SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2409.04434 · ICLR · 2025

Accelerating Training with Neuron Interaction and Nowcasting Networks

Guillaume Lajoie, Simon Lacoste-Julien, Eugene Belilovsky, Boris Knyazev, Abhinav Moudgil

arXiv · PDF · Open in the Atlas

Code that ran

3 of 7 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 2 have no contract verdict; 1 failure of our instrument; 4 no recorded run), measured 2026-11-21; 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/2409.04434.json · badge: /api/ran/2409.04434.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
samsungsailmontreal/nino canonical (classifier label)
classifier basis: classifier_2025
3 ran · 4 no recorded run
FunctionStatusWhere it lives
EdgeMLP Ran
ran on a synthesized input
samsungsailmontreal/nino/optim/nino.py
code served (permissive licence) · get_code("5274257eb33e5408")
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, shape (3, 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 torch.nn as nn
from torch import Tensor

class EdgeMLP(nn.Module):

    def __init__(self, edge_dim: int, node_dim: int, act, chunk_size: int=0):
        super().__init__()
        self.edge_dim = edge_dim
        self.node_dim = node_dim
        self.chunk_size = chunk_size
        self.lin_e = nn.Linear(edge_dim, edge_dim)
        self.lin_s = nn.Linear(node_dim, edge_dim)
        self.lin_t = nn.Linear(node_dim, edge_dim)
        self.act = act
        self.lin1 = nn.Linear(edge_dim, edge_dim)

    def forward(self, x: Tensor, edge_index, edge_attr: Tensor) -> Tensor:
        if self.training:
            edge_attr = self.lin_e(edge_attr) + self.lin_s(x)[edge_index[0]] + self.lin_t(x)[edge_index[1]]
            edge_attr = self.act(edge_attr)
            edge_attr = self.lin1(edge_attr)
        else:
            assert not torch.is_grad_enabled(), 'this must be run with torch.no_grad() to avoid memory leaks'
            x_s = self.lin_s(x)
            x = self.lin_t(x)
            chunk_size = len(edge_attr) if self.chunk_size in [0, -1, None] else self.chunk_size
            for i in range(0, len(edge_attr), chunk_size):
                edge_attr[i:i + chunk_size] = self.lin_e(edge_attr[i:i + chunk_size])
                edge_attr[i:i + chunk_size] += x_s[edge_index[0][i:i + chunk_size]] + x[edge_index[1][i:i + chunk_size]]
                edge_attr[i:i + chunk_size] = self.act(edge_attr[i:i + chunk_size])
                edge_attr[i:i + chunk_size] = self.lin1(edge_attr[i:i + chunk_size])
        return edge_attr


INIT_ARGS = {
    'edge_dim': 4,
    'node_dim': 3,
    'act': nn.ReLU(),
    'chunk_size': 0,
}

ARGS = {
    'x': torch.randn(2, 3),
    'edge_index': torch.tensor([[0, 1, 0], [1, 0, 1]], dtype=torch.long),
    'edge_attr': torch.randn(3, 4),
}

if __name__ == '__main__':
    model = EdgeMLP(**INIT_ARGS)
    model.train()
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {output}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch, torch_geometric — 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 (3, 4) — shape (3, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
MLP Ran
ran on a synthesized input
samsungsailmontreal/nino/optim/nino.py
code served (permissive licence) · get_code("680f498e65d3b2da")
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, 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 torch.nn as nn
from torch_geometric.nn.resolver import activation_resolver

class MLP(nn.Module):

    def __init__(self, in_dim, hid_dim=None, out_dim=None, n_layers=1, act_name='silu'):
        super().__init__()
        hid_dim = hid_dim or in_dim
        out_dim = out_dim or hid_dim
        layers = []
        for layer in range(n_layers):
            in_dim_ = in_dim if layer == 0 else hid_dim
            out_dim_ = out_dim if layer == n_layers - 1 else hid_dim
            layers.append(nn.Linear(in_dim_, out_dim_))
            if layer < n_layers - 1:
                layers.append(activation_resolver(act_name))
        self.fc = nn.Sequential(*layers)

    def forward(self, x):
        return self.fc(x)


INIT_ARGS = {
    'in_dim': 4,
    'hid_dim': 8,
    'out_dim': 3,
    'n_layers': 2,
    'act_name': 'silu',
}

ARGS = {
    'x': torch.randn(2, 4),
}

if __name__ == '__main__':
    model = MLP(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 3)")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch, torch_geometric — 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
test Ran
ran; our contract draft was wrong, not the code
samsungsailmontreal/nino/train_vision.py
code served (permissive licence) · get_code("9f17a5e0c5f927ac")
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 dict. 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)

# Build a minimal model that:
# - accepts data as input
# - returns output with shape (batch_size, num_classes)
# - can be put in eval mode
class MinimalModel(nn.Module):
    def __init__(self, input_dim=4, num_classes=3):
        super().__init__()
        self.linear = nn.Linear(input_dim, num_classes)
    
    def forward(self, x):
        return self.linear(x)

# Create model instance
model = MinimalModel(input_dim=4, num_classes=3)

# Create data: batch_size=2, input_dim=4
data = torch.randn(2, 4)

# Create target: batch_size=2, with class indices in range [0, 2]
target = torch.tensor([0, 2])

ARGS = {
    'model': model,
    'data': data,
    'target': target,
    'verbose': 0
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
NiNo Not yet run
unverified (harvested, no recorded run)
samsungsailmontreal/nino/optim/nino.py
code served (permissive licence) · get_code("92deafb9b889ae70")
How it ran

no invocation recorded.

NiNoModel Not yet run
unverified (harvested, no recorded run)
samsungsailmontreal/nino/optim/nino.py
code served (permissive licence) · get_code("3937d6b9c027ef1e")
How it ran

no invocation recorded.

PNA Not yet run
unverified (harvested, no recorded run)
samsungsailmontreal/nino/optim/nino.py
code served (permissive licence) · get_code("a309fb0999e41b64")
How it ran

no invocation recorded.

PNAConv Not yet run
unverified (harvested, no recorded run)
samsungsailmontreal/nino/optim/nino.py
code served (permissive licence) · get_code("ad628f616981f9be")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 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 7 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 2026-11-21.

Repositories linked to this paper

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

Abstract

Neural network training can be accelerated when a learnable update rule is used in lieu of classic adaptive optimizers (e.g. Adam). However, learnable update rules can be costly and unstable to train and use. Recently, Jang et al. (2023) proposed a simpler approach to accelerate training based on weight nowcaster networks (WNNs). In their approach, Adam is used for most of the optimization steps and periodically, only every few steps, a WNN nowcasts (predicts near future) parameters. We improve WNNs by proposing neuron interaction and nowcasting (NiNo) networks. In contrast to WNNs, NiNo leverages neuron connectivity and graph neural networks to more accurately nowcast parameters. We further show that in some networks, such as Transformers, modeling neuron connectivity accurately is challenging. We address this and other limitations, which allows NiNo to accelerate Adam training by up to 50% in vision and language tasks.

For agents

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

get_harvested_code_for_paper("2409.04434")
get_code_for_paper("2409.04434")
have("2409.04434")

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

curl https://syntology.ai/api/ran/2409.04434.json

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

[![Syntology run record](https://syntology.ai/api/ran/2409.04434.svg)](https://syntology.ai/paper/2409.04434)

Connect an agent — have() is free.