SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.10637 · 2020

Temporal Graph Networks for Deep Learning on Dynamic Graphs

arXiv · PDF · Open in the Atlas

Code that ran

11 of 19 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 7 have no contract verdict; 4 failures of our instrument; 8 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/2006.10637.json · badge: /api/ran/2006.10637.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
twitter-research/tgn community (classifier label)
classifier basis not recorded on the link
5 ran · 3 no recorded run
wyd1502/DGNN reimplementation (classifier label)
classifier basis: classifier_2025
2 ran · 1 no recorded run
KimMeen/TGN reimplementation (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
jasperzhong/GNNFlow community (classifier label)
classifier basis not recorded on the link
1 ran · 3 no recorded run
petersh6/mspipe community (classifier label)
classifier basis not recorded on the link
1 ran · 1 no recorded run
FunctionStatusWhere it lives
choose_target Ran
ran; our contract draft was wrong, not the code
wyd1502/DGNN/evaluation.py
code served (permissive licence) · get_code("0cefebdbffe30631")
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) 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

torch.manual_seed(0)

class SimpleMemoryMergeModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.merge_layer = nn.Linear(8, 4)
    
    def memory_merge(self, mem1, mem2):
        # Concatenate and merge two memory tensors
        combined = torch.cat([mem1, mem2], dim=1)
        return self.merge_layer(combined)

# Create model
model = SimpleMemoryMergeModel()

# Create memory tensors
# memory_s and memory_g are tuples/lists where index [1] is accessed
# Each should have a tensor of shape (batch_size, feature_dim)
batch_size = 2
feature_dim = 4

memory_s = [torch.randn(batch_size, feature_dim), torch.randn(batch_size, feature_dim)]
memory_g = [torch.randn(batch_size, feature_dim), torch.randn(batch_size, feature_dim)]

# src_mem is a tensor of shape (src_batch_size, feature_dim)
src_mem = torch.randn(3, feature_dim)

ARGS = {
    "model": model,
    "memory_s": memory_s,
    "memory_g": memory_g,
    "src_mem": src_mem
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
compute_time_statistics Ran
ran on a synthesized input
twitter-research/tgn/utils/data_processing.py
code served (permissive licence) · get_code("6a8c69478a373575")
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, ndim 1, 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 numpy as np

# Analyze the function:
# - sources: array-like of source node IDs (integers)
# - destinations: array-like of destination node IDs (integers)
# - timestamps: array-like of timestamp values (floats/ints)
# The function iterates through indices and accesses elements by index
# It builds dictionaries keyed by source/destination IDs
# It computes time differences and statistics

# Create small but distinct test data
num_events = 5

# Source IDs: small integers (0-3)
sources = np.array([0, 1, 0, 2, 1], dtype=np.int32)

# Destination IDs: small integers (0-3), distinct from sources for clarity
destinations = np.array([2, 3, 2, 3, 0], dtype=np.int32)

# Timestamps: increasing sequence with varying gaps
timestamps = np.array([1.0, 3.5, 7.2, 10.1, 15.8], dtype=np.float32)

ARGS = {
    "sources": sources,
    "destinations": destinations,
    "timestamps": timestamps,
}
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 tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
eval_edge_prediction Ran
ran on a synthesized input
twitter-research/tgn/evaluation/evaluation.py
code served (permissive licence) · get_code("f5baa0e9f4ee3f74")
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, ndim 1, 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 math
import numpy as np
import torch
import torch.nn as nn
from torch import Tensor

torch.manual_seed(0)

class MockNegativeEdgeSampler:
    def __init__(self):
        self.seed = 42
    
    def reset_random_state(self):
        pass
    
    def sample(self, size):
        # Returns (_, negative_samples)
        negative_samples = np.random.randint(0, 10, size=size)
        return None, negative_samples

class MockData:
    def __init__(self):
        self.sources = np.array([0, 1, 2, 3, 4])
        self.destinations = np.array([1, 2, 3, 4, 5])
        self.timestamps = np.array([0.1, 0.2, 0.3, 0.4, 0.5])
        self.edge_idxs = np.array([0, 1, 2, 3, 4])

class MockModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def eval(self):
        return super().eval()
    
    def compute_edge_probabilities(self, sources_batch, destinations_batch, negative_samples, timestamps_batch, edge_idxs_batch, n_neighbors):
        # Returns (pos_prob, neg_prob) as tensors
        size = len(sources_batch)
        pos_prob = torch.sigmoid(torch.randn(size))
        neg_prob = torch.sigmoid(torch.randn(size))
        return pos_prob, neg_prob

ARGS = {
    "model": MockModel(),
    "negative_edge_sampler": MockNegativeEdgeSampler(),
    "data": MockData(),
    "n_neighbors": 10,
    "batch_size": 2,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
get_memory_updater Ran
ran on a synthesized input
twitter-research/tgn/modules/memory_updater.py
code served (permissive licence) · get_code("5ac1084a72c973aa")
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 GRUMemoryUpdater. 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
from torch import nn

torch.manual_seed(0)

