[True-backward] | 0.9563ms | 0.9059ms | 1.1038 KOps/s | 909.2286 Ops/s | |
test_dqn_speed[reduce-overhead-None] | 0.7112ms | 0.4936ms | 2.0260 KOps/s | 2.0394 KOps/s | |
test_dqn_speed[reduce-overhead-backward] | 0.9890ms | 0.9136ms | 1.0946 KOps/s | 1.0722 KOps/s | |
test_ddpg_speed[False-None] | 3.9407ms | 2.9735ms | 336.3020 Ops/s | 334.4878 Ops/s | |
test_ddpg_speed[False-backward] | 4.4258ms | 4.1277ms | 242.2684 Ops/s | 242.4775 Ops/s | |
test_ddpg_speed[True-None] | 1.8401ms | 1.2378ms | 807.9150 Ops/s | 803.5753 Ops/s | |
test_ddpg_speed[True-backward] | 2.1910ms | 2.1342ms | 468.5661 Ops/s | 419.1653 Ops/s | |
test_ddpg_speed[reduce-overhead-None] | 1.7117ms | 1.2369ms | 808.4464 Ops/s | 811.3037 Ops/s | |
test_ddpg_speed[reduce-overhead-backward] | 2.2908ms | 2.1358ms | 468.2074 Ops/s | 463.2291 Ops/s | |
test_sac_speed[False-None] | 8.6191ms | 8.1725ms | 122.3614 Ops/s | 121.1889 Ops/s | |
test_sac_speed[False-backward] | 12.1506ms | 10.9818ms | 91.0600 Ops/s | 90.4599 Ops/s | |
test_sac_speed[True-None] | 2.6602ms | 2.0935ms | 477.6740 Ops/s | 471.1064 Ops/s | |
test_sac_speed[True-backward] | 4.3504ms | 3.8378ms | 260.5672 Ops/s | 265.0584 Ops/s | |
test_sac_speed[reduce-overhead-None] | 2.5896ms | 2.1016ms | 475.8354 Ops/s | 474.7573 Ops/s | |
test_sac_speed[reduce-overhead-backward] | 3.9153ms | 3.8176ms | 261.9433 Ops/s | 262.8223 Ops/s | |
test_redq_speed[False-None] | 14.8691ms | 13.3388ms | 74.9690 Ops/s | 75.7892 Ops/s | |
test_redq_speed[False-backward] | 23.8230ms | 22.8594ms | 43.7457 Ops/s | 44.1594 Ops/s | |
test_redq_speed[True-None] | 5.5661ms | 4.9555ms | 201.7948 Ops/s | 195.7609 Ops/s | |
test_redq_speed[True-backward] | 13.3599ms | 12.5631ms | 79.5984 Ops/s | 77.5902 Ops/s | |
test_redq_speed[reduce-overhead-None] | 6.1849ms | 5.1366ms | 194.6827 Ops/s | 194.7702 Ops/s | |
test_redq_speed[reduce-overhead-backward] | 13.2352ms | 12.5546ms | 79.6522 Ops/s | 76.8510 Ops/s | |
test_redq_deprec_speed[False-None] | 14.9018ms | 13.0821ms | 76.4405 Ops/s | 76.1479 Ops/s | |
test_redq_deprec_speed[False-backward] | 21.9246ms | 18.8657ms | 53.0061 Ops/s | 52.1427 Ops/s | |
test_redq_deprec_speed[True-None] | 4.4671ms | 3.9116ms | 255.6525 Ops/s | 200.8119 Ops/s | |
test_redq_deprec_speed[True-backward] | 10.4768ms | 8.9520ms | 111.7068 Ops/s | 116.4033 Ops/s | |
test_redq_deprec_speed[reduce-overhead-None] | 4.2842ms | 3.8670ms | 258.6001 Ops/s | 258.6651 Ops/s | |
test_redq_deprec_speed[reduce-overhead-backward] | 9.4182ms | 8.4895ms | 117.7924 Ops/s | 118.8772 Ops/s | |
test_td3_speed[False-None] | 8.4200ms | 8.1747ms | 122.3282 Ops/s | 121.7917 Ops/s | |
test_td3_speed[False-backward] | 12.4221ms | 10.6584ms | 93.8229 Ops/s | 93.1054 Ops/s | |
test_td3_speed[True-None] | 2.0252ms | 1.7900ms | 558.6595 Ops/s | 556.9502 Ops/s | |
test_td3_speed[True-backward] | 4.6119ms | 3.4870ms | 286.7814 Ops/s | 291.1147 Ops/s | |
test_td3_speed[reduce-overhead-None] | 2.0413ms | 1.7877ms | 559.3738 Ops/s | 559.1760 Ops/s | |
test_td3_speed[reduce-overhead-backward] | 3.4837ms | 3.4093ms | 293.3118 Ops/s | 292.8839 Ops/s | |
test_cql_speed[False-None] | 40.2062ms | 37.0945ms | 26.9582 Ops/s | 26.6069 Ops/s | |
test_cql_speed[False-backward] | 50.6296ms | 47.5040ms | 21.0508 Ops/s | 20.7944 Ops/s | |
test_cql_speed[True-None] | 18.5668ms | 16.2284ms | 61.6204 Ops/s | 61.3553 Ops/s | |
test_cql_speed[True-backward] | 24.8546ms | 22.9863ms | 43.5042 Ops/s | 43.6497 Ops/s | |
