SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.16046 · CVPR · 2021

Unsupervised Hyperbolic Representation Learning via Message Passing Auto-Encoders

Jin Choi, Jiwoong Park, Hyung Chang, Junho Cho

arXiv · PDF · Open in the Atlas

Code that ran

10 of 12 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (6 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 7 have no contract verdict; 3 failures of our instrument; 2 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/2103.16046.json · badge: /api/ran/2103.16046.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
junhocho/HGCAE marked official in Papers with Code
Papers with Code archive
10 ran · 2 no recorded run
FunctionStatusWhere it lives
Encoder Ran
ran on a synthesized input · constructed an object; no result was computed
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("77e4a84f7b06b98d")
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, Encoder: 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)

# Create a minimal graph convolution layer that the encoder will use
class GraphConvLayer(nn.Module):
    def __init__(self, in_features, out_features):
        super(GraphConvLayer, self).__init__()
        self.linear = nn.Linear(in_features, out_features)
    
    def forward(self, xx):
        x, adj = xx
        # Simple graph convolution: apply linear transformation
        out = self.linear(x)
        return (out, adj)

# Create a minimal encoder subclass for testing
class TestEncoder(nn.Module):
    def __init__(self, c, use_cnn=None):
        super(TestEncoder, self).__init__()
        self.c = c
        self.encode_graph = True
        # Create two graph conv layers
        self.layers = nn.ModuleList([
            GraphConvLayer(4, 5),
            GraphConvLayer(5, 6)
        ])
    
    def encode(self, x, adj):
        self.features = []
        if self.encode_graph:
            input = (x, adj)
            xx = input
            for i in range(len(self.layers)):
                out = self.layers[i].forward(xx)
                self.features.append(out[0])
                xx = out
            output, _ = xx
        else:
            output = self.layers.forward(x)
        return output

# Set up valid arguments
c = 10  # some config or channel parameter
use_cnn = False  # optional parameter

ARGS = {
    'c': c,
    'use_cnn': use_cnn
}
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 Encoder — 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 Encoder 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
GeometricAwareHypAggAtt Ran
ran on a synthesized input · constructed an object; no result was computed
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("12913b08638cdab8")
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, GeometricAwareHypAggAtt: 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)

# Create a minimal manifold class that implements sqdist method
class MinimalManifold:
    def sqdist(self, x1, x2, c):
        # Return squared distance as a tensor
        return torch.sum((x1 - x2) ** 2, dim=1, keepdim=True)

# Create input tensors
n_nodes = 2
in_features = 4

# Create feature tensor x: shape (n_nodes, in_features)
x = torch.randn(n_nodes, in_features)

# Create sparse adjacency matrix as COO tensor
# We need edges between nodes
edge_indices = torch.tensor([[0, 1], [1, 0]], dtype=torch.long).t().contiguous()  # shape (2, 2)
edge_values = torch.ones(2, dtype=torch.float32)
adj = torch.sparse_coo_tensor(edge_indices, edge_values, (n_nodes, n_nodes))

# Create activation function
def relu_act(x):
    return torch.relu(x)

# Create the arguments dictionary
ARGS = {
    'in_features': in_features,
    'manifold': MinimalManifold(),
    'dropout': 0.1,
    'act': relu_act,
    'att_logit': torch.tanh,
    'beta': 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 GeometricAwareHypAggAtt — 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 GeometricAwareHypAggAtt 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
HypAct Ran
ran on a synthesized input · constructed an object; no result was computed
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("3f3addbba6aadb18")
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, HypAct: 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
from torch.nn.modules.module import Module

torch.manual_seed(0)

# Define a minimal manifold class that supports the required interface
class PoincareBallManifold:
    def __init__(self):
        self.name = 'PoincareBall'
    
    def activation(self, x, act, c_in, c_out):
        # Minimal implementation: apply activation and return
        return act(x)
    
    def logmap0(self, x, c):
        # Minimal implementation: return x scaled by c
        return x / (c + 1e-8)

# Define a simple activation function
def simple_activation(x):
    return torch.relu(x)

# Create the manifold instance
manifold = PoincareBallManifold()

# Create a tiny input tensor: batch_size=2, dim=4
x = torch.randn(2, 4)

# Define curvature parameters
c_in = 1.0
c_out = 0.5

# Define activation function
act = simple_activation

ARGS = {
    'manifold': manifold,
    'c_in': c_in,
    'c_out': c_out,
    'act': act
}
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 HypAct — 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 HypAct 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
HypAgg Ran
ran on a synthesized input · constructed an object; no result was computed
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("d3afe60057d3d7fa")
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, HypAgg: 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
from torch.nn.modules.module import Module

torch.manual_seed(0)

