SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.01645 · ICML · 2024

Sign Gradient Descent-based Neuronal Dynamics: ANN-to-SNN Conversion Beyond ReLU Network

Hyunseok Oh, Youngki Lee

arXiv · PDF · Open in the Atlas

Code that ran

12 of 13 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; 3 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 9 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/2407.01645.json · badge: /api/ran/2407.01645.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
snuhcs/snn_signgd canonical (classifier label)
classifier basis: census
12 ran · 1 no recorded run
FunctionStatusWhere it lives
copy_if_exists Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/core/layer.py
code served (permissive licence) · get_code("b9c21ece96a51e25")
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 Parameter, 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
from torch.nn.parameter import Parameter

torch.manual_seed(0)

# The function copy_if_exists takes a param argument.
# It checks if param is not None, and if so:
# - calls param.clone() to clone the parameter
# - wraps it in Parameter()
# Otherwise returns None

# We need to test both cases: with a Parameter and with None

# Case 1: Create a valid Parameter with a small tensor
param_tensor = torch.randn(2, 4)
param_with_value = Parameter(param_tensor)

# Case 2: None case
param_none = None

# Build ARGS dict with the parameter argument
ARGS = {
    "param": param_with_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 Parameter, ndim 2, dtype family float — no output shape is stated: observe_signature.py recorded Tensor ndim 2, which disagrees with the fixture run
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_inputs_and_outputs Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/graph_functional/utils.py
code served (permissive licence) · get_code("84111747a4743f85")
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 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
from torch import fx

torch.manual_seed(0)

# Create a simple model to trace
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# Create a model instance and trace it
model = SimpleModel()
example_input = torch.randn(2, 4)
traced = fx.symbolic_trace(model)

# Get the graph from the traced module
graph = traced.graph

ARGS = {
    "graph": graph
}
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, ndim 2, 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_parent_name Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/graph_functional/utils.py
code served (permissive licence) · get_code("28353e985d866fe1")
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 torch
import torch.nn as nn

# Analyze the function:
# - get_parent_name takes a single parameter `target` which is a string
# - It splits the string using rsplit('.', 1) to separate parent path from last atom
# - Returns a tuple of (parent_path, name)
# - Examples: "foo.bar.baz" -> ("foo.bar", "baz")

ARGS = {
    "target": "foo.bar.baz"
}
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, 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
import_module_from_path Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/utils.py
code served (permissive licence) · get_code("f5954937d705ef0a")
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 type. 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 sys
import tempfile
from pathlib import Path

# Create a temporary directory structure for testing
temp_dir = tempfile.mkdtemp()
module_dir = os.path.join(temp_dir, "test_module")
os.makedirs(module_dir, exist_ok=True)

# Create a minimal __init__.py file with a class/object named "test_module"
init_content = """
class test_module:
    def __init__(self):
        self.value = "test"
    
    def method(self):
        return "executed"
"""

with open(os.path.join(module_dir, "__init__.py"), "w") as f:
    f.write(init_content)

# Define ARGS dictionary with valid arguments for the function
ARGS = {
    "path": module_dir,
    "module_name": "test_module"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
multiply_inverse_of_square_root Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/core/layer.py
code served (permissive licence) · get_code("a2295cefbeb3475a")
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, 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

torch.manual_seed(0)

x = torch.randn(2, 4)
y = torch.randn(2, 4) + 1.0  # Add 1.0 to ensure positive values for sqrt

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
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, 4) — shape (2, 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
reduce_duplicates Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
snuhcs/snn_signgd/src/packages/snn_signgd/system_optimizations.py
code served (permissive licence) · get_code("a4ea239e37558106")
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 1, dtype family float, shape (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

torch.manual_seed(0)

class TensorPair:
    def __init__(self, x, y):
        self.x = x
        self.y = y

# Create a small tensor with duplicates along dimension 0
tensor_with_dupes = torch.tensor([[1.0, 2.0, 3.0, 4.0], 
                                   [1.0, 2.0, 3.0, 4.0]], dtype=torch.float32)