test_cql_speed[reduce-overhead-None] | 17.8980ms | 16.1438ms | 61.9434 Ops/s | 60.5564 Ops/s | |
test_cql_speed[reduce-overhead-backward] | 24.6440ms | 22.9638ms | 43.5468 Ops/s | 42.3044 Ops/s | |
test_a2c_speed[False-None] | 8.6922ms | 7.2559ms | 137.8186 Ops/s | 134.2229 Ops/s | |
test_a2c_speed[False-backward] | 15.0099ms | 14.5538ms | 68.7104 Ops/s | 67.0312 Ops/s | |
test_a2c_speed[True-None] | 5.2917ms | 3.8145ms | 262.1559 Ops/s | 263.7465 Ops/s | |
test_a2c_speed[True-backward] | 10.5660ms | 10.1697ms | 98.3317 Ops/s | 93.3463 Ops/s | |
test_a2c_speed[reduce-overhead-None] | 4.1079ms | 3.7581ms | 266.0899 Ops/s | 258.2304 Ops/s | |
test_a2c_speed[reduce-overhead-backward] | 10.7545ms | 10.3145ms | 96.9513 Ops/s | 93.9316 Ops/s | |
test_ppo_speed[False-None] | 8.9221ms | 7.6398ms | 130.8927 Ops/s | 129.3263 Ops/s | |
test_ppo_speed[False-backward] | 16.9989ms | 15.1018ms | 66.2171 Ops/s | 64.3792 Ops/s | |
test_ppo_speed[True-None] | 4.7780ms | 4.0995ms | 243.9349 Ops/s | 235.1746 Ops/s | |
test_ppo_speed[True-backward] | 10.5354ms | 10.0290ms | 99.7111 Ops/s | 89.2607 Ops/s | |
test_ppo_speed[reduce-overhead-None] | 4.8741ms | 4.1044ms | 243.6418 Ops/s | 233.1278 Ops/s | |
test_ppo_speed[reduce-overhead-backward] | 11.1200ms | 10.1310ms | 98.7070 Ops/s | 95.2981 Ops/s | |
test_reinforce_speed[False-None] | 8.1278ms | 6.6660ms | 150.0146 Ops/s | 146.4154 Ops/s | |
test_reinforce_speed[False-backward] | 10.3877ms | 10.0322ms | 99.6791 Ops/s | 96.9830 Ops/s | |
test_reinforce_speed[True-None] | 3.8548ms | 3.0740ms | 325.3048 Ops/s | 316.4977 Ops/s | |
test_reinforce_speed[True-backward] | 9.3219ms | 9.0311ms | 110.7286 Ops/s | 110.0069 Ops/s | |
test_reinforce_speed[reduce-overhead-None] | 3.8839ms | 3.0806ms | 324.6130 Ops/s | 321.2275 Ops/s | |
test_reinforce_speed[reduce-overhead-backward] | 9.8189ms | 9.0271ms | 110.7771 Ops/s | 109.4994 Ops/s | |
test_iql_speed[False-None] | 34.9493ms | 32.5477ms | 30.7241 Ops/s | 29.3239 Ops/s | |
test_iql_speed[False-backward] | 50.2271ms | 45.8335ms | 21.8181 Ops/s | 21.8039 Ops/s | |
test_iql_speed[True-None] | 12.5664ms | 11.3092ms | 88.4238 Ops/s | 86.1622 Ops/s | |
test_iql_speed[True-backward] | 23.9083ms | 22.7959ms | 43.8676 Ops/s | 43.4940 Ops/s | |
test_iql_speed[reduce-overhead-None] | 12.6225ms | 11.3311ms | 88.2524 Ops/s | 86.3564 Ops/s | |
test_iql_speed[reduce-overhead-backward] | 24.1885ms | 22.5783ms | 44.2903 Ops/s | 42.9081 Ops/s | |
test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] | 7.5128ms | 4.8973ms | 204.1931 Ops/s | 197.3199 Ops/s | |
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] | 2.5144ms | 0.5354ms | 1.8678 KOps/s | 1.8869 KOps/s | |
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] | 0.8047ms | 0.5157ms | 1.9391 KOps/s | 2.0112 KOps/s | |
test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] | 5.0331ms | 4.7022ms | 212.6651 Ops/s | 213.7276 Ops/s | |
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] | 2.1599ms | 0.5191ms | 1.9263 KOps/s | 1.9268 KOps/s | |
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] | 0.7522ms | 0.4922ms | 2.0317 KOps/s | 2.0178 KOps/s | |
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] | 2.3393ms | 1.6881ms | 592.3916 Ops/s | 596.3778 Ops/s | |
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] | 1.8112ms | 1.5913ms | 628.4290 Ops/s | 633.2228 Ops/s | |
test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] | 5.1781ms | 4.7520ms | 210.4355 Ops/s | 208.0704 Ops/s | |
