[RLlib] IMPALA on new API stack (w/ EnvRunner- and ConnectorV2 APIs). #42085
+2,835
−1,739
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
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
-- no more context objects -- calls take rl_module -- ctors take obs- and action spaces (and maybe env) -- ctors do NOT take RLModule anymore -- env-to-module and learner connectors get constructed before(!) RLModule -- module-to-env connector gets constructed after(!) RLModule - StatelessCartPole still learning as well as before (see previous PRs) Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
…ner and ConnectorV2. New stack Atari - Pong deterministic + no frameskip + reduced action space - w/ EnvRunner and ConnectorV2s - 8 GPUs (4000/256 batch/minibatch per Learner) - SGD iters 10 - model: "vf_share_layers": True, "conv_filters": [[16, 4, 2], [32, 4, 2], [64, 4, 2], [128, 4, 2]], "conv_activation": "relu", "post_fcnet_hiddens": [256] - 59 rollout workers - 1 env per worker - other training settings: lambda_=0.95, kl_coeff=0.5, clip_param=0.1, vf_clip_param=10.0, entropy_coeff=0.01, grad_clip=100.0, grad_clip_by="global_norm" ----------------------------------- w/ ONLY 1 GPU (to compare to old stack) LR: 0.0005 num_rollout_workers: 95 (from 59) num_sgd_iter: 10 (back to original) actual RLlib Atari wrappers (grayscale, frameskip, episodic life were missing!) PRETTY DECENT! Trial status: 1 RUNNING Current time: 2023-12-13 07:08:15. Total running time: 16min 1s Logical resource usage: 96.0/96 CPUs, 1.0/8 GPUs (0.0/1.0 accelerator_type:V100) ╭──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮ │ Trial name status iter total time (s) ts reward episode_reward_max episode_reward_min episode_len_mean episodes_this_iter │ ├──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┤ │ PPO_env_33a1e_00000 RUNNING 239 885.251 956000 19.39 21 12 1762 3 │ ╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯ ----------------------------------- w/ 8 GPUs again LR: 0.002 num_rollout_workers: 95 (from 59) num_sgd_iter: 10 (back to original) actual RLlib Atari wrappers (grayscale, frameskip, episodic life were missing!) Trial status: 1 RUNNING Current time: 2023-12-13 07:16:35. Total running time: 7min 30s Logical resource usage: 96.0/96 CPUs, 8.0/8 GPUs (0.0/1.0 accelerator_type:V100) ╭───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮ │ Trial name status iter total time (s) ts reward episode_reward_max episode_reward_min episode_len_mean episodes_this_iter │ ├───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┤ │ PPO_env_8dc32_00000 RUNNING 46 369.115 1472000 19.8 21 15 1738.74 25 │ ╰───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯ ---------------------------------- Signed-off-by: sven1977 <[email protected]>
…ace_learner_hps_with_algo_config
Signed-off-by: sven1977 <[email protected]>
…nfig' into replace_learner_hps_with_algo_config
…runner_support_connectors
Signed-off-by: sven1977 <[email protected]>
…ace_learner_hps_with_algo_config
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
…runner_support_connectors
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
…nfig' into replace_learner_hps_with_algo_config
Signed-off-by: sven1977 <[email protected]>
…ace_learner_hps_with_algo_config
Signed-off-by: sven1977 <[email protected]>
…ace_learner_hps_with_algo_config
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
…_on_new_api_stack_w_env_runner_and_connectorv2
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
…_on_new_api_stack_w_env_runner_and_connectorv2
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
…_on_new_api_stack_w_env_runner_and_connectorv2
Signed-off-by: sven1977 <[email protected]>
Signed-off-by: sven1977 <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
IMPALA on new API stack (w/ EnvRunner- and ConnectorV2 APIs).
Why are these changes needed?
Related issue number
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.