SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2410.08540 · NeurIPS · 2024

Kaleidoscope: Learnable Masks for Heterogeneous Multi-agent Reinforcement Learning

Ling Pan, Jun Zhang, Xinran Li

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 (2 of them constructed an object rather than computing a result; 3 honoured the contract we drafted, 0 violated it and 2 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/2410.08540.json · badge: /api/ran/2410.08540.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
LXXXXR/Kaleidoscope canonical (classifier label)
classifier basis: census
6 ran · 0 no recorded run
lxxxxr/kaleidoscope canonical (classifier label)
classifier basis: classifier_2025
2 ran · 1 no recorded run
uoe-agents/epymarl extension (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
KaleiLinear Ran
ran on a synthesized input · constructed an object; no result was computed
lxxxxr/kaleidoscope/Kalei_MPE/src/modules/agents/rnn_agent.py
code served (permissive licence) · get_code("3e4047b55fdd8d46")
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, KaleiLinear: 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
import math

torch.manual_seed(0)

# Define the KaleiLinear class (copy from the function definition)
class compare_STE(torch.autograd.Function):
    @staticmethod
    def forward(ctx, input):
        return torch.nn.functional.relu(torch.sign(input))

    @staticmethod
    def backward(ctx, grad_output):
        return torch.tanh_(grad_output)

class KaleiLinear(nn.Linear):
    def __init__(self, n_masks, threshold_init_scale, threshold_init_bias, threshold_reset_scale, threshold_reset_bias, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.n_masks = n_masks
        self.threshold_init_scale = threshold_init_scale
        self.threshold_init_bias = threshold_init_bias
        self.threshold_reset_scale = threshold_reset_scale
        self.threshold_reset_bias = threshold_reset_bias
        init_sacles = torch.rand_like(self.weight[None, ...].repeat(n_masks, 1, 1))
        init_sacles = -(init_sacles * threshold_init_scale + threshold_init_bias)
        self.sparse_thresholds = nn.Parameter(init_sacles)
        self.activation = torch.relu
        self.f = torch.sigmoid

    def _sparse_function(self, w, mask_ids):
        all_mask_ids = torch.arange(self.n_masks).to(self.weight.device)
        thresholds = self.sparse_thresholds[all_mask_ids]
        t = self.f(thresholds)
        w = torch.sign(w[None, ...]) * self.activation(torch.abs(w[None, ...]) - t)
        return w[mask_ids]

    def _get_masks(self, w, mask_ids):
        all_mask_ids = torch.arange(self.n_masks).to(self.weight.device)
        thresholds = self.sparse_thresholds[all_mask_ids]
        w = compare_STE.apply(torch.abs(w[None, ...].detach()) - self.f(thresholds))
        return w[mask_ids]

    def _get_weighted_masks(self, w, mask_ids):
        masks = self._get_masks(w, mask_ids)
        return masks * torch.abs(w[None, ...].detach())

    def forward(self, x, mask_ids):
        assert len(x.shape) == 2, f'Invalid input shape {x.shape}.'
        w = self._sparse_function(self.weight, mask_ids)
        result = torch.matmul(w, x[:, :, None]).squeeze(-1) + self.bias[None, :]
        return result

    def get_sparsities(self):
        mask_ids = torch.arange(self.n_masks).to(self.weight.device)
        w = self._sparse_function(self.weight, mask_ids)
        temp = w.detach()
        temp[temp != 0] = 1
        temp = temp.reshape(self.n_masks, -1)
        return (1.0 - temp.mean(dim=-1), temp[0].numel())

    def _reset(self, reset_ratio):
        with torch.no_grad():
            reset_flag = torch.rand_like(self.weight)
            reset_flag = reset_flag < reset_ratio
            all_mask_ids = torch.arange(self.n_masks).to(self.weight.device)
            thresholds = self.sparse_thresholds[all_mask_ids]
            m = compare_STE.apply(torch.abs(self.weight[None, ...].detach()) - self.f(thresholds))
            m = m.sum(dim=0) == 0
            reset_flag = torch.logical_and(reset_flag, m).float()
            init_sacles = torch.rand_like(self.weight[None, ...].repeat(self.n_masks, 1, 1))
            init_sacles = -(init_sacles * self.threshold_reset_scale + self.threshold_reset_bias)
            new_sparse_thresholds = reset_flag[None, ...] * init_sacles + (1 - reset_flag[None, ...]) * self.sparse_thresholds
            self.sparse_thresholds.copy_(new_sparse_thresholds)
            new_weights = torch.zeros_like(self.weight)
            nn.init.kaiming_uniform_(new_weights, a=math.sqrt(5))
            new_weights = reset_flag * new_weights + (1 - reset_flag) * self.weight
            self.weight.copy_(new_weights)

