SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2603.24366 · 2026

CoordLight: Learning Decentralized Coordination for Network-Wide Traffic Signal Control

Sandeep Chinchali, Guillaume Sartoretti, Yifeng Zhang, Harsh Goel, Mehul Damani, Peizhuo Li

arXiv · PDF · Open in the Atlas

Code that ran

7 of 10 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; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 3 failures of our instrument; 3 no recorded run), measured 2026-11-22; 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/2603.24366.json · badge: /api/ran/2603.24366.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
marmotlab/CoordLight community
Papers with Code archive
7 ran · 3 no recorded run
FunctionStatusWhere it lives
Decoder Ran
ran on a synthesized input
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("bf6237a8d8d92117")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, 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 torch
import numpy as np

# INIT_ARGS for Decoder.__init__(embedding_dim, n_head, n_layer=1)
INIT_ARGS = {
    'embedding_dim': 8,
    'n_head': 2,
    'n_layer': 2
}

# ARGS for Decoder.forward(tgt, memory, mask=None)
batch_size = 2
n_query = 3
graph_size = 5
embedding_dim = 8

ARGS = {
    'tgt': torch.randn(batch_size, n_query, embedding_dim),
    'memory': torch.randn(batch_size, graph_size, embedding_dim),
    'mask': torch.zeros(batch_size, n_query, graph_size)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 3, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
DecoderLayer Ran
ran on a synthesized input
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("30303624bb3dec79")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, 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 torch
import numpy as np

# INIT_ARGS for DecoderLayer.__init__(embedding_dim, n_head)
INIT_ARGS = {
    'embedding_dim': 8,
    'n_head': 2
}

# ARGS for DecoderLayer.forward(tgt, memory, mask=None)
batch_size = 2
n_query = 3
graph_size = 5
embedding_dim = 8

tgt = torch.randn(batch_size, n_query, embedding_dim)
memory = torch.randn(batch_size, graph_size, embedding_dim)
mask = torch.zeros(batch_size, n_query, graph_size)
mask[0, 0, 0] = 1  # Add some masking

ARGS = {
    'tgt': tgt,
    'memory': memory,
    'mask': mask
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 3, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
MultiHeadAttention Ran
ran on a synthesized input · constructed an object; no result was computed
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("a76fb898ce3994a8")
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, MultiHeadAttention: 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 numpy as np
import math

torch.manual_seed(0)

class MultiHeadAttention(nn.Module):
    """
    Multi-head attention module from https://arxiv.org/abs/1706.03762
    """

    def __init__(self, embedding_dim, n_heads):
        super(MultiHeadAttention, self).__init__()
        self.n_heads = n_heads
        self.input_dim = embedding_dim
        self.embedding_dim = embedding_dim
        self.value_dim = self.embedding_dim // self.n_heads
        self.key_dim = self.value_dim
        self.norm_factor = 1 / math.sqrt(self.key_dim)
        self.w_query = nn.Parameter(torch.Tensor(self.n_heads, self.input_dim, self.key_dim))
        self.w_key = nn.Parameter(torch.Tensor(self.n_heads, self.input_dim, self.key_dim))
        self.w_value = nn.Parameter(torch.Tensor(self.n_heads, self.input_dim, self.value_dim))
        self.w_out = nn.Parameter(torch.Tensor(self.n_heads, self.value_dim, self.embedding_dim))
        self.init_parameters()

    def init_parameters(self):
        for param in self.parameters():
            stdv = 1.0 / math.sqrt(param.size(-1))
            param.data.uniform_(-stdv, stdv)

    def forward(self, q, h=None, mask=None):
        """
        :param q: queries (batch_size, n_query, input_dim)
        :param h: data (batch_size, graph_size, input_dim)
        :param mask: mask (batch_size, n_query, graph_size) or viewable as that (i.e. can be 2 dim if n_query == 1)
        Mask should contain 1 if attention is not possible (i.e. mask is negative adjacency)
        :return:
        """
        if h is None:
            h = q
        batch_size, target_size, input_dim = h.size()
        n_query = q.size(1)
        h_flat = h.contiguous().view(-1, input_dim)
        q_flat = q.contiguous().view(-1, input_dim)
        shape_v = (self.n_heads, batch_size, target_size, -1)
        shape_k = (self.n_heads, batch_size, target_size, -1)
        shape_q = (self.n_heads, batch_size, n_query, -1)
        Q = torch.matmul(q_flat, self.w_query).view(shape_q)
        K = torch.matmul(h_flat, self.w_key).view(shape_k)
        V = torch.matmul(h_flat, self.w_value).view(shape_v)
        U = torch.tensor(self.norm_factor) * torch.matmul(Q, K.transpose(2, 3))
        if mask is not None:
            mask = mask.view(1, batch_size, -1, target_size).expand_as(U)
            U[mask.bool()] = -np.inf
        attention = torch.softmax(U, dim=-1)
        if mask is not None:
            attnc = attention.clone()
            attnc[mask.bool()] = 0
            attention = attnc
        heads = torch.matmul(attention, V)
        out = torch.mm(heads.permute(1, 2, 0, 3).reshape(-1, self.n_heads * self.value_dim), self.w_out.view(-1, self.embedding_dim)).view(batch_size, n_query, self.embedding_dim)
        return out


embedding_dim = 4
n_heads = 2
batch_size = 2
graph_size = 3
n_query = 2

model = MultiHeadAttention(embedding_dim=embedding_dim, n_heads=n_heads)

q = torch.randn(batch_size, n_query, embedding_dim)
h = torch.randn(batch_size, graph_size, embedding_dim)
mask = torch.zeros(batch_size, n_query, graph_size)