# Create a minimal Memory class that satisfies the interface
class MinimalMemory(nn.Module):
    def __init__(self, num_nodes, memory_dimension, device):
        super().__init__()
        self.register_buffer('memory', torch.randn(num_nodes, memory_dimension))
        self.register_buffer('last_update', torch.zeros(num_nodes, dtype=torch.float32))
        self.device = device
    
    def get_memory(self, node_ids):
        return self.memory[node_ids]
    
    def set_memory(self, node_ids, values):
        self.memory[node_ids] = values
    
    def get_last_update(self, node_ids):
        return self.last_update[node_ids]

# Create memory instance
num_nodes = 8
memory_dimension = 4
message_dimension = 3
device = 'cpu'

memory = MinimalMemory(num_nodes, memory_dimension, device)

ARGS = {
    'module_type': 'gru',
    'memory': memory,
    'message_dimension': message_dimension,
    'memory_dimension': memory_dimension,
    'device': device
}
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 GRUMemoryUpdater — a GRUMemoryUpdater output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a GRUMemoryUpdater 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_message_aggregator Ran
ran on a synthesized input
twitter-research/tgn/modules/message_aggregator.py
code served (permissive licence) · get_code("74cf96ac89330114")
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 LastMessageAggregator. 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)

# Define the arguments for get_message_aggregator
ARGS = {
    'aggregator_type': 'last',
    'device': 'cpu'
}
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 LastMessageAggregator — a LastMessageAggregator output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a LastMessageAggregator 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_message_function Ran
ran on a synthesized input
twitter-research/tgn/modules/message_function.py
code served (permissive licence) · get_code("e6724c779cc863ee")
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 MLPMessageFunction. 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)

ARGS = {
    'module_type': 'mlp',
    'raw_message_dimension': 8,
    'message_dimension': 4
}
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 MLPMessageFunction — a MLPMessageFunction output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MLPMessageFunction 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
load_dataset Ran
ran on a synthesized input
jasperzhong/GNNFlow/gnnflow/utils.py
code served (permissive licence) · get_code("59bf623a3b88b301")
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 os
import pandas as pd
import tempfile
import shutil

# Create a temporary directory structure with a mock dataset
temp_dir = tempfile.mkdtemp()
data_dir = os.path.join(temp_dir, 'data', 'test_dataset')
os.makedirs(data_dir, exist_ok=True)

# Create a mock edges.csv file with the required columns
mock_data = pd.DataFrame({
    'eid': range(10),
    'ext_roll': [0, 0, 0, 1, 1, 1, 2, 2, 2, 2],
    'col1': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
    'col2': ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j']
})
csv_path = os.path.join(data_dir, 'edges.csv')
mock_data.to_csv(csv_path, index=False)

# Define ARGS
ARGS = {
    'dataset': 'test_dataset',
    'data_dir': os.path.join(temp_dir, 'data')
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — 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
neg_sample Ran
ran on a synthesized input
petersh6/mspipe/tgl/offline_tgl_presample.py
code served (permissive licence) · get_code("5434bcfd6e9908b3")
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 numpy as np

# Construct arguments for neg_sample function
# The function expects:
# - target_nodes: array of length that's divisible by 3
# - ts: array of timestamps
# - nodes: array of all node IDs to sample negatives from

# Based on the function logic:
# batch_size = len(target_nodes) // 3
# target_pos = len(target_nodes) * 2 // 3
# So if target_nodes has length 9: batch_size=3, target_pos=6
# target_nodes[:6] are positive targets
# target_nodes[6:12] would be dst (but we only have up to 9, so target_nodes[6:9])

# Let's create arrays with length 9 for target_nodes
target_nodes = np.array([10, 11, 12, 20, 21, 22, 5, 6, 7], dtype=np.int64)
# target_nodes[:6] = [10, 11, 12, 20, 21, 22] (positive)
# target_nodes[6:9] = [5, 6, 7] (dst for negative sampling)

# ts array - should be related to batch_size which is 3
ts = np.array([1.0, 2.0, 3.0], dtype=np.float32)

# nodes array - the universe of all nodes to sample negatives from
# Should not contain the dst nodes [5, 6, 7] ideally, but function handles deletion
nodes = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12], dtype=np.int64)

ARGS = {
    "target_nodes": target_nodes,
    "ts": ts,
    "nodes": nodes,
}
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 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
preprocess Ran
ran; our contract draft was wrong, not the code
KimMeen/TGN/utils/preprocess_data.py
code served (permissive licence) · get_code("3e0dea4b496f8cd2")
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) 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 numpy as np
import pandas as pd
import tempfile
import os

# Create a temporary CSV file with sample data
temp_dir = tempfile.mkdtemp()
data_file = os.path.join(temp_dir, "test_data.csv")

# Write sample data to the file
with open(data_file, 'w') as f:
    f.write("0,1,100.5,1.0,0.1,0.2,0.3,0.4\n")
    f.write("1,2,101.5,0.0,0.5,0.6,0.7,0.8\n")
    f.write("0,3,102.5,1.0,0.9,1.0,1.1,1.2\n")

