SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.07985 · NeurIPS · 2023

Neural Combinatorial Optimization with Heavy Decoder: Toward Large Scale Generalization

Fei Liu, Xi Lin, Qingfu Zhang, Fu Luo, Zhenkun Wang

arXiv · PDF · Open in the Atlas

Code that ran

1 of 3 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; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 0 failures of our instrument; 2 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/2310.07985.json · badge: /api/ran/2310.07985.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
ciam-group/nco_code canonical (classifier label)
classifier basis: classifier_2025
1 ran · 2 no recorded run
FunctionStatusWhere it lives
DecoderLayer Ran
ran on a synthesized input
ciam-group/nco_code/single_objective/LEHD/CVRP/VRPModel.py
code served (permissive licence) · get_code("c559544d2b05dfa9")
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 torch.nn as nn
import torch.nn.functional as F

class Feed_Forward_Module(nn.Module):

    def __init__(self, **model_params):
        super().__init__()
        embedding_dim = model_params['embedding_dim']
        ff_hidden_dim = model_params['ff_hidden_dim']
        self.W1 = nn.Linear(embedding_dim, ff_hidden_dim)
        self.W2 = nn.Linear(ff_hidden_dim, embedding_dim)

    def forward(self, input1):
        return self.W2(F.relu(self.W1(input1)))

def multi_head_attention(q, k, v):
    batch_s = q.size(0)
    head_num = q.size(1)
    n = q.size(2)
    key_dim = q.size(3)
    score = torch.matmul(q, k.transpose(2, 3))
    score_scaled = score / torch.sqrt(torch.tensor(key_dim, dtype=torch.float))
    weights = nn.Softmax(dim=3)(score_scaled)
    out = torch.matmul(weights, v)
    out_transposed = out.transpose(1, 2)
    out_concat = out_transposed.reshape(batch_s, n, head_num * key_dim)
    return out_concat

def reshape_by_heads(qkv, head_num):
    batch_s = qkv.size(0)
    n = qkv.size(1)
    q_reshaped = qkv.reshape(batch_s, n, head_num, -1)
    q_transposed = q_reshaped.transpose(1, 2)
    return q_transposed

class DecoderLayer(nn.Module):

    def __init__(self, **model_params):
        super().__init__()
        self.model_params = model_params
        embedding_dim = self.model_params['embedding_dim']
        head_num = self.model_params['head_num']
        qkv_dim = self.model_params['qkv_dim']
        self.Wq = nn.Linear(embedding_dim, head_num * qkv_dim, bias=False)
        self.Wk = nn.Linear(embedding_dim, head_num * qkv_dim, bias=False)
        self.Wv = nn.Linear(embedding_dim, head_num * qkv_dim, bias=False)
        self.multi_head_combine = nn.Linear(head_num * qkv_dim, embedding_dim)
        self.feedForward = Feed_Forward_Module(**model_params)

    def forward(self, input1):
        head_num = self.model_params['head_num']
        q = reshape_by_heads(self.Wq(input1), head_num=head_num)
        k = reshape_by_heads(self.Wk(input1), head_num=head_num)
        v = reshape_by_heads(self.Wv(input1), head_num=head_num)
        out_concat = multi_head_attention(q, k, v)
        multi_head_out = self.multi_head_combine(out_concat)
        out1 = input1 + multi_head_out
        out2 = self.feedForward(out1)
        out3 = out1 + out2
        return out3

INIT_ARGS = {
    'embedding_dim': 8,
    'head_num': 2,
    'qkv_dim': 4,
    'ff_hidden_dim': 16,
}

ARGS = {
    'input1': torch.randn(2, 3, 8),
}

if __name__ == '__main__':
    decoder_layer = DecoderLayer(**INIT_ARGS)
    output = decoder_layer(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: {ARGS['input1'].shape}")
    assert output.shape == ARGS['input1'].shape, "Output shape mismatch"
    print("Success!")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 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)
CVRP_Decoder Not yet run
unverified (harvested, no recorded run)
ciam-group/nco_code/single_objective/LEHD/CVRP/VRPModel.py
code served (permissive licence) · get_code("d8afc21ac6216353")
How it ran

no invocation recorded.

VRPModel Not yet run
unverified (harvested, no recorded run)
ciam-group/nco_code/single_objective/LEHD/CVRP/VRPModel.py
code served (permissive licence) · get_code("9c8f6c89b8e8caf7")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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

Neural combinatorial optimization (NCO) is a promising learning-based approach for solving challenging combinatorial optimization problems without specialized algorithm design by experts. However, most constructive NCO methods cannot solve problems with large-scale instance sizes, which significantly diminishes their usefulness for real-world applications. In this work, we propose a novel Light Encoder and Heavy Decoder (LEHD) model with a strong generalization ability to address this critical issue. The LEHD model can learn to dynamically capture the relationships between all available nodes of varying sizes, which is beneficial for model generalization to problems of various scales. Moreover, we develop a data-efficient training scheme and a flexible solution construction mechanism for the proposed LEHD model. By training on small-scale problem instances, the LEHD model can generate nearly optimal solutions for the Travelling Salesman Problem (TSP) and the Capacitated Vehicle Routing Problem (CVRP) with up to 1000 nodes, and also generalizes well to solve real-world TSPLib and CVRPLib problems. These results confirm our proposed LEHD model can significantly improve the state-of-the-art performance for constructive NCO.

For agents

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

get_harvested_code_for_paper("2310.07985")
get_code_for_paper("2310.07985")
have("2310.07985")

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

curl https://syntology.ai/api/ran/2310.07985.json

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

[![Syntology run record](https://syntology.ai/api/ran/2310.07985.svg)](https://syntology.ai/paper/2310.07985)

Connect an agent — have() is free.