SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.07869 · NeurIPS · 2020

Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks

Filippos Christianos, Lukas Schäfer, Stefano Albrecht, Georgios Papoudakis

arXiv · PDF · Open in the Atlas

Code that ran

11 of 12 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; 1 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 6 failures 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/2006.07869.json · badge: /api/ran/2006.07869.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
ShashwatNigam99/MARBLER community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
dtabas/epymarl community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
3 ran · 0 no recorded run
semitable/robotic-warehouse community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
FunctionStatusWhere it lives
convert_to_robotarium_poses Ran
ran on a synthesized input
ShashwatNigam99/MARBLER/robotarium_gym/utilities/misc.py
code served (permissive licence) · get_code("176e514fbf32fef9")
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 2, dtype family float, shape (4, 4) 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

# The function expects 'locations' which should be array-like
# It treats locations as N points, gets shape[0] (number of points)
# and transposes them, then vstacks with zeros
# So locations should be (N, D) shaped array where N is number of poses, D is dimensions

locations = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0],
    [7.0, 8.0, 9.0],
    [10.0, 11.0, 12.0]
])

ARGS = {
    "locations": locations
}
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 2, dtype family float, shape (4, 4) — shape (4, 4) 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
extract_alg_name_from_config Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from uoe-agents/epymarl
pointer only · get_code("fa9a1eda0ac523d2")
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.

extract_env_name_from_config Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from uoe-agents/epymarl
pointer only · get_code("cf0832c848510d11")
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.

get_max_reward Ran
ran on a synthesized input
ShashwatNigam99/MARBLER/epymarl-logparse.py
code served (permissive licence) · get_code("e766f65debc9e598")
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 os
import json
import tempfile
import numpy as np

# Create a temporary directory structure with the required JSON files
temp_dir = tempfile.mkdtemp()

# Create subdirectories for hyperparameter combinations (numeric folder names)
for i in range(2):
    folder_path = os.path.join(temp_dir, str(i))
    os.makedirs(folder_path, exist_ok=True)
    
    # Create metrics.json
    metrics_data = {
        'test_return_mean': {
            'steps': [100, 200, 300, 400],
            'values': [10.5, 15.3, 20.1, 18.9]
        },
        'test_return_std': {
            'steps': [100, 200, 300, 400],
            'values': [0.5, 0.6, 0.7, 0.65]
        }
    }
    with open(os.path.join(folder_path, 'metrics.json'), 'w') as f:
        json.dump(metrics_data, f)
    
    # Create run.json
    run_data = {
        'meta': {
            'config_updates': {
                'learning_rate': 0.001 * (i + 1),
                'batch_size': 32 + (i * 16),
                'num_layers': 2 + i
            }
        }
    }
    with open(os.path.join(folder_path, 'run.json'), 'w') as f:
        json.dump(run_data, f)

ARGS = {
    'path': temp_dir
}
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 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_nearest_neighbors Ran
ran on a synthesized input
ShashwatNigam99/MARBLER/robotarium_gym/utilities/misc.py
code served (permissive licence) · get_code("7fbfb1102413922b")
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 1, dtype family int, shape (2,) 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 get_nearest_neighbors
# The function expects:
# - poses: a numpy array of shape (2, N) where N is number of agents
# - agent: an integer index (agent ID)
# - num_neighbors: an integer for how many neighbors to return

# Create a small poses array with 2D positions for 5 agents
# Shape: (2, 5) - first dimension is x,y coordinates, second is agent index
poses = np.array([
    [0.0, 1.0, 2.0, 3.0, 4.0],  # x coordinates
    [0.0, 1.0, 0.5, 1.5, 2.0]   # y coordinates
], dtype=np.float32)

# Pick an agent index (must be valid, between 0 and poses.shape[1]-1)
agent = 2

# Number of neighbors to return (must be less than total agents - 1)
num_neighbors = 2

ARGS = {
    "poses": poses,
    "agent": agent,
    "num_neighbors": num_neighbors
}
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 1, dtype family int, shape (2,) — shape (2,) 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
gumbel_softmax_sample Ran
ran; our contract draft was wrong, not the code
dtabas/epymarl/src/controllers/maddpg_controller.py
pointer only (licence: Apache-2.0) · get_code("355d7b51c7ffd042")
How it ran

The body of this sample is withheld: Apache-2.0, as GitHub detects it on this paper's repository, dtabas/epymarl, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with 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.

is_close Ran
ran on a synthesized input
ShashwatNigam99/MARBLER/robotarium_gym/utilities/misc.py
code served (permissive licence) · get_code("a4f3813a0404f041")
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

# agent_poses is indexed with [:2, agent_index], so it needs:
# - First dimension of at least 2 (for x, y coordinates)
# - Second dimension needs enough columns for agent_index
# Using shape (2, 3) to have 2 coordinates and 3 agents
agent_poses = np.array([
    [1.0, 5.0, 3.0],  # x coordinates for agents 0, 1, 2
    [2.0, 6.0, 4.0]   # y coordinates for agents 0, 1, 2
])