ARGS = {
    "data_name": data_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, pandas — 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
recall Ran
ran; our contract draft was wrong, not the code
wyd1502/DGNN/evaluation.py
code served (permissive licence) · get_code("4d2b9553dedc1537")
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) and returned float64, ndim 0, dtype family float, shape () 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

# Construct arguments for the recall function
# The function expects:
# - des_node: array-like of desired node indices (compared element-wise with idx rows)
# - idx: 2D array of shape (bs, something) containing indices
# - top_k: integer for slicing idx to top_k columns

# Create a small batch size
bs = 2

# des_node: array of desired node indices, one per batch element
# These will be checked against the rows of idx
des_node = np.array([0, 1])

# idx: 2D array of shape (bs, num_candidates) with candidate indices
# We'll make it (2, 5) so we have room to slice with different top_k values
idx = np.array([
    [0, 2, 3, 4, 5],
    [1, 0, 2, 3, 4]
], dtype=np.int64)

# top_k: integer for how many top candidates to consider
top_k = 3

ARGS = {
    "des_node": des_node,
    "idx": idx,
    "top_k": top_k
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float, shape () — shape () 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
reindex Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
KimMeen/TGN/utils/preprocess_data.py
code served (permissive licence) · get_code("4b8167621054e800")
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) and returned DataFrame. 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 pandas as pd
import numpy as np

# Create a small dataframe with the expected structure
data = {
    'u': [0, 1, 1, 2, 1, 3, 2],
    'i': [0, 1, 1, 2, 1, 3, 2],
    'ts': [0.0, 36.0, 77.0, 131.0, 150.0, 200.0, 250.0],
    'label': [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
    'idx': [0, 1, 2, 3, 4, 5, 6]
}

df = pd.DataFrame(data)

ARGS = {
    'df': df,
    'bipartite': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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
MRR Not yet run
unverified (harvested, no recorded run)
wyd1502/DGNN/evaluation.py
code served (permissive licence) · get_code("3537ea56146f8de9")
How it ran

no invocation recorded.

eval_node_classification Not yet run
unverified (harvested, no recorded run)
twitter-research/tgn/evaluation/evaluation.py
code served (permissive licence) · get_code("ce54929f6e81f2e2")
How it ran

no invocation recorded.

get_data Not yet run
unverified (harvested, no recorded run)
twitter-research/tgn/utils/data_processing.py
code served (permissive licence) · get_code("d0e882ad0635a4cb")
How it ran

no invocation recorded.

get_data_node_classification Not yet run
unverified (harvested, no recorded run)
twitter-research/tgn/utils/data_processing.py
code served (permissive licence) · get_code("1b8c08588991e985")
How it ran

no invocation recorded.

get_default_config Not yet run
unverified (harvested, no recorded run)
jasperzhong/GNNFlow/gnnflow/config.py
code served (permissive licence) · get_code("2abeac7066079ca7")
How it ran

no invocation recorded.

get_default_config Not yet run
unverified (harvested, no recorded run)
petersh6/mspipe/gnnflow/config.py
code served (permissive licence) · get_code("61d785f64b05116b")
How it ran

no invocation recorded.

load_partition_table Not yet run
unverified (harvested, no recorded run)
jasperzhong/GNNFlow/gnnflow/utils.py
code served (permissive licence) · get_code("7de6a93ec6c03c40")
How it ran

no invocation recorded.

load_partition_table Not yet run
unverified (harvested, no recorded run)
jasperzhong/GNNFlow/gnnflow/utils.py
code served (permissive licence) · get_code("01f22d83afe320b6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 19 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 18 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
reindex 1 other paper
1 later
2308.14129 (2023-08)
SPEED: Streaming Partition and Parallel Acceleration for Temporal Interaction Graph Embedding

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

Graph Neural Networks (GNNs) have recently become increasingly popular due to their ability to learn complex systems of relations or interactions arising in a broad spectrum of problems ranging from biology and particle physics to social networks and recommendation systems. Despite the plethora of different models for deep learning on graphs, few approaches have been proposed thus far for dealing with graphs that present some sort of dynamic nature (e.g. evolving features or connectivity over time). In this paper, we present Temporal Graph Networks (TGNs), a generic, efficient framework for deep learning on dynamic graphs represented as sequences of timed events. Thanks to a novel combination of memory modules and graph-based operators, TGNs are able to significantly outperform previous approaches being at the same time more computationally efficient. We furthermore show that several previous models for learning on dynamic graphs can be cast as specific instances of our framework. We perform a detailed ablation study of different components of our framework and devise the best configuration that achieves state-of-the-art performance on several transductive and inductive prediction tasks for dynamic graphs.

For agents

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

get_harvested_code_for_paper("2006.10637")
get_code_for_paper("2006.10637")
have("2006.10637")

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

curl https://syntology.ai/api/ran/2006.10637.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.10637.svg)](https://syntology.ai/paper/2006.10637)

Connect an agent — have() is free.