SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.03535 · NeurIPS · 2022

Influencing Long-Term Behavior in Multiagent Reinforcement Learning

Jakob Foerster, Dong-Ki Kim, Jonathan How, Miao Liu, Matthew Riemer, Gerald Tesauro, Michael Everett

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 (3 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 6 failures 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/2203.03535.json · badge: /api/ran/2203.03535.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
dkkim93/further marked official in Papers with Code
Papers with Code archive
10 ran · 3 no recorded run
FunctionStatusWhere it lives
CategoricalMLP Ran
ran on a synthesized input · constructed an object; no result was computed
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("e953502a76ebafce")
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, CategoricalMLP: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import logging

torch.manual_seed(0)

# Create a simple logger
logger = logging.getLogger('test_logger')
logger.setLevel(logging.INFO)
handler = logging.StreamHandler()
logger.addHandler(handler)

# Create log dictionary
log = {'test_log': logger}

# Create args object with required attributes
class Args:
    def __init__(self):
        self.n_hidden = 8
        self.log_name = 'test_log'

args = Args()

# Define ARGS dictionary with all parameters for CategoricalMLP
ARGS = {
    'n_input': 4,
    'n_output': 3,
    'name': 'policy',
    'log': log,
    'args': args,
    'device': torch.device('cpu')
}
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 CategoricalMLP — 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 CategoricalMLP 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
GaussianMLP Ran
ran on a synthesized input · constructed an object; no result was computed
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("e2120bd5095898b6")
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, GaussianMLP: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import logging

torch.manual_seed(0)

# Create a simple args object with required attributes
class Args:
    def __init__(self):
        self.n_hidden = 8
        self.log_name = 'test_logger'

args = Args()

# Create a logger and log dictionary
logger = logging.getLogger('test_logger')
logger.setLevel(logging.INFO)
log = {args.log_name: logger}

# Device is CPU only
device_cpu = torch.device('cpu')

# Build ARGS dictionary with all required parameters
ARGS = {
    'n_input': 4,
    'n_output': 3,
    'name': 'actor',
    'log': log,
    'args': args,
    'device': device_cpu
}
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 GaussianMLP — 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 GaussianMLP 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
ReplayMemory Ran
ran on a synthesized input · constructed an object; no result was computed
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("6cc0afec8e5e5562")
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, ReplayMemory: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 numpy as np
from collections import namedtuple

torch.manual_seed(0)

class Args:
    def __init__(self):
        self.memory_capacity = 10
        self.batch_size = 2

ARGS = {
    'args': Args()
}
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 ReplayMemory — 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 ReplayMemory 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_env_action_type Ran
ran; our contract draft was wrong, not the code
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("5d15c5576b6f1119")
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
import gym

# Create a simple discrete action space environment
class DiscreteEnv(gym.Env):
    def __init__(self):
        self.action_space = gym.spaces.Discrete(4)
        self.observation_space = gym.spaces.Box(low=0, high=1, shape=(2,), dtype=np.float32)

# Create the environment
env = DiscreteEnv()

# Define ARGS dict with the environment parameter
ARGS = {
    "env": env
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
gym, 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
get_env_dim Ran
ran, honoured the contract we drafted
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("3cc888ba40ec94e8")
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 int; 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 gym
import numpy as np

# Create a gym environment with Box observation and action spaces
env = gym.make('Pendulum-v1')

ARGS = {
    'env': env
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
gym, 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 int — 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
get_parameters Ran
ran; our contract draft was wrong, not the code
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("a460d87b686cc889")
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 odict_values. 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
from collections import OrderedDict

torch.manual_seed(0)

# Test case 1: OrderedDict
ordered_dict_network = OrderedDict([
    ('layer1', nn.Parameter(torch.randn(4, 3))),
    ('layer2', nn.Parameter(torch.randn(2, 4)))
])

# Test case 2: nn.Parameter
param_network = nn.Parameter(torch.randn(4, 3))

# Test case 3: torch.Tensor
tensor_network = torch.randn(4, 3)

# Test case 4: nn.Module with parameters
class SimpleNetwork(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
module_network = SimpleNetwork()

