SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.16168 · ACL · 2024

Progressively Modality Freezing for Multi-Modal Entity Alignment

Junfan Chen, Richong Zhang, Jaein Kim, Xuefeng Zhang, Yani Huang

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 9 have no contract verdict; 1 failure 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/2407.16168.json · badge: /api/ran/2407.16168.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
ninibymilk/PMF-MMEA community
Papers with Code archive
10 ran · 2 no recorded run
FunctionStatusWhere it lives
CKG_Loss Ran
ran on a synthesized input
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("bcceb1a21bf0d6de")
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 0, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
from scipy.sparse import csr_matrix

INIT_ARGS = {
    'tau': 0.05,
    'ab_weight': 0.5,
    'n_view': 2,
    'intra_weight': 1.0
}

# Create small tensors for the forward pass
batch_size = 2
feature_dim = 4
num_nodes = 6  # Must be >= batch_size * 2 for train_links

# Embedding tensor: (num_nodes, feature_dim)
emb = torch.randn(num_nodes, feature_dim)

# Train links: (batch_size, 2) - pairs of node indices
train_links = torch.tensor([[0, 3], [1, 4]], dtype=torch.long)

# Optional mask: (num_nodes,) - binary mask for nodes
mask = torch.tensor([1, 1, 0, 1, 1, 0], dtype=torch.float32)

ARGS = {
    'emb': emb,
    'train_links': train_links,
    'mask': mask
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, 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)
CM_Loss Ran
ran on a synthesized input
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("5e980ee90e9e3dbd")
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 0, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 numpy as np

# INIT_ARGS for CM_Loss.__init__
INIT_ARGS = {
    'tau': 0.05,
    'ab_weight': 0.5,
    'weight': 1.0
}

# ARGS for CM_Loss.forward
# Create embeddings for two modalities
batch_size = 6  # Total entities in each modality
embedding_dim = 8
num_train_links = 4  # Number of training links

# Embeddings from modality 1 and modality 2
m1_emb = torch.randn(batch_size, embedding_dim)
m2_emb = torch.randn(batch_size, embedding_dim)

# Train links: pairs of indices linking entities between modalities
# Shape: (num_train_links, 2) where each row is [idx_in_m1, idx_in_m2]
train_links = torch.tensor([
    [0, 1],
    [2, 3],
    [1, 4],
    [3, 5]
], dtype=torch.int64)

# Optional loss masks (can be None or tensors of shape [batch_size])
# These indicate which entities should contribute to the loss
loss_mask_l = torch.tensor([1.0, 1.0, 1.0, 1.0, 0.0, 0.0])
loss_mask_r = torch.tensor([0.0, 1.0, 0.0, 1.0, 1.0, 1.0])

ARGS = {
    'm1_emb': m1_emb,
    'm2_emb': m2_emb,
    'train_links': train_links,
    'loss_mask_l': loss_mask_l,
    'loss_mask_r': loss_mask_r
}
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 0, 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)
CustomMultiLossLayer Ran
ran on a synthesized input · constructed an object; no result was computed
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("40f8a8faf24955d4")
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, CustomMultiLossLayer: 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 CustomMultiLossLayer.__init__ takes loss_num as parameter
# The forward method takes loss_list (list of loss values/tensors)

# Create a small list of loss values (scalars or 0-d tensors)
loss_list = [
    torch.tensor(0.5),
    torch.tensor(0.3),
    torch.tensor(0.7)
]

ARGS = {
    "loss_num": len(loss_list)
}
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 CustomMultiLossLayer — 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 CustomMultiLossLayer 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
GAT Ran
ran on a synthesized input · constructed an object; no result was computed
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("d6be02987f151a14")
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, GAT: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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)

# Create a minimal sparse adjacency matrix
# For a graph with N=2 nodes, we need edge indices and values
N = 2
edge_indices = torch.tensor([[0, 1], [1, 0]], dtype=torch.long).t()
edge_values = torch.ones(2, dtype=torch.float32)
adj = torch.sparse_coo_tensor(edge_indices, edge_values, (N, N))

# Create input features: (N, f_in) where N=2, f_in=4
x = torch.randn(2, 4)

# Define GAT parameters
n_units = [4, 8, 3]  # input_dim=4, hidden_dim=8, output_dim=3
n_heads = [2, 2]     # 2 heads for each of the 2 layers
dropout = 0.1
attn_dropout = 0.1
instance_normalization = True
diag = True