# agent_index selects which agent to check
agent_index = 1

# prey_loc will be reshaped to (1, 2) and indexed [0], so needs to have 2 elements
prey_loc = np.array([5.5, 6.5])

# sensing_radius is compared with distance
sensing_radius = 1.0

ARGS = {
    "agent_poses": agent_poses,
    "agent_index": agent_index,
    "prey_loc": prey_loc,
    "sensing_radius": sensing_radius
}
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
load_results Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from uoe-agents/epymarl
pointer only · get_code("73de3ace29add240")
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.

onehot_from_logits Ran
ran; our contract draft was wrong, not the code
dtabas/epymarl/src/controllers/maddpg_controller.py
pointer only (licence: Apache-2.0) · get_code("046b92243aae4f76")
How it ran

The body of this sample is withheld: Apache-2.0, as GitHub detects it on this paper's repository, dtabas/epymarl, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with 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.

sample_gumbel Ran
ran; our contract draft was wrong, not the code
dtabas/epymarl/src/controllers/maddpg_controller.py
pointer only (licence: Apache-2.0) · get_code("3695d66ab681cacf")
How it ran

The body of this sample is withheld: Apache-2.0, as GitHub detects it on this paper's repository, dtabas/epymarl, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with 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.

work Ran
ran, honoured the contract we drafted
dtabas/epymarl/src/search.py
pointer only (licence: Apache-2.0) · get_code("930812c3b06fc858")
How it ran

The body of this sample is withheld: Apache-2.0, as GitHub detects it on this paper's repository, dtabas/epymarl, is in the permissive set, but a licence decision keeps the body a pointer: the reason for that decision is not recorded with 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.

get_display Not yet run
unverified (harvested, no recorded run)
semitable/robotic-warehouse/rware/rendering.py
code served (permissive licence) · get_code("d681f8b5572da32a")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 8 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 1 the same month, 8 later) and 4 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
onehot_from_logits 6 other papers
1 the same month, 5 later
2006.09447 (2020-06)
Agent Modelling under Partial Observability for Deep Reinforcement Learning
2210.06835 (2022-10)
Multi-agent Dynamic Algorithm Configuration
2212.05331 (2022-12)
Effects of Spectral Normalization in Multi-agent Reinforcement Learning
and 3 more not named here
gumbel_softmax_sample 5 other papers
5 later
2210.06835 (2022-10)
Multi-agent Dynamic Algorithm Configuration
2212.05331 (2022-12)
Effects of Spectral Normalization in Multi-agent Reinforcement Learning
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
and 2 more not named here
sample_gumbel 5 other papers
5 later
2210.06835 (2022-10)
Multi-agent Dynamic Algorithm Configuration
2212.05331 (2022-12)
Effects of Spectral Normalization in Multi-agent Reinforcement Learning
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
and 2 more not named here
work 5 other papers
1 dated earlier, 4 later
1910.07483 (2019-10)
MAVEN: Multi-Agent Variational Exploration
2212.05331 (2022-12)
Effects of Spectral Normalization in Multi-agent Reinforcement Learning
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
and 2 more not named here
extract_alg_name_from_config 4 other papers
1 dated earlier, 3 later
1902.04043 (2019-02)
The StarCraft Multi-Agent Challenge
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
2410.08540 (2024-10)
Kaleidoscope: Learnable Masks for Heterogeneous Multi-agent Reinforcement Learning
and 1 more not named here
extract_env_name_from_config 4 other papers
1 dated earlier, 3 later
1902.04043 (2019-02)
The StarCraft Multi-Agent Challenge
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
2410.08540 (2024-10)
Kaleidoscope: Learnable Masks for Heterogeneous Multi-agent Reinforcement Learning
and 1 more not named here
load_results 4 other papers
1 dated earlier, 3 later
1902.04043 (2019-02)
The StarCraft Multi-Agent Challenge
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
2410.08540 (2024-10)
Kaleidoscope: Learnable Masks for Heterogeneous Multi-agent Reinforcement Learning
and 1 more not named here
get_display 3 other papers
1 dated earlier, 2 later
1810.02912 (2018-10)
Actor-Attention-Critic for Multi-Agent Reinforcement Learning
2008.02840 (2020-08)
Assisted Perception: Optimizing Observations to Communicate State
2505.05262 (2025-05)
Enhancing Cooperative Multi-Agent Reinforcement Learning with State Modelling and Adversarial Exploration

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

Multi-agent deep reinforcement learning (MARL) suffers from a lack of commonlyused evaluation tasks and criteria, making comparisons between approaches difficult. In this work, we provide a systematic evaluation and comparison of three different classes of MARL algorithms (independent learning, centralised multiagent policy gradient, value decomposition) in a diverse range of cooperative multi-agent learning tasks. Our experiments serve as a reference for the expected performance of algorithms across different learning tasks, and we provide insights regarding the effectiveness of different learning approaches. We open-source EPy-MARL, which extends the PyMARL codebase to include additional algorithms and allow for flexible configuration of algorithm implementation details such as parameter sharing. Finally, we open-source two environments for multi-agent research which focus on coordination under sparse rewards.

For agents

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

get_harvested_code_for_paper("2006.07869")
get_code_for_paper("2006.07869")
have("2006.07869")

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

curl https://syntology.ai/api/ran/2006.07869.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.07869.svg)](https://syntology.ai/paper/2006.07869)

Connect an agent — have() is free.