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] | 2.3309ms | 0.6596ms | 1.5160 KOps/s | 1.5142 KOps/s | |
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] | 0.9156ms | 0.6323ms | 1.5815 KOps/s | 1.5840 KOps/s | |
test_rb_iterate[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] | 7.4175ms | 4.7211ms | 211.8170 Ops/s | 214.7078 Ops/s | |
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] | 2.3445ms | 0.5247ms | 1.9057 KOps/s | 1.7952 KOps/s | |
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] | 0.8072ms | 0.5064ms | 1.9749 KOps/s | 1.9603 KOps/s | |
test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] | 5.4702ms | 4.6660ms | 214.3180 Ops/s | 213.6244 Ops/s | |
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] | 2.3846ms | 0.5194ms | 1.9253 KOps/s | 1.9336 KOps/s | |
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] | 0.7300ms | 0.4882ms | 2.0484 KOps/s | 2.0282 KOps/s | |
test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] | 5.1808ms | 4.8248ms | 207.2633 Ops/s | 211.2120 Ops/s | |
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] | 0.4787s | 1.3441ms | 744.0001 Ops/s | 1.5149 KOps/s | |
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] | 0.8263ms | 0.6363ms | 1.5716 KOps/s | 1.5745 KOps/s | |
test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] | 5.5559ms | 4.2563ms | 234.9471 Ops/s | 244.3263 Ops/s | |
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] | 7.9090ms | 2.3714ms | 421.6965 Ops/s | 421.5039 Ops/s | |
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] | 1.9273ms | 1.3423ms | 744.9785 Ops/s | 722.3525 Ops/s | |
test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] | 5.6566ms | 4.2927ms | 232.9547 Ops/s | 242.1526 Ops/s | |
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] | 5.9488ms | 2.3801ms | 420.1555 Ops/s | 430.8502 Ops/s | |
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] | 6.0647ms | 1.4445ms | 692.2713 Ops/s | 707.8488 Ops/s | |
test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] | 0.4379s | 13.1948ms | 75.7874 Ops/s | 32.7891 Ops/s | |
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] | 8.5752ms | 2.7474ms | 363.9817 Ops/s | 400.3796 Ops/s | |
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] | 5.7897ms | 1.6134ms | 619.7989 Ops/s | 632.1373 Ops/s | |
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] | 14.2223ms | 12.2122ms | 81.8851 Ops/s | 77.3982 Ops/s | |
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] | 19.2390ms | 14.5030ms | 68.9512 Ops/s | 70.0617 Ops/s | |
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] | 22.5150ms | 21.0125ms | 47.5906 Ops/s | 46.6781 Ops/s | |
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] | 16.1045ms | 14.6012ms | 68.4876 Ops/s | 68.2189 Ops/s | |
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] | 23.2778ms | 20.9544ms | 47.7227 Ops/s | 47.1472 Ops/s | |
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] | 26.3936ms | 16.1536ms | 61.9059 Ops/s | 62.6293 Ops/s |
Result of GPU Benchmark TestsExpand to view detailed results
|
This was referenced Feb 28, 2025
Merged
This was referenced Mar 3, 2025
vmoens
added a commit
that referenced
this pull request
Mar 3, 2025
ghstack-source-id: a9a677f24fc1e6a47312d0a96ab60daae543ff78 Pull Request resolved: #2732
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
enhancement
New feature or request
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack (oldest at bottom):