ARGS = {
    'n_units': n_units,
    'n_heads': n_heads,
    'dropout': dropout,
    'attn_dropout': attn_dropout,
    'instance_normalization': instance_normalization,
    'diag': diag,
}
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 GAT — 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 GAT 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
GCN Ran
ran on a synthesized input · constructed an object; no result was computed
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("55399663a530eb02")
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, GCN: 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 torch.nn.functional as F
import math
from torch.nn.parameter import Parameter
from torch.nn.modules.module import Module

torch.manual_seed(0)

class GraphConvolution(Module):
    """
    Simple GCN layer, similar to https://arxiv.org/abs/1609.02907
    """

    def __init__(self, in_features, out_features, bias=True):
        super(GraphConvolution, self).__init__()
        self.in_features = in_features
        self.out_features = out_features
        self.weight = Parameter(torch.FloatTensor(in_features, out_features))
        if bias:
            self.bias = Parameter(torch.FloatTensor(out_features))
        else:
            self.register_parameter('bias', None)
        self.reset_parameters()

    def reset_parameters(self):
        stdv = 1.0 / math.sqrt(self.weight.size(1))
        self.weight.data.uniform_(-stdv, stdv)
        if self.bias is not None:
            self.bias.data.uniform_(-stdv, stdv)

    def forward(self, input, adj):
        support = torch.mm(input, self.weight)
        output = torch.spmm(adj, support)
        if self.bias is not None:
            return output + self.bias
        else:
            return output

    def __repr__(self):
        return self.__class__.__name__ + ' (' + str(self.in_features) + ' -> ' + str(self.out_features) + ')'

class GCN(nn.Module):

    def __init__(self, nfeat, nhid, nout, dropout):
        super(GCN, self).__init__()
        self.gc1 = GraphConvolution(nfeat, nhid)
        self.gc2 = GraphConvolution(nhid, nout)
        self.dropout = dropout

    def forward(self, x, adj):
        x = F.relu(self.gc1(x, adj))
        x = F.dropout(x, self.dropout, training=self.training)
        x = self.gc1(x, adj)
        return x

# Create small tensors
num_nodes = 2
nfeat = 4
nhid = 3
nout = 5

# Feature matrix: (num_nodes, nfeat)
x = torch.randn(num_nodes, nfeat)

# Adjacency matrix as sparse tensor: (num_nodes, num_nodes)
indices = torch.LongTensor([[0, 1], [1, 0]])
values = torch.FloatTensor([1.0, 1.0])
adj = torch.sparse_coo_tensor(indices.t(), values, (num_nodes, num_nodes))

ARGS = {
    'nfeat': nfeat,
    'nhid': nhid,
    'nout': nout,
    'dropout': 0.5
}
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 GCN — 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 GCN 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
MultiHeadGraphAttention Ran
ran on a synthesized input · constructed an object; no result was computed
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("c23ba7846955e383")
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, MultiHeadGraphAttention: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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)

# Create a small sparse adjacency matrix
# For a graph with N=2 nodes, we need edge indices and values
N = 2
edge_indices = torch.tensor([[0, 1], [1, 0]], dtype=torch.long).t()  # 2 edges: 0->1, 1->0
edge_values = torch.tensor([1.0, 1.0], dtype=torch.float)
adj = torch.sparse_coo_tensor(edge_indices, edge_values, (N, N))

# Create input features: N nodes, f_in=3 features
input_data = torch.randn(N, 3, dtype=torch.float)

# Define parameters for MultiHeadGraphAttention
n_head = 2
f_in = 3
f_out = 4
attn_dropout = 0.5
diag = True
init = None
bias = False

ARGS = {
    'n_head': n_head,
    'f_in': f_in,
    'f_out': f_out,
    'attn_dropout': attn_dropout,
    'diag': diag,
    'init': init,
    'bias': bias,
}
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 MultiHeadGraphAttention — 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 MultiHeadGraphAttention 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
PMFusion Ran
ran on a synthesized input · constructed an object; no result was computed
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("cb74d190b5c6e785")
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, PMFusion: 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)

# Build a minimal args object with the attributes used in forward()
class Args:
    def __init__(self):
        self.with_weight = 1

args = Args()

# Parameters for PMFusion.__init__
ent_num = 10
modal_num = 2
with_weight = 1

# Build sample data for forward() call
# embs_dict: dict with keys mapping to embeddings (or None)
# Each embedding should be shape (batch_size, embedding_dim)
batch_size = 2
embedding_dim = 4

embs_dict = {
    'modal_0': torch.randn(batch_size, embedding_dim),
    'modal_1': torch.randn(batch_size, embedding_dim),
}