# Construct arguments
n_masks = 3
threshold_init_scale = 0.1
threshold_init_bias = 0.05
threshold_reset_scale = 0.1
threshold_reset_bias = 0.05
in_features = 4
out_features = 3

ARGS = {
    'n_masks': n_masks,
    'threshold_init_scale': threshold_init_scale,
    'threshold_init_bias': threshold_init_bias,
    'threshold_reset_scale': threshold_reset_scale,
    'threshold_reset_bias': threshold_reset_bias,
    'in_features': in_features,
    'out_features': out_features,
}
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 KaleiLinear — 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 KaleiLinear 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
compare_STE Ran
ran on a synthesized input · constructed an object; no result was computed
lxxxxr/kaleidoscope/Kalei_MPE/src/modules/agents/rnn_agent.py
code served (permissive licence) · get_code("d10c169fc33ec23c")
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, compare_STE: 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 as th
import torch.nn.functional as F

th.manual_seed(0)

# The compare_STE.forward method takes an input tensor and applies F.relu(th.sign(input))
# The compare_STE.backward method takes grad_output and applies th.tanh_(grad_output)
# So we need to provide a tensor as the 'input' parameter

input_tensor = th.randn(2, 4)

ARGS = {
    'input': input_tensor
}
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 compare_STE — 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 compare_STE 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
extract_alg_name_from_config Ran
ran; our contract draft was wrong, not the code
uoe-agents/epymarl/plot_results.py
code served (permissive licence) · get_code("868161eada09b424")
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) 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 torch
import numpy as np
from collections import defaultdict

# The function extract_alg_name_from_config takes a config parameter
# and accesses config['name'] to return it.
# We need to provide a config dict with a 'name' key.

config = {
    'name': 'test_algorithm',
    'other_param': 'some_value'
}

ARGS = {
    'config': config
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
matplotlib, numpy, seaborn, torch — 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
extract_env_name_from_config Ran
ran; our contract draft was wrong, not the code
uoe-agents/epymarl/plot_results.py
code served (permissive licence) · get_code("b23d20b93b147852")
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) 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 argparse
from collections import defaultdict
import json
from pathlib import Path
import warnings
import matplotlib.pyplot as plt
import numpy as np
import seaborn as sns

# Create a minimal config dict that satisfies the function's requirements
config_with_map_name = {
    'env': 'StarCraft',
    'env_args': {
        'map_name': '3m'
    }
}

config_with_key = {
    'env': 'Atari',
    'env_args': {
        'key': 'Pong-v0'
    }
}

config_with_neither = {
    'env': 'CustomEnv',
    'env_args': {}
}

# ARGS maps the parameter name to a value
ARGS = {
    'config': config_with_map_name
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
matplotlib, numpy, seaborn — 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
flatdim_box_multibinary Ran
ran, honoured the contract we drafted
LXXXXR/Kaleidoscope/Kalei_MPE/src/envs/gym_spaces_utils.py
code served (permissive licence) · get_code("35b9956a050d423d")
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.

from gym.spaces import Box, MultiBinary
import numpy as np

# Create a Box space with shape (4, 3)
box_space = Box(low=0, high=1, shape=(4, 3), dtype=np.float32)

# Create a MultiBinary space with shape (2, 5)
multibinary_space = MultiBinary((2, 5))

# Test with Box space
ARGS = {'space': box_space}
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 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
flatdim_discrete Ran
ran, honoured the contract we drafted
LXXXXR/Kaleidoscope/Kalei_MPE/src/envs/gym_spaces_utils.py
code served (permissive licence) · get_code("33d389e8d257639f")
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.

