SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2004.14309 · 2020

How to Learn a Useful Critic? Model-based Action-Gradient-Estimator Policy Optimization

arXiv · PDF · Open in the Atlas

Code that ran

5 of 6 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 1 failure of our instrument; 1 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/2004.14309.json · badge: /api/ran/2004.14309.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
angelolovatto/raylab community (classifier label)
classifier basis not recorded on the link
4 ran · 1 no recorded run
maximerobeyns/taylortd extension (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
configure Ran
ran on a synthesized input
angelolovatto/raylab/raylab/options.py
code served (permissive licence) · get_code("5bd21b59b9184273")
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 type. 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)

class MockOptions:
    """Mock options class that supports copy_and_set_queued_options."""
    def __init__(self):
        self.queued = {}
    
    def copy_and_set_queued_options(self):
        """Return a copy of self with queued options applied."""
        new_options = MockOptions()
        new_options.queued = self.queued.copy()
        return new_options

class MockClass:
    """Mock class to decorate with configure."""
    options = MockOptions()

ARGS = {
    "cls": MockClass
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
get_last_checkpoint_path Ran
ran on a synthesized input
angelolovatto/raylab/raylab/cli/best_checkpoint.py
code served (permissive licence) · get_code("aafa30b64abfcbaf")
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 str. 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 os
import tempfile
import shutil

# Create a temporary directory structure that mimics a Ray Tune logdir
tmpdir = tempfile.mkdtemp()
logdir = tmpdir

# Create checkpoint directories with numbered suffixes
checkpoint_dir_1 = os.path.join(logdir, 'checkpoint_1')
checkpoint_dir_2 = os.path.join(logdir, 'checkpoint_2')
checkpoint_dir_3 = os.path.join(logdir, 'checkpoint_10')

os.makedirs(checkpoint_dir_1, exist_ok=True)
os.makedirs(checkpoint_dir_2, exist_ok=True)
os.makedirs(checkpoint_dir_3, exist_ok=True)

# Create dummy files in the checkpoint directories so they appear valid
with open(os.path.join(checkpoint_dir_1, 'dummy.txt'), 'w') as f:
    f.write('checkpoint 1')
with open(os.path.join(checkpoint_dir_2, 'dummy.txt'), 'w') as f:
    f.write('checkpoint 2')
with open(os.path.join(checkpoint_dir_3, 'dummy.txt'), 'w') as f:
    f.write('checkpoint 10')

ARGS = {
    'logdir': logdir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
inner_product_last_dim Ran
ran; our contract draft was wrong, not the code
maximerobeyns/taylortd/MC_expctd_TD3/expected_td3.py
pointer only (licence: NONE) · get_code("1003d2f5e7abc4bc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, maximerobeyns/taylortd 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.

is_experiment_root Ran
ran on a synthesized input
angelolovatto/raylab/raylab/cli/experiment_dashboard.py
code served (permissive licence) · get_code("994a6a2e9045d553")
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 bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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.

# Setup code for is_experiment_root function

def is_experiment_root(path: str) -> bool:
    return path.startswith('experiment_state') and path.endswith('.json')

# The function takes a single parameter: path (a string)
# It checks if the path starts with 'experiment_state' and ends with '.json'

ARGS = {
    'path': 'experiment_state_config.json'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
produce_rollout Ran
ran on a synthesized input
angelolovatto/raylab/raylab/cli/module_dashboard.py
code served (permissive licence) · get_code("4b99f0e38ded336a")
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 dict. 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)

class MockWorker:
    def sample(self):
        # Return a simple dict representing a rollout sample
        return {
            "observations": torch.randn(2, 4),
            "actions": torch.randn(2, 3),
            "rewards": torch.randn(2),
        }

class MockWorkers:
    def __init__(self):
        self._local_worker = MockWorker()
    
    def local_worker(self):
        return self._local_worker

class MockAgent:
    def __init__(self):
        self.workers = MockWorkers()

agent = MockAgent()

ARGS = {
    "agent": agent
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
get_agent_cls Not yet run
unverified (harvested, no recorded run)
angelolovatto/raylab/raylab/agents/registry.py
code served (permissive licence) · get_code("8ed9cb782d14f9e5")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 6 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.

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

Deterministic-policy actor-critic algorithms for continuous control improve the actor by plugging its actions into the critic and ascending the action-value gradient, which is obtained by chaining the actor's Jacobian matrix with the gradient of the critic with respect to input actions. However, instead of gradients, the critic is, typically, only trained to accurately predict expected returns, which, on their own, are useless for policy optimization. In this paper, we propose MAGE, a model-based actor-critic algorithm, grounded in the theory of policy gradients, which explicitly learns the action-value gradient. MAGE backpropagates through the learned dynamics to compute gradient targets in temporal difference learning, leading to a critic tailored for policy improvement. On a set of MuJoCo continuous-control tasks, we demonstrate the efficiency of the algorithm in comparison to model-free and model-based state-of-the-art baselines.

For agents

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

get_harvested_code_for_paper("2004.14309")
get_code_for_paper("2004.14309")
have("2004.14309")

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

curl https://syntology.ai/api/ran/2004.14309.json

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

[![Syntology run record](https://syntology.ai/api/ran/2004.14309.svg)](https://syntology.ai/paper/2004.14309)

Connect an agent — have() is free.