# Build a minimal manifold class that satisfies the interface used in HypAgg
class MinimalManifold(Module):
    def __init__(self):
        super(MinimalManifold, self).__init__()
    
    def logmap0(self, x, c=1):
        # Return tangent space representation (same shape as input)
        return x
    
    def expmap0(self, x, c=1):
        # Map from tangent space back to manifold (same shape as input)
        return x
    
    def proj(self, x, c=1):
        # Project onto manifold (same shape as input)
        return x
    
    def sqdist(self, x, y, c):
        # Squared distance between points
        return torch.sum((x - y) ** 2, dim=1, keepdim=True)

# Create a sparse adjacency matrix (COO format)
n_nodes = 2
edge_indices = torch.tensor([[0, 1, 1, 0], [1, 0, 1, 0]], dtype=torch.long)
edge_values = torch.ones(4, dtype=torch.float32)
adj = torch.sparse_coo_tensor(edge_indices, edge_values, (n_nodes, n_nodes))

# Create node features
in_features = 4
x = torch.randn(n_nodes, in_features)

# Create manifold
manifold = MinimalManifold()

# Build ARGS dict with all parameters for HypAgg
ARGS = {
    'manifold': manifold,
    'c': 1,
    'use_att': True,
    'in_features': in_features,
    'dropout': 0.1,
    'att_type': 'sparse_adjmask_dist',
    'att_logit': 'tanh',
    'beta': 0.0,
    'decode': False,
}
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 HypAgg — 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 HypAgg 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
HypAggAtt Ran
ran; our contract draft was wrong, not the code
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("5af2c641fe540582")
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 GeometricAwareHypAggAtt. 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 minimal manifold class that implements the required methods
class MinimalManifold:
    def sqdist(self, x1, x2, c):
        # Return squared distance between two points
        return torch.sum((x1 - x2) ** 2, dim=1, keepdim=True)

# Create test data
n_nodes = 2
in_features = 4

# Create feature matrix
x = torch.randn(n_nodes, in_features)

# Create a sparse adjacency matrix
edge_indices = torch.LongTensor([[0, 1], [1, 0]]).t()
edge_values = torch.ones(2)
adj = torch.sparse_coo_tensor(edge_indices, edge_values, (n_nodes, n_nodes))

# Create minimal manifold
manifold = MinimalManifold()

# Build arguments
ARGS = {
    "in_features": in_features,
    "manifold": manifold,
    "dropout": 0.5,
    "act": lambda x: x,
    "att_type": None,
    "att_logit": "tanh",
    "beta": 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 GeometricAwareHypAggAtt — a GeometricAwareHypAggAtt output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a GeometricAwareHypAggAtt 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
HypLinear Ran
ran on a synthesized input · constructed an object; no result was computed
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("e437debd2b06548e")
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, HypLinear: 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)

class MockManifold:
    """Mock manifold with required methods for HypLinear."""
    
    def mobius_matvec(self, weight, x, c):
        """Matrix-vector multiplication in Möbius space."""
        # Simple mock: just do standard matmul
        return torch.matmul(weight, x.t()).t()
    
    def proj(self, x, c):
        """Project onto manifold."""
        # Mock: return as-is
        return x
    
    def expmap0(self, x, c):
        """Exponential map from origin."""
        # Mock: return as-is
        return x
    
    def mobius_add(self, x, y, c):
        """Möbius addition."""
        # Mock: simple addition
        return x + y

# Create mock manifold
manifold = MockManifold()

# Create input tensor: batch_size=2, in_features=4
# Shape: (2, 4)
x = torch.randn(2, 4)

# Arguments for HypLinear
ARGS = {
    'manifold': manifold,
    'in_features': 4,
    'out_features': 3,
    'c': torch.tensor(1.0),
    'dropout': 0.5,
    'use_bias': True
}
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 HypLinear — 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 HypLinear 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
SpecialSpmm Ran
ran on a synthesized input
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("a0ffa737ca2c2186")
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 2, 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

class SpecialSpmmFunction(torch.autograd.Function):
    """Special function for only sparse region backpropataion layer."""

    @staticmethod
    def forward(ctx, indices, values, shape, b):
        assert indices.requires_grad == False
        a = torch.sparse_coo_tensor(indices, values, shape, device=b.device)
        ctx.save_for_backward(a, b)
        ctx.N = shape[0]
        return torch.matmul(a, b)

    @staticmethod
    def backward(ctx, grad_output):
        assert not torch.isnan(grad_output).any()
        a, b = ctx.saved_tensors
        grad_values = grad_b = None
        if ctx.needs_input_grad[1]:
            grad_a_dense = grad_output.matmul(b.t())
            edge_idx = a._indices()[0, :] * ctx.N + a._indices()[1, :]
            grad_values = grad_a_dense.view(-1)[edge_idx]
        if ctx.needs_input_grad[3]:
            grad_b = a.t().matmul(grad_output)
        return (None, grad_values, None, grad_b)

class SpecialSpmm(nn.Module):

    def forward(self, indices, values, shape, b):
        return SpecialSpmmFunction.apply(indices, values, shape, b)


