SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.09359 · 2020

AWAC: Accelerating Online Reinforcement Learning with Offline Datasets

Abhishek Gupta, Sergey Levine, Ashvin Nair, Murtaza Dalal

arXiv · PDF · Open in the Atlas

Code that ran

10 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (9 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 9 have no contract verdict; 1 failure of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2006.09359.json · badge: /api/ran/2006.09359.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
zzmtsvv/rl_task community
Papers with Code archive
4 ran · 1 no recorded run
HzcIrving/DLRL-PlayGround community
Papers with Code archive
3 ran · 1 no recorded run
linhlpv/Offline-and-OfflineOnline-RL community
Papers with Code archive
1 ran · 1 no recorded run
corl-team/CORL community
Papers with Code archive
1 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
AWACActor Ran
ran on a synthesized input · constructed an object; no result was computed
HzcIrving/DLRL-PlayGround/Offline RL/AWAC/Networks.py
code served (permissive licence) · get_code("848704f003bcbe0a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, AWACActor: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'obs_dim': 4,
    'act_dim': 3,
    'hidden_sizes': (8, 8),
    'activation': nn.ReLU,
    'act_limit': 1.0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type AWACActor — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a AWACActor output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Actor Ran
ran on a synthesized input · constructed an object; no result was computed
zzmtsvv/rl_task/awac/awac.py
pointer only (licence: NONE) · get_code("eecebb628e9b243d")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zzmtsvv/rl_task when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

AdvantageWeightedActorCritic Ran
ran on a synthesized input · constructed an object; no result was computed
linhlpv/Offline-and-OfflineOnline-RL/AWAC/awac.py
pointer only (licence: NONE) · get_code("ad518f56352757d0")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, linhlpv/Offline-and-OfflineOnline-RL when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

AdvantageWeightedActorCritic Ran
ran on a synthesized input · constructed an object; no result was computed
corl-team/CORL/algorithms/finetune/awac.py
code served (permissive licence) · get_code("d79f983080b6e99f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, AdvantageWeightedActorCritic: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Define a simple actor network that outputs actions and log probs
class SimpleActor(nn.Module):
    def __init__(self, state_dim=4, action_dim=3):
        super().__init__()
        self.net = nn.Linear(state_dim, 64)
        self.action_head = nn.Linear(64, action_dim)
        self.log_std = nn.Parameter(torch.zeros(action_dim))
    
    def forward(self, states):
        # Returns (actions, log_probs)
        x = torch.relu(self.net(states))
        actions = torch.tanh(self.action_head(x))
        log_probs = torch.zeros(states.shape[0], 1)
        return actions, log_probs
    
    def log_prob(self, states, actions):
        # Returns log probability of actions given states
        return torch.zeros(states.shape[0])

# Define a simple critic network
class SimpleCritic(nn.Module):
    def __init__(self, state_dim=4, action_dim=3):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(state_dim + action_dim, 64),
            nn.ReLU(),
            nn.Linear(64, 1)
        )
    
    def forward(self, states, actions):
        # Concatenate states and actions
        x = torch.cat([states, actions], dim=-1)
        return self.net(x)

# Create actor and critic networks
actor = SimpleActor(state_dim=4, action_dim=3)
critic_1 = SimpleCritic(state_dim=4, action_dim=3)
critic_2 = SimpleCritic(state_dim=4, action_dim=3)

# Create optimizers
actor_optimizer = torch.optim.Adam(actor.parameters(), lr=0.001)
critic_1_optimizer = torch.optim.Adam(critic_1.parameters(), lr=0.001)
critic_2_optimizer = torch.optim.Adam(critic_2.parameters(), lr=0.001)

ARGS = {
    'actor': actor,
    'actor_optimizer': actor_optimizer,
    'critic_1': critic_1,
    'critic_1_optimizer': critic_1_optimizer,
    'critic_2': critic_2,
    'critic_2_optimizer': critic_2_optimizer,
    'gamma': 0.99,
    'tau': 0.005,
    'awac_lambda': 1.0,
    'exp_adv_max': 100.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type AdvantageWeightedActorCritic — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a AdvantageWeightedActorCritic output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
EnsembleLinear Ran
ran on a synthesized input · constructed an object; no result was computed
zzmtsvv/rl_task/awac/awac.py
pointer only (licence: NONE) · get_code("4440ae74ec82d1e3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zzmtsvv/rl_task when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

EnsembledCritic Ran
ran on a synthesized input · constructed an object; no result was computed
zzmtsvv/rl_task/awac/awac.py
pointer only (licence: NONE) · get_code("2bde489d14627922")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zzmtsvv/rl_task when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

