SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2010.04683 · 2020

Smooth Variational Graph Embeddings for Efficient Neural Architecture Search

arXiv · PDF · Open in the Atlas

Code that ran

5 of 6 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; 2 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 0 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/2010.04683.json · badge: /api/ran/2010.04683.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
jovitalukasik/SVGe canonical (classifier label)
classifier basis: census
3 ran · 1 no recorded run
automl/SVGe canonical (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
FunctionStatusWhere it lives
batch2graph Ran
ran on a synthesized input
jovitalukasik/SVGe/models/SVGe.py
code served (permissive licence) · get_code("fbdbd8d75bfe83f2")
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 list. 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

torch.manual_seed(0)

# The function batch2graph expects:
# - graphs: a tensor that will be converted to numpy and iterated over
# - stop: an integer (default 10)
# - backward: a boolean (default False)

# Create a small tensor for graphs
# The function expects graphs to be iterable as numpy array where each element
# is indexable like graph[idx] to get scalars and graph[idx:idx+run] to get slices
# Looking at the logic: graph[0] is node_att (a scalar), then graph[1:idx+run] are edges
# Let's create a small graph tensor with enough elements

# Minimal graph structure:
# graph[0] = first node attribute
# graph[1] = edge list start
# graph[2:4] = more edges for run=3
# graph[4:7] = more edges for run=4
# etc.

# Create a batch of 2 small graphs
graph1 = torch.tensor([1.0, 1, 0, 1, 0, 1, 0, 1, 0], dtype=torch.float32)
graph2 = torch.tensor([0.0, 1, 1, 0, 1, 0, 1, 0, 1], dtype=torch.float32)

graphs = torch.stack([graph1, graph2], dim=0)

ARGS = {
    "graphs": graphs,
    "stop": 10,
    "backward": False
}
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 list — a list output has no array shape of its own; none is 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
compute_kernel_numpy Ran
ran on a synthesized input
jovitalukasik/SVGe/Bayesian_Optimization/gauss.py
code served (permissive licence) · get_code("6c74edc2c0dc695e")
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, 3) 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

# Analyze the function:
# - lls: log length scale(s), used as np.exp(lls), then in np.outer with ls
#   appears to be 1D array of shape (n_features,)
# - lsf: log signal frequency, scalar or array, used as np.exp(lsf)
# - x: data array, can be 1D or 2D, shape (n_samples, n_features) after processing
# - z: data array, can be 1D or 2D, shape (m_samples, n_features) after processing

# The function computes a kernel matrix between x and z
# Let's create small distinct dimensions:
# - n_features = 4 (for lls dimension)
# - n_samples_x = 2 (for x)
# - n_samples_z = 3 (for z, different from x to avoid ambiguity)

np.random.seed(0)

# lls: log length scale, one per feature, shape (4,)
lls = np.array([-0.5, 0.0, 0.2, -0.3])

# lsf: log signal frequency, scalar
lsf = np.array(0.5)

# x: input data, shape (2, 4)
x = np.random.randn(2, 4)

# z: reference data, shape (3, 4)
z = np.random.randn(3, 4)

ARGS = {
    "lls": lls,
    "lsf": lsf,
    "x": x,
    "z": z
}
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 (2, 3) — shape (2, 3) 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
edges2index Ran
ran on a synthesized input
jovitalukasik/SVGe/models/SVGe.py
code served (permissive licence) · get_code("4d5aa0fb9e228553")
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 int, shape (2, 9) 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)

# The function expects:
# - edges: a 2D tensor of shape (batch_size, size) containing 0s and 1s
# - finish: a boolean flag