# Use the OrderedDict case as the primary test
ARGS = {
    'network': ordered_dict_network
}
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 odict_values — a odict_values output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a odict_values 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
reparameterization Ran
ran; our contract draft was wrong, not the code
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("92c1700ad0ee80cf")
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 Tensor, ndim 2, dtype family float, shape (2, 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 torch
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors for mean and logvar
# The function expects 2D tensors (batch_size, latent_dim) based on the operations
mean = torch.randn(2, 4)  # batch_size=2, latent_dim=4
logvar = torch.randn(2, 4)  # batch_size=2, latent_dim=4

ARGS = {
    "mean": mean,
    "logvar": logvar,
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
to_numpy Ran
ran; our contract draft was wrong, not the code
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("4f00b7f646079156")
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 (2, 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 torch
import numpy as np

torch.manual_seed(0)

# Create a small tensor on CPU
tensor = torch.randn(2, 4)

# Define target dtype for numpy conversion
dtype = np.float32

ARGS = {
    "tensor": tensor,
    "dtype": dtype
}
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 ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
to_onehot Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("818ef30cb3e0a466")
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 torch
import numpy as np

torch.manual_seed(0)

# Test case 1: numpy array input
numpy_value = np.array([0, 1, 2, 1], dtype=np.int64)
numpy_dim = 4

# Test case 2: torch tensor input
torch_value = torch.tensor([0, 1, 2, 1], dtype=torch.long)
torch_dim = 4

# We'll use the numpy case as the primary test
ARGS = {
    "value": numpy_value,
    "dim": numpy_dim
}
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 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
to_transition Ran
ran; our contract draft was wrong, not the code
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("9b6789716d0eab8d")
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 Tensor, ndim 2, dtype family float, shape (1, 15) 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)

class MockAgent:
    def __init__(self):
        self.device = 'cpu'
        self.env_n_action = 3

class MockArgs:
    def __init__(self):
        self.n_agent = 2

# Create mock objects
agent = MockAgent()
args = MockArgs()

# Create sample transition data
obs = np.array([[1.0, 2.0, 3.0, 4.0]], dtype=np.float32)
actions = [[0, 1]]  # 2 agents, each with an action index
reward = np.array([1.5], dtype=np.float32)
next_obs = np.array([[2.0, 3.0, 4.0, 5.0]], dtype=np.float32)

ARGS = {
    'obs': obs,
    'actions': actions,
    'reward': reward,
    'next_obs': next_obs,
    'agent': agent,
    'args': args
}
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 Tensor, ndim 2, dtype family float, shape (1, 15) — shape (1, 15) 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
Base Not yet run
unverified (harvested, no recorded run)
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("992fa4129e010d32")
How it ran

no invocation recorded.

FURTHER Not yet run
unverified (harvested, no recorded run)
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("662aff050d002949")
How it ran

no invocation recorded.

weight_init Not yet run
unverified (harvested, no recorded run)
dkkim93/further/algorithm/further/agent.py
code served (permissive licence) · get_code("cb1c431c7a635b3f")
How it ran

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, 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 13 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

The main challenge of multiagent reinforcement learning is the difficulty of learning useful policies in the presence of other simultaneously learning agents whose changing behaviors jointly affect the environment's transition and reward dynamics. An effective approach that has recently emerged for addressing this non-stationarity is for each agent to anticipate the learning of other agents and influence the evolution of future policies towards desirable behavior for its own benefit. Unfortunately, previous approaches for achieving this suffer from myopic evaluation, considering only a finite number of policy updates. As such, these methods can only influence transient future policies rather than achieving the promise of scalable equilibrium selection approaches that influence the behavior at convergence. In this paper, we propose a principled framework for considering the limiting policies of other agents as time approaches infinity. Specifically, we develop a new optimization objective that maximizes each agent's average reward by directly accounting for the impact of its behavior on the limiting set of policies that other agents will converge to. Our paper characterizes desirable solution concepts within this problem setting and provides practical approaches for optimizing over possible outcomes. As a result of our farsighted objective, we demonstrate better long-term performance than state-of-the-art baselines across a suite of diverse multiagent benchmark domains.

For agents

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

get_harvested_code_for_paper("2203.03535")
get_code_for_paper("2203.03535")
have("2203.03535")

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

curl https://syntology.ai/api/ran/2203.03535.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.03535.svg)](https://syntology.ai/paper/2203.03535)

Connect an agent — have() is free.