ARGS = {
    "embedding_dim": embedding_dim,
    "n_heads": n_heads,
}
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 MultiHeadAttention — 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 MultiHeadAttention 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
Normalization Ran
ran on a synthesized input · constructed an object; no result was computed
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("0c6fcba4acde220a")
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, Normalization: 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

torch.manual_seed(0)

# The Normalization class takes embedding_dim as its only parameter
# It uses this to create a LayerNorm layer
# The forward method reshapes input to (-1, x.size(-1)) and applies normalization

embedding_dim = 4

ARGS = {
    "embedding_dim": embedding_dim,
}
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 Normalization — 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 Normalization 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
init Ran
ran; our contract draft was wrong, not the code
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("c31c51db289b4c16")
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 SimpleModule. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleModule(nn.Module):
    def __init__(self):
        super().__init__()
        self.weight = nn.Parameter(torch.randn(4, 3))
        self.bias = nn.Parameter(torch.randn(4))

module = SimpleModule()

def weight_init_fn(weight_data, gain=1):
    nn.init.orthogonal_(weight_data, gain=gain)

def bias_init_fn(bias_data):
    nn.init.zeros_(bias_data)

ARGS = {
    "module": module,
    "weight_init": weight_init_fn,
    "bias_init": bias_init_fn,
    "gain": 1
}
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 SimpleModule — a SimpleModule output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleModule 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
init_ Ran
ran; our contract draft was wrong, not the code
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("af4e55fde2363077")
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 Linear. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a small linear module with weight and bias attributes
# The function init_ expects a module with .weight.data and .bias.data
m = nn.Linear(in_features=4, out_features=3)

ARGS = {
    "m": m
}
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 Linear — a Linear output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Linear 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
init_rnn_ Ran
ran; our contract draft was wrong, not the code
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("b6e207c5dad4e9df")
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 RNN. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a simple RNN module to initialize
rnn = nn.RNN(input_size=4, hidden_size=3, num_layers=1, batch_first=True)

# Build the ARGS dict with the single required parameter
ARGS = {
    'rnn': rnn
}
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 RNN — a RNN output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a RNN 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
ActorNetwork Not yet run
unverified (harvested, no recorded run)
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("cd770c0ea46e5ffc")
How it ran

no invocation recorded.

CoordLight Not yet run
unverified (harvested, no recorded run)
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("488e1c25ef189c64")
How it ran

no invocation recorded.

CriticNetwork Not yet run
unverified (harvested, no recorded run)
marmotlab/CoordLight/Models/CoordLightModel.py
code served (permissive licence) · get_code("bda7507c3cc10545")
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, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 9 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
init
brick of this paper's assembly: helper
34 other papers
34 dated earlier · 5 only as an assembly brick
1705.05363 (2017-05)
Curiosity-driven Exploration by Self-supervised Prediction
1706.02275 (2017-06)
Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments
1810.12894 (2018-10)
Exploration by Random Network Distillation
and 31 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

Adaptive traffic signal control (ATSC) is crucial in alleviating congestion, maximizing throughput and promoting sustainable mobility in ever-expanding cities. Multi-Agent Reinforcement Learning (MARL) has recently shown significant potential in addressing complex traffic dynamics, but the intricacies of partial observability and coordination in decentralized environments still remain key challenges in formulating scalable and efficient control strategies. To address these challenges, we present CoordLight, a MARL-based framework designed to improve intra-neighborhood traffic by enhancing decisionmaking at individual junctions (agents), as well as coordination with neighboring agents, thereby scaling up to network-level traffic optimization. Specifically, we introduce the Queue Dynamic State Encoding (QDSE), a novel state representation based on vehicle queuing models, which strengthens the agents' capability to analyze, predict, and respond to local traffic dynamics. We further propose an advanced MARL algorithm, named Neighboraware Policy Optimization (NAPO). It integrates an attention mechanism that discerns the state and action dependencies among adjacent agents, aiming to facilitate more coordinated decision-making, and to improve policy learning updates through robust advantage calculation. This enables agents to identify and prioritize crucial interactions with influential neighbors, thus enhancing the targeted coordination and collaboration among agents. Through comprehensive evaluations against state-of-theart traffic signal control methods over three real-world traffic datasets composed of up to 196 intersections, we empirically show that CoordLight consistently exhibits superior performance across diverse traffic networks with varying traffic flows. The code is available at https://github.com/marmotlab/CoordLight.

For agents

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

get_harvested_code_for_paper("2603.24366")
get_code_for_paper("2603.24366")
have("2603.24366")

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

curl https://syntology.ai/api/ran/2603.24366.json

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

[![Syntology run record](https://syntology.ai/api/ran/2603.24366.svg)](https://syntology.ai/paper/2603.24366)

Connect an agent — have() is free.