INIT_ARGS = {}

ARGS = {
    'indices': torch.tensor([[0, 1, 2], [1, 2, 0]], dtype=torch.long),
    'values': torch.tensor([1.0, 2.0, 3.0], dtype=torch.float32, requires_grad=True),
    'shape': (3, 4),
    'b': torch.randn(4, 2, dtype=torch.float32, requires_grad=True)
}
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 2, 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)
SpecialSpmmFunction Ran
ran on a synthesized input · constructed an object; no result was computed
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("074c8b15e5bd660d")
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, SpecialSpmmFunction: 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

torch.manual_seed(0)

# Build sparse tensor indices and values
# Create a small sparse matrix: 4x3 with a few non-zero entries
indices = torch.tensor([[0, 1, 2, 3], [0, 1, 2, 0]], dtype=torch.long)
values = torch.tensor([1.0, 2.0, 3.0, 4.0], dtype=torch.float32)
shape = (4, 3)

# Build dense tensor b that will be multiplied with sparse matrix
# Shape should be (3, 2) to match sparse matrix columns
b = torch.randn(3, 2, dtype=torch.float32, requires_grad=True)

ARGS = {
    'indices': indices,
    'values': values,
    'shape': shape,
    'b': b
}
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 SpecialSpmmFunction — 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 SpecialSpmmFunction 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_att_logit Ran
ran; our contract draft was wrong, not the code
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("eabedf90b49f2f4b")
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 builtin_function_or_method. 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

torch.manual_seed(0)

# The function get_att_logit takes two parameters:
# 1. att_logit: a string (or falsy value) that will be used with getattr(torch, att_logit)
# 2. att_type: appears to be unused in the function body, but we need to provide it

# Based on the function body:
# - att_logit is checked for truthiness
# - If truthy, it's passed to getattr(torch, att_logit) to retrieve a torch attribute
# - att_type is not used in the function, so any value works

# Valid torch attribute names that could be used:
att_logit_value = "sin"  # or any other valid torch function/attribute name
att_type_value = "some_type"  # unused but required parameter

ARGS = {
    "att_logit": att_logit_value,
    "att_type": att_type_value,
}
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 builtin_function_or_method — a builtin_function_or_method output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a builtin_function_or_method 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_dim_act_curv Ran
ran; our contract draft was wrong, not the code
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("16ac17aa0ccb2abf")
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. 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 torch.nn.functional as F

torch.manual_seed(0)

class Args:
    def __init__(self):
        self.act = 'relu'
        self.num_layers = 3
        self.feat_dim = 4
        self.hidden_dim = '8,6'
        self.dim = 5
        self.c = 1.0
        self.c_trainable = 0
        self.cuda = -1
        self.device = 'cpu'

ARGS = {'args': Args()}
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 tuple — a tuple 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
HGCAE Not yet run
unverified (harvested, no recorded run)
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("480b551018b3650b")
How it ran

no invocation recorded.

HyperbolicGraphConvolution Not yet run
unverified (harvested, no recorded run)
junhocho/HGCAE/models/encoders.py
code served (permissive licence) · get_code("09a0d5d032765b3c")
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, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 11 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
SpecialSpmm
brick of this paper's assembly: helper
3 other papers
2 dated earlier, 1 later · 1 only as an assembly brick
1710.10903 (2017-10)
Graph Attention Networks
2005.06312 (2020-05)
Reasoning with Latent Structure Refinement for Document-Level Relation Extraction
2407.16168 (2024-07)
Progressively Modality Freezing for Multi-Modal Entity Alignment

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

Most of the existing literature regarding hyperbolic embedding concentrate upon supervised learning, whereas the use of unsupervised hyperbolic embedding is less well explored. In this paper, we analyze how unsupervised tasks can benefit from learned representations in hyperbolic space. To explore how well the hierarchical structure of unlabeled data can be represented in hyperbolic spaces, we design a novel hyperbolic message passing auto-encoder whose overall auto-encoding is performed in hyperbolic space. The proposed model conducts auto-encoding the networks via fully utilizing hyperbolic geometry in message passing. Through extensive quantitative and qualitative analyses, we validate the properties and benefits of the unsupervised hyperbolic representations. Codes are available at https://github.com/junhocho/HGCAE.

For agents

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

get_harvested_code_for_paper("2103.16046")
get_code_for_paper("2103.16046")
have("2103.16046")

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

curl https://syntology.ai/api/ran/2103.16046.json

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

[![Syntology run record](https://syntology.ai/api/ran/2103.16046.svg)](https://syntology.ai/paper/2103.16046)

Connect an agent — have() is free.