ARGS = {
    "tensor": tensor_with_dupes
}
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 Tensor, ndim 1, dtype family float, shape (4,) — shape (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
repeat_shape_base Ran
ran, honoured the contract we drafted
snuhcs/snn_signgd/src/packages/snn_signgd/utils.py
code served (permissive licence) · get_code("3bf7592ec2efc27c")
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, ndim 1, dtype family int; 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

torch.manual_seed(0)

# Create a small tensor with distinct dimensions for shape testing
x = torch.randn(2, 4, 3, 5)

ARGS = {
    "x": x
}
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 list, ndim 1, dtype family int — a list 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
replace_node_module Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/graph_functional/utils.py
code served (permissive licence) · get_code("728ef6b02e252dd8")
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 str. 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 import fx
from typing import Dict, Any

torch.manual_seed(0)

# Create a simple module hierarchy to work with
class SimpleModule(nn.Module):
    def __init__(self):
        super().__init__()
        self.child = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.child(x)

class ParentModule(nn.Module):
    def __init__(self):
        super().__init__()
        self.foo = nn.Module()
        self.foo.bar = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.foo.bar(x)

# Create the parent module instance
parent_module = ParentModule()

# Create a traced FX graph to get a valid Node
def dummy_forward(x):
    return parent_module.foo.bar(x)

try:
    traced = fx.symbolic_trace(parent_module)
    # Get a node from the traced graph
    nodes = list(traced.graph.nodes)
    # Find a call_module node
    node = None
    for n in nodes:
        if n.op == 'call_module':
            node = n
            break
    
    if node is None:
        # Create a mock node if symbolic trace doesn't work as expected
        node = fx.Node(traced.graph, 'call_module', 'call_module', 'foo.bar', (), {})
        node.name = 'foo_bar'
        node.target = 'foo.bar'
except:
    # Fallback: create a minimal mock node
    class MockNode:
        def __init__(self):
            self.target = 'foo.bar'
            self.name = 'foo_bar'
    node = MockNode()

# Create modules dict that mirrors the structure
modules = {
    '': parent_module,
    'foo': parent_module.foo,
    'foo_bar': nn.Identity()  # placeholder
}

# Create a new module to replace with
new_module = nn.Linear(4, 2)

ARGS = {
    'node': node,
    'modules': modules,
    'new_module': new_module
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
running_average Ran
ran, honoured the contract we drafted
snuhcs/snn_signgd/src/packages/snn_signgd/core/hook.py
code served (permissive licence) · get_code("62209dabe65cb4da")
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, 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

torch.manual_seed(0)

# Analyze the function:
# running_average(state, input, timestep)
# - state: used in arithmetic operations (multiplication, addition)
# - input: used in arithmetic operations (multiplication, addition)
# - timestep: used in arithmetic operations (division, addition)
# All three parameters appear to be numeric types that support standard operations

# Create small tensors for state and input
state = torch.randn(2, 4)
input_tensor = torch.randn(2, 4)
timestep = torch.tensor(5.0)

ARGS = {
    "state": state,
    "input": input_tensor,
    "timestep": timestep
}
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, 4) — shape (2, 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
save_and_reimport_nn_module Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/utils.py
code served (permissive licence) · get_code("b6e46dc0b37a272f")
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 type. 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, fx
import tempfile
import os

torch.manual_seed(0)

# Create a simple nn.Module to trace
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# Create an instance and trace it to get a GraphModule
simple_model = SimpleModel()
example_input = torch.randn(2, 4)
traced_model = fx.symbolic_trace(simple_model)

# Create a temporary directory for the test
temp_dir = tempfile.mkdtemp()

ARGS = {
    'model': traced_model,
    'temporary_directory': temp_dir
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
traversal_pre Ran
ran; our contract draft was wrong, not the code
snuhcs/snn_signgd/src/packages/snn_signgd/graph_functional/pattern_matching.py
code served (permissive licence) · get_code("45bc1a945c9346c6")
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, 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
from torch import nn, fx
from typing import List, Tuple

torch.manual_seed(0)