# input_idx: not used in forward, but required as parameter
input_idx = torch.tensor([0, 1])

# adj: not used in forward, but required as parameter
adj = torch.randn(ent_num, ent_num)

# mask: dict with keys matching embs_dict, values are 1D tensors of shape (batch_size,)
mask = {
    'modal_0': torch.ones(batch_size),
    'modal_1': torch.ones(batch_size),
}

ARGS = {
    'args': args,
    'ent_num': ent_num,
    'modal_num': modal_num,
    'with_weight': with_weight,
}
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 PMFusion — 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 PMFusion 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
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("32c82917ea1d7b9c")
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 is False
        a = torch.sparse_coo_tensor(indices, values, shape)
        ctx.save_for_backward(a, b)
        ctx.N = shape[0]
        return torch.matmul(a, b)

    @staticmethod
    def backward(ctx, grad_output):
        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 = {}

indices = torch.tensor([[0, 1, 0, 1], [0, 1, 1, 0]], dtype=torch.long)
values = torch.tensor([1.0, 2.0, 3.0, 4.0], dtype=torch.float32, requires_grad=True)
shape = (2, 2)
b = torch.randn(2, 3, requires_grad=True)

ARGS = {
    'indices': indices,
    'values': values,
    'shape': shape,
    'b': b
}
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
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("4c08e0d2ffc7bc34")
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)

# Construct arguments for SpecialSpmmFunction.forward(ctx, indices, values, shape, b)
# indices: LongTensor of shape (2, nnz) for COO format
# values: Tensor of non-zero values, shape (nnz,)
# shape: tuple for sparse tensor shape (m, n)
# b: dense tensor to multiply with sparse matrix, shape (n, batch_size)

nnz = 3  # number of non-zero elements
m, n = 4, 5  # sparse matrix dimensions
batch_size = 2  # batch dimension for b

# Create indices for sparse COO tensor
# Must be 2D: first row is row indices, second row is column indices
indices = torch.tensor(
    [[0, 1, 3],
     [0, 2, 4]],
    dtype=torch.long,
    requires_grad=False
)

# Create values for sparse tensor
values = torch.tensor([1.0, 2.0, 3.0], dtype=torch.float32, requires_grad=True)

# Shape of sparse matrix (m x n)
shape = (m, n)

# Dense tensor b with shape (n, batch_size)
b = torch.randn(n, batch_size, 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
pairwise_distances Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("70fa9d2ca12c8bc3")
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, 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 torch
import numpy as np

torch.manual_seed(0)

# Create small tensors for testing
# x: Nxd matrix where N=2 (batch size), d=4 (dimensions)
x = torch.randn(2, 4)

# y: Mxd matrix where M=3 (different batch size), d=4 (same dimensions)
y = torch.randn(3, 4)

ARGS = {
    "x": x,
    "y": y,
}
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 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
PMF Not yet run
unverified (harvested, no recorded run)
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("7bdecc4812362ce3")
How it ran

no invocation recorded.

PMF_Encoder Not yet run
unverified (harvested, no recorded run)
ninibymilk/PMF-MMEA/model/PMF.py
code served (permissive licence) · get_code("a9f2ada726003d51")
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, 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 10 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
3 dated earlier · 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
2103.16046 (2021-03)
Unsupervised Hyperbolic Representation Learning via Message Passing Auto-Encoders
pairwise_distances
brick of this paper's assembly: helper
2 other papers
2 dated earlier
2212.14454 (2022-12)
MEAformer: Multi-modal Entity Alignment Transformer for Meta Modality Hybrid
2307.16210 (2023-07)
Rethinking Uncertainly Missing and Ambiguous Visual Modality in 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

Multi-Modal Entity Alignment aims to discover identical entities across heterogeneous knowledge graphs. While recent studies have delved into fusion paradigms to represent entities holistically, the elimination of features irrelevant to alignment and modal inconsistencies is overlooked, which are caused by inherent differences in multi-modal features. To address these challenges, we propose a novel strategy of progressive modality freezing, called PMF, that focuses on alignmentrelevant features and enhances multi-modal feature fusion. Notably, our approach introduces a pioneering cross-modal association loss to foster modal consistency. Empirical evaluations across nine datasets confirm PMF's superiority, demonstrating stateof-the-art performance and the rationale for freezing modalities. Our code is available at https://github.com/ninibymilk/PMF-MMEA.

For agents

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

get_harvested_code_for_paper("2407.16168")
get_code_for_paper("2407.16168")
have("2407.16168")

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

curl https://syntology.ai/api/ran/2407.16168.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.16168.svg)](https://syntology.ai/paper/2407.16168)

Connect an agent — have() is free.