MLPQFunction Ran
ran on a synthesized input · constructed an object; no result was computed
HzcIrving/DLRL-PlayGround/Offline RL/AWAC/Networks.py
code served (permissive licence) · get_code("3817b56031f35acd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, MLPQFunction: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Define parameters for MLPQFunction
obs_dim = 4
act_dim = 3
hidden_sizes = [8, 8]
activation = nn.ReLU

ARGS = {
    "obs_dim": obs_dim,
    "act_dim": act_dim,
    "hidden_sizes": hidden_sizes,
    "activation": activation,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MLPQFunction — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MLPQFunction output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
MLPVFunction Ran
ran on a synthesized input · constructed an object; no result was computed
HzcIrving/DLRL-PlayGround/Offline RL/AWAC/Networks.py
code served (permissive licence) · get_code("ba9e014f75dcd55d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, MLPVFunction: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'obs_dim': 4,
    'hidden_sizes': (8, 16),
    'activation': nn.ReLU
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MLPVFunction — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MLPVFunction output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
awac_config Ran
ran on a synthesized input · constructed an object; no result was computed
zzmtsvv/rl_task/awac/awac.py
pointer only (licence: NONE) · get_code("10cced0b9bb6e19f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zzmtsvv/rl_task when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

mlp Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from idsia/policyevaluator
pointer only; the stored copy get_code serves, from idsia/policyevaluator, is pointer only (licence: NONE) · get_code("7569e2a512ec417c")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

AWAC Not yet run
unverified (harvested, no recorded run)
zzmtsvv/rl_task/awac/awac.py
pointer only (licence: NONE) · get_code("3cb57c71a483a705")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zzmtsvv/rl_task when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

AWActorCritic Not yet run
unverified (harvested, no recorded run)
HzcIrving/DLRL-PlayGround/Offline RL/AWAC/Networks.py
code served (permissive licence) · get_code("9d49fcc185cc1eb5")
How it ran

no invocation recorded.

soft_update Not yet run
unverified (harvested, no recorded run)
linhlpv/Offline-and-OfflineOnline-RL/AWAC/awac.py
pointer only (licence: NONE) · get_code("1ddf841eed24fa3a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, linhlpv/Offline-and-OfflineOnline-RL when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 2 the same month, 2 later) and 11 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
mlp
brick of this paper's assembly: model
12 other papers
1 dated earlier, 1 the same month, 10 later · 2 only as an assembly brick
1905.01240 (2019-05)
Information asymmetry in KL-regularized RL
2006.07442 (2020-06)
Self-Imitation Learning via Generalized Lower Bound Q-learning
2010.09635 (2020-10)
Deep Reinforcement Learning with Population-Coded Spiking Neural Network for Continuous Control
and 9 more not named here
soft_update
brick of this paper's assembly: update_rule
6 other papers
1 the same month, 5 later · 5 only as an assembly brick
2006.04779 (2020-06)
Conservative Q-Learning for Offline Reinforcement Learning
2110.01548 (2021-10)
Uncertainty-Based Offline Reinforcement Learning with Diversified Q-Ensemble
2110.06169 (2021-10)
Offline Reinforcement Learning with Implicit Q-Learning
and 3 more not named here

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Reinforcement learning (RL) provides an appealing formalism for learning control policies from experience. However, the classic active formulation of RL necessitates a lengthy active exploration process for each behavior, making it difficult to apply in real-world settings such as robotic control. If we can instead allow RL algorithms to effectively use previously collected data to aid the online learning process, such applications could be made substantially more practical: the prior data would provide a starting point that mitigates challenges due to exploration and sample complexity, while the online training enables the agent to perfect the desired skill. Such prior data could either constitute expert demonstrations or, more generally, sub-optimal prior data that illustrates potentially useful transitions. While a number of prior methods have either used optimal demonstrations to bootstrap reinforcement learning, or have used sub-optimal data to train purely offline, it remains exceptionally difficult to train a policy with potentially sub-optimal offline data and actually continue to improve it further with online RL. In this paper we systematically analyze why this problem is so challenging, and propose an algorithm that combines sampleefficient dynamic programming with maximum likelihood policy updates, providing a simple and effective framework that is able to leverage large amounts of offline data and then quickly perform online fine-tuning of RL policies. We show that our method, advantage weighted actor critic (AWAC), enables rapid learning of skills with a combination of prior demonstration data and online experience. We demonstrate these benefits on a variety of simulated and real-world robotics domains, including dexterous manipulation with a real multi-fingered hand, drawer opening with a robotic arm, and rotating a valve. Our results show that incorporating prior data can reduce the time required to learn a range of robotic skills to practical time-scales.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2006.09359")
get_code_for_paper("2006.09359")
have("2006.09359")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2006.09359.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2006.09359.svg)](https://syntology.ai/paper/2006.09359)

Connect an agent — have() is free.