from gym.spaces import Discrete

# Create a Discrete space with n=5
space = Discrete(n=5)

# Map all parameters of flatdim_discrete to values
ARGS = {
    'space': space
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
gym — 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
gumbel_softmax_sample Ran
ran; our contract draft was wrong, not the code
LXXXXR/Kaleidoscope/Kalei_MPE/src/controllers/maddpg_controller.py
code served (permissive licence) · get_code("355d7b51c7ffd042")
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 as th
from torch.autograd import Variable
import torch.nn.functional as F

th.manual_seed(0)

# Create a small logits tensor (batch_size=2, num_classes=4)
logits = th.randn(2, 4, requires_grad=True)

# Set temperature
temperature = 0.5

# Construct ARGS dict
ARGS = {
    "logits": logits,
    "temperature": temperature,
}
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
load_results Ran
ran; our contract draft was wrong, not the code
uoe-agents/epymarl/plot_results.py
code served (permissive licence) · get_code("e8a91a919b93f3e1")
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) and returned defaultdict. 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 json
import tempfile
from pathlib import Path
from collections import defaultdict

# Create a temporary directory structure with test data
temp_dir = tempfile.mkdtemp()
temp_path = Path(temp_dir)

# Create a subdirectory for results
results_dir = temp_path / "run_1"
results_dir.mkdir(parents=True, exist_ok=True)

# Create a config.json file
config = {
    "name": "QMIX",
    "env": "SMAC",
    "env_args": {
        "map_name": "3m"
    },
    "seed": 42,
    "common_reward": True,
    "reward_scalarisation": "sum"
}

config_file = results_dir / "config.json"
with open(config_file, 'w') as f:
    json.dump(config, f)

# Create a metrics.json file
metrics = {
    "episode_reward": {
        "steps": [0, 100, 200, 300],
        "values": [10.5, 20.3, 25.1, 28.7]
    },
    "win_rate": {
        "steps": [0, 100, 200, 300],
        "values": [0.0, 0.2, 0.5, 0.8]
    }
}

metrics_file = results_dir / "metrics.json"
with open(metrics_file, 'w') as f:
    json.dump(metrics, f)

# Define ARGS dictionary with all parameters for load_results function
ARGS = {
    "path": str(temp_path),
    "metric": "episode_reward"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
matplotlib, numpy, seaborn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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
onehot_from_logits Ran
ran; our contract draft was wrong, not the code
LXXXXR/Kaleidoscope/Kalei_MPE/src/controllers/maddpg_controller.py
code served (permissive licence) · get_code("046b92243aae4f76")
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

torch.manual_seed(0)

# The function expects:
# - logits: a tensor that supports .max(-1, keepdim=True) and comparison operations
# - eps: a float (default 0.0, not used in the function body)

# Create a small batch of logits
# Shape: (batch_size=2, num_actions=4)
logits = torch.randn(2, 4)

ARGS = {
    "logits": logits,
    "eps": 0.0
}
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 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
sample_gumbel Ran
ran; our contract draft was wrong, not the code
LXXXXR/Kaleidoscope/Kalei_MPE/src/controllers/maddpg_controller.py
code served (permissive licence) · get_code("3695d66ab681cacf")
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 as th
from torch.autograd import Variable

th.manual_seed(0)

# Define arguments for sample_gumbel function
shape = (2, 4)  # Small tensor shape: batch size 2, dimension 4
eps = 1e-20  # Default epsilon value
tens_type = th.FloatTensor  # Tensor type

ARGS = {
    "shape": shape,
    "eps": eps,
    "tens_type": tens_type
}
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
work Ran
ran, honoured the contract we drafted
LXXXXR/Kaleidoscope/Kalei_MPE/src/search.py
code served (permissive licence) · get_code("930812c3b06fc858")
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 subprocess

def work(cmd):
    cmd = cmd.split(' ')
    return subprocess.call(cmd, shell=False)

# The function takes a cmd string, splits it by spaces, and calls subprocess.call
# We need to provide a valid command that can be executed
# Using 'echo' with a simple message as a safe, universal command

