SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1810.09103 · 2018

Greedy Actor-Critic: A New Conditional Cross-Entropy Method for Policy Improvement

arXiv · PDF · Open in the Atlas

Code that ran

6 of 10 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; 3 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 0 failures of our instrument; 4 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/1810.09103.json · badge: /api/ran/1810.09103.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
samuelfneumann/greedyac canonical (classifier label)
classifier basis: census
6 ran · 4 no recorded run
FunctionStatusWhere it lives
angle_normalize Ran
ran, honoured the contract we drafted
samuelfneumann/greedyac/env/PendulumEnv.py
code served (permissive licence) · get_code("eef88b4b150868c5")
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 float, 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.

import numpy as np

# angle_normalize takes a single float parameter x
# The function normalizes an angle to the range [-π, π]
# We'll test with a variety of angles

ARGS = {
    "x": 2 * np.pi + 0.5  # An angle outside [-π, π] that should be normalized
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
bound Ran
ran, honoured the contract we drafted
samuelfneumann/greedyac/env/Acrobot.py
code served (permissive licence) · get_code("9491f5abebf0343a")
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 int, ndim 0, dtype family int, 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.

import torch

torch.manual_seed(0)

# The function bound takes three parameters: x, m, and M
# x is a scalar value to be bounded
# m can be either a scalar (when M is provided) or a length-2 vector [min, max]
# M is optional - if None, m is treated as a vector [min, max]

# Test case 1: x as scalar, m as scalar, M as scalar
x_scalar = 5
m_scalar = 1
M_scalar = 10

ARGS = {
    "x": x_scalar,
    "m": m_scalar,
    "M": M_scalar
}
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 int, ndim 0, dtype family int, 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
get_returns Ran
ran on a synthesized input
samuelfneumann/greedyac/utils/experiment_utils.py
code served (permissive licence) · get_code("9c4b0e2408f1e88c")
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 ndarray, ndim 3, dtype family float, shape (2, 1, 5) 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.

import numpy as np

# Build the data dictionary structure expected by get_returns
data = {
    'experiment': {
        'environment': {
            'eval_episodes': 5
        }
    },
    'experiment_data': [
        {
            'runs': [
                {
                    'eval_episode_rewards': np.array([[1.0, 2.0, 3.0, 4.0, 5.0]]),
                    'train_episode_rewards': np.array([10.0, 20.0, 30.0, 40.0])
                },
                {
                    'eval_episode_rewards': np.array([[6.0, 7.0, 8.0, 9.0, 10.0]]),
                    'train_episode_rewards': np.array([50.0, 60.0, 70.0, 80.0])
                }
            ]
        }
    ]
}

ARGS = {
    'data': data,
    'type_': 'eval',
    'ind': 0,
    'env_type': 'continuing'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (2, 1, 5) — shape (2, 1, 5) 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
sweeps Ran
ran on a synthesized input
samuelfneumann/greedyac/utils/hypers.py
code served (permissive licence) · get_code("c16b6724c146799d")
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 tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 numpy as np

# Create a parameters dictionary with batch_size and replay_capacity
parameters = {
    'batch_size': [16, 32, 64],
    'replay_capacity': [100, 200, 500, 1000],
    'learning_rate': [0.001, 0.01, 0.1],
    'discount_factor': [0.9, 0.99]
}

index = 5

ARGS = {
    'parameters': parameters,
    'index': index
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
total Ran
ran on a synthesized input
samuelfneumann/greedyac/utils/hypers.py
code served (permissive licence) · get_code("d7a05ffbc1dcccfc")
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 int, ndim 0, dtype family int, 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.

import numpy as np

# Construct valid arguments for the `total` function
# The function takes a `parameters` dict and calls sweeps(parameters, 0)[1]

parameters = {
    'learning_rate': [0.001, 0.01, 0.1],
    'batch_size': np.array([32, 64, 128]),
    'replay_capacity': np.array([1000, 2000, 4000, 8000]),
    'hidden_units': [64, 128, 256],
}

ARGS = {
    'parameters': parameters
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, 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
wrap Ran
ran, honoured the contract we drafted
samuelfneumann/greedyac/env/Acrobot.py
code served (permissive licence) · get_code("a91d2ed86e80c635")
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 float, 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.

import torch
import numpy as np

torch.manual_seed(0)

# The function wrap() takes three scalar parameters: x, m, M
# x is the value to wrap
# m is the minimum bound
# M is the maximum bound
# The function wraps x to be within [m, M] by repeatedly adding/subtracting the difference

# Simple scalar values work here
x = 360.0  # A value that needs wrapping
m = -180.0  # Minimum bound
M = 180.0   # Maximum bound

ARGS = {
    "x": x,
    "m": m,
    "M": M
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
create_agent Not yet run
unverified (harvested, no recorded run)
samuelfneumann/greedyac/utils/experiment_utils.py
code served (permissive licence) · get_code("b59edaafb5dc160b")
How it ran

no invocation recorded.

get_mean_err Not yet run
unverified (harvested, no recorded run)
samuelfneumann/greedyac/utils/experiment_utils.py
code served (permissive licence) · get_code("a867ab3556920304")
How it ran

no invocation recorded.

rk4 Not yet run
unverified (harvested, no recorded run)
samuelfneumann/greedyac/env/Acrobot.py
code served (permissive licence) · get_code("849841b638e7d288")
How it ran

no invocation recorded.

satisfies Not yet run
unverified (harvested, no recorded run)
samuelfneumann/greedyac/utils/hypers.py
code served (permissive licence) · get_code("33b22af6a3fbd5a8")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 3 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 3 later) and 7 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
angle_normalize 2 other papers
1 dated earlier, 1 later
1605.06432 (2016-05)
Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data
2106.12782 (2021-06)
Hamiltonian-based Neural ODE Networks on the SE(3) Manifold For Dynamics Learning and Control
bound 1 other paper
1 later
2002.06836 (2020-02)
Control Frequency Adaptation via Action Persistence in Batch Reinforcement Learning
wrap 1 other paper
1 later
2002.06836 (2020-02)
Control Frequency Adaptation via Action Persistence in Batch Reinforcement Learning

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

Many policy gradient methods are variants of Actor-Critic (AC), where a value function (critic) is learned to facilitate updating the parameterized policy (actor). The update to the actor involves a log-likelihood update weighted by the action-values, with the addition of entropy regularization for soft variants. In this work, we explore an alternative update for the actor, based on an extension of the cross entropy method (CEM) to condition on inputs (states). The idea is to start with a broader policy and slowly concentrate around maximal actions, using a maximum likelihood update towards actions in the top percentile per state. The speed of this concentration is controlled by a proposal policy, that concentrates at a slower rate than the actor. We first provide a policy improvement result in an idealized setting, and then prove that our conditional CEM (CCEM) strategy tracks a CEM update per state, even with changing action-values. We empirically show that our Greedy AC algorithm, that uses CCEM for the actor update, performs better than Soft Actor-Critic and is much less sensitive to entropy-regularization.

For agents

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

get_harvested_code_for_paper("1810.09103")
get_code_for_paper("1810.09103")
have("1810.09103")

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

curl https://syntology.ai/api/ran/1810.09103.json

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

[![Syntology run record](https://syntology.ai/api/ran/1810.09103.svg)](https://syntology.ai/paper/1810.09103)

Connect an agent — have() is free.