# Looking at the function logic:
# - It processes edges as a batch of edge encodings
# - size must be a triangular number to produce valid num_nodes
# - For a triangular number: size = n*(n+1)/2, so num_nodes = n+1
# - Let's use size = 6 (which gives n=3, so num_nodes=4)
# - batch_size = 2 (small batch)

batch_size = 2
size = 6  # triangular number: 3*4/2 = 6, gives num_nodes = 4

# Create a small tensor with edges (0s and 1s)
edges = torch.randint(0, 2, (batch_size, size), dtype=torch.float32)

ARGS = {
    "edges": edges,
    "finish": False
}
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 int, shape (2, 9) — shape (2, 9) 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
generate_adj Ran
ran, honoured the contract we drafted
automl/SVGe/Extrapolation_Ability/eval_extrapolation.py
code served (permissive licence) · get_code("b6fa639365ad18db")
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 int, 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

class MockGraph:
    def __init__(self, edge_index, num_nodes):
        self.edge_index = edge_index
        self.num_nodes = num_nodes

torch.manual_seed(0)

# Create a simple graph with 4 nodes and 5 edges
edge_index = torch.tensor([
    [0, 1, 2, 3, 1],  # source nodes
    [1, 2, 3, 0, 3]   # target nodes
], dtype=torch.long)

pth_graph = MockGraph(
    edge_index=edge_index,
    num_nodes=4
)

ARGS = {
    "pth_graph": pth_graph
}
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 int, 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
parse_graph_to_nx Ran
ran, honoured the contract we drafted
automl/SVGe/Extrapolation_Ability/eval_extrapolation.py
code served (permissive licence) · get_code("7e6eb552e178473b")
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 DiGraph. 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 numpy as np
import torch

class MockGraph:
    def __init__(self):
        # edge_index: shape (2, num_edges) - pairs of node indices
        self.edge_index = torch.tensor([[0, 1, 1, 2], [1, 0, 2, 1]], dtype=torch.long)
        # node_atts: shape (num_nodes,) - node attributes
        self.node_atts = np.array([2, 3, 4, 5], dtype=np.int32)
        # acc: shape (num_nodes,) - accuracy values
        self.acc = np.array([0.8, 0.9, 0.85, 0.95], dtype=np.float32)

graph_instance = MockGraph()

ARGS = {
    "graph_name": graph_instance
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
networkx, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DiGraph — a DiGraph output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DiGraph output the recorded ndim 1 and dtype family int 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_highest_over_time Not yet run
unverified (harvested, no recorded run)
jovitalukasik/SVGe/Bayesian_Optimization/summarize.py
code served (permissive licence) · get_code("b877cbe6c942ca45")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 0 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
compute_kernel_numpy 2 other papers
2 dated earlier
1904.11088 (2019-04)
D-VAE: A Variational Autoencoder for Directed Acyclic Graphs
1910.00698 (2019-10)
Re-balancing Variational Autoencoder Loss for Molecule Sequence Generation
get_highest_over_time 1 other paper
1 dated earlier
1904.11088 (2019-04)
D-VAE: A Variational Autoencoder for Directed Acyclic Graphs

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 architecture search (NAS) has recently been addressed from various directions, including discrete, sampling-based methods and efficient differentiable approaches. While the former are notoriously expensive, the latter suffer from imposing strong constraints on the search space. Architecture optimization from a learned embedding space for example through graph neural network based variational autoencoders builds a middle ground and leverages advantages from both sides. Such approaches have recently shown good performance on several benchmarks. Yet, their stability and predictive power heavily depends on their capacity to reconstruct networks from the embedding space. In this paper, we propose a two-sided variational graph autoencoder, which allows to smoothly encode and accurately reconstruct neural architectures from various search spaces. We evaluate the proposed approach on neural architectures defined by the ENAS approach, the NAS-Bench-101 and the NAS-Bench-201 search space and show that our smooth embedding space allows to directly extrapolate the performance prediction to architectures outside the seen domain (e.g. with more operations). Thus, it facilitates to predict good network architectures even without expensive Bayesian optimization or reinforcement learning.

For agents

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

get_harvested_code_for_paper("2010.04683")
get_code_for_paper("2010.04683")
have("2010.04683")

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

curl https://syntology.ai/api/ran/2010.04683.json

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

[![Syntology run record](https://syntology.ai/api/ran/2010.04683.svg)](https://syntology.ai/paper/2010.04683)

Connect an agent — have() is free.