ARGS = {
    'cmd': 'echo hello'
}
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 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
Kalei_RNNAgent_1R3 Not yet run
unverified (harvested, no recorded run)
lxxxxr/kaleidoscope/Kalei_MPE/src/modules/agents/rnn_agent.py
code served (permissive licence) · get_code("73ec7f831d9eb7bc")
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, 7 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 7 the same month, 7 later) and 5 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
4 dated earlier, 1 the same month, 1 later
2006.07869 (2020-06)
Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks
2006.09447 (2020-06)
Agent Modelling under Partial Observability for Deep Reinforcement Learning
2210.06835 (2022-10)
Multi-agent Dynamic Algorithm Configuration
and 3 more not named here
gumbel_softmax_sample 5 other papers
3 dated earlier, 1 the same month, 1 later
2006.07869 (2020-06)
Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks
2210.06835 (2022-10)
Multi-agent Dynamic Algorithm Configuration
2212.05331 (2022-12)
Effects of Spectral Normalization in Multi-agent Reinforcement Learning
and 2 more not named here
sample_gumbel 5 other papers
3 dated earlier, 1 the same month, 1 later
2006.07869 (2020-06)
Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks
2210.06835 (2022-10)
Multi-agent Dynamic Algorithm Configuration
2212.05331 (2022-12)
Effects of Spectral Normalization in Multi-agent Reinforcement Learning
and 2 more not named here
work 5 other papers
3 dated earlier, 1 the same month, 1 later
1910.07483 (2019-10)
MAVEN: Multi-Agent Variational Exploration
2006.07869 (2020-06)
Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks
2212.05331 (2022-12)
Effects of Spectral Normalization in Multi-agent Reinforcement Learning
and 2 more not named here
extract_alg_name_from_config 4 other papers
2 dated earlier, 1 the same month, 1 later
1902.04043 (2019-02)
The StarCraft Multi-Agent Challenge
2006.07869 (2020-06)
Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
and 1 more not named here
extract_env_name_from_config 4 other papers
2 dated earlier, 1 the same month, 1 later
1902.04043 (2019-02)
The StarCraft Multi-Agent Challenge
2006.07869 (2020-06)
Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
and 1 more not named here
load_results 4 other papers
2 dated earlier, 1 the same month, 1 later
1902.04043 (2019-02)
The StarCraft Multi-Agent Challenge
2006.07869 (2020-06)
Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks
2410.03997 (2024-10)
YOLO-MARL: You Only LLM Once for Multi-Agent Reinforcement Learning
and 1 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

In multi-agent reinforcement learning (MARL), parameter sharing is commonly employed to enhance sample efficiency. However, the popular approach of full parameter sharing often leads to homogeneous policies among agents, potentially limiting the performance benefits that could be derived from policy diversity. To address this critical limitation, we introduce Kaleidoscope, a novel adaptive partial parameter sharing scheme that fosters policy heterogeneity while still maintaining high sample efficiency. Specifically, Kaleidoscope maintains one set of common parameters alongside multiple sets of distinct, learnable masks for different agents, dictating the sharing of parameters. It promotes diversity among policy networks by encouraging discrepancy among these masks, without sacrificing the efficiencies of parameter sharing. This design allows Kaleidoscope to dynamically balance high sample efficiency with a broad policy representational capacity, effectively bridging the gap between full parameter sharing and non-parameter sharing across various environments. We further extend Kaleidoscope to critic ensembles in the context of actor-critic algorithms, which could help improve value estimations. Our empirical evaluations across extensive environments, including multi-agent particle environment, multi-agent MuJoCo and StarCraft multi-agent challenge v2, demonstrate the superior performance of Kaleidoscope compared with existing parameter sharing approaches, showcasing its potential for performance enhancement in MARL. The code is publicly available at https://github.com/LXXXXR/Kaleidoscope.

For agents

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

get_harvested_code_for_paper("2410.08540")
get_code_for_paper("2410.08540")
have("2410.08540")

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

curl https://syntology.ai/api/ran/2410.08540.json

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

[![Syntology run record](https://syntology.ai/api/ran/2410.08540.svg)](https://syntology.ai/paper/2410.08540)

Connect an agent — have() is free.