# Create a simple model to trace
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

# Create a model and trace it
model = SimpleModel()
example_input = torch.randn(2, 4)
traced = fx.symbolic_trace(model)

# Get a node from the traced graph that has arguments
# We need to find a node with fx.Node arguments
node_with_args = None
for node in traced.graph.nodes:
    if len(node.args) > 0:
        # Check if any args are fx.Node instances
        has_node_arg = any(isinstance(arg, fx.Node) for arg in node.args)
        if has_node_arg:
            node_with_args = node
            break

# If no node with args found, create a simple scenario
if node_with_args is None:
    # Get any node that has args
    for node in traced.graph.nodes:
        if len(node.args) > 0:
            node_with_args = node
            break

# Fallback: use the first non-placeholder node
if node_with_args is None:
    for node in traced.graph.nodes:
        if node.op != 'placeholder':
            node_with_args = node
            break

ARGS = {
    'node': node_with_args,
    'depth': 0,
    'max_depth': 3
}
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 list, ndim 2, dtype family float — a list 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
walk_backward Ran
ran, honoured the contract we drafted
snuhcs/snn_signgd/src/packages/snn_signgd/graph_functional/pattern_matching.py
code served (permissive licence) · get_code("987da864ece2cd79")
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, 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 torch
from torch import nn, fx

torch.manual_seed(0)

# Create a simple model to trace
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 8)
        self.linear2 = nn.Linear(8, 3)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

# Create and trace the model to get an fx.GraphModule
model = SimpleModel()
example_input = torch.randn(2, 4)
traced_model = fx.symbolic_trace(model)

# Get a node from the traced graph
# We need to find a node that has args (not a placeholder or parameter)
target_node = None
for node in traced_model.graph.nodes:
    if node.op == 'call_module' or node.op == 'call_function':
        if len(node.args) > 0:
            target_node = node
            break

# If we couldn't find a suitable node, create one manually
if target_node is None:
    # Get any non-placeholder node with arguments
    for node in traced_model.graph.nodes:
        if node.op != 'placeholder':
            target_node = node
            break

# Fallback: use the first call_module or call_function node
if target_node is None:
    for node in traced_model.graph.nodes:
        if node.op in ['call_module', 'call_function']:
            target_node = node
            break

# Final fallback: create a simple graph with nodes that have arguments
if target_node is None:
    # Build a minimal graph manually
    graph = fx.GraphModule(nn.Module(), fx.GraphModule(nn.Module(), fx.Graph()).graph)
    with graph.graph.inserting_before(None):
        placeholder = graph.graph.placeholder('x')
        call_node = graph.graph.call_function(torch.relu, args=(placeholder,))
        graph.graph.output(call_node)
    target_node = call_node

ARGS = {
    'node': target_node,
    'length': 3
}
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 list, ndim 1, dtype family float — a list 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
porting Not yet run
unverified (harvested, no recorded run)
snuhcs/snn_signgd/src/packages/snn_signgd/port_ann.py
code served (permissive licence) · get_code("fe922121061b3d44")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 13 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Spiking neural network (SNN) is studied in multidisciplinary domains to (i) enable order-ofmagnitudes energy-efficient AI inference and (ii) computationally simulate neuroscientific mechanisms. The lack of discrete theory obstructs the practical application of SNN by limiting its performance and nonlinearity support. We present a new optimization-theoretic perspective of the discrete dynamics of spiking neurons. We prove that a discrete dynamical system of simple integrateand-fire models approximates the subgradient method over unconstrained optimization problems. We practically extend our theory to introduce a novel sign gradient descent (signGD)based neuronal dynamics that can (i) approximate diverse nonlinearities beyond ReLU and (ii) advance ANN-to-SNN conversion performance in low time steps. Experiments on large-scale datasets show that our technique achieves (i) stateof-the-art performance in ANN-to-SNN conversion and (ii) is the first to convert new DNN architectures, e.g., ConvNext, MLP-Mixer, and ResMLP. We publicly share our source code at www.github.com/snuhcs/snn signgd .

For agents

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

get_harvested_code_for_paper("2407.01645")
get_code_for_paper("2407.01645")
have("2407.01645")

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

curl https://syntology.ai/api/ran/2407.01645.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.01645.svg)](https://syntology.ai/paper/2407.01645)

Connect an agent — have() is free.