SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1901.08276 · 2019

Traditional and Heavy-Tailed Self Regularization in Neural Network Models

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 3 have no contract verdict; 0 failures of our instrument; 4 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/1901.08276.json · badge: /api/ran/1901.08276.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
CalculatedContent/WeightWatcher community (classifier label)
classifier basis not recorded on the link
3 ran · 4 no recorded run
FunctionStatusWhere it lives
apply_trap_mp_fit Ran
ran on a synthesized input
CalculatedContent/WeightWatcher/weightwatcher/remove_traps.py
code served (permissive licence) · get_code("0a2c8739a51e4984")
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 SimpleLayer. 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 stand-in class for ww that has apply_esd and apply_mp_fit methods
class WeightWatcher:
    def __init__(self):
        self.esd_applied = False
        self.mp_fit_applied = False
    
    def apply_esd(self, ww_layer, params):
        """Stand-in for ESD application"""
        self.esd_applied = True
    
    def apply_mp_fit(self, ww_layer, random=False, params=None):
        """Stand-in for MP fit application"""
        self.mp_fit_applied = True

# Build a simple layer that can be passed to the ww methods
class SimpleLayer(nn.Module):
    def __init__(self):
        super().__init__()
        self.weight = nn.Parameter(torch.randn(4, 3))
        self.bias = nn.Parameter(torch.randn(4))
    
    def forward(self, x):
        return torch.nn.functional.linear(x, self.weight, self.bias)

# Create instances
ww = WeightWatcher()
ww_layer = SimpleLayer()

# Define params dict with some reasonable defaults
params = {
    'layer_id': 0,
    'esd_power': 1.0,
    'max_moments': 100,
}

ARGS = {
    'ww': ww,
    'ww_layer': ww_layer,
    'params': params
}
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 SimpleLayer — a SimpleLayer output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleLayer 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
identify_trap_mode_indices Ran
ran on a synthesized input
CalculatedContent/WeightWatcher/weightwatcher/remove_traps.py
code served (permissive licence) · get_code("792f4c10b5c1b21a")
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 numpy as np
from scipy.linalg import svd

def svd_full(W):
    """Helper function to perform SVD on a weight matrix."""
    U, s, Vt = svd(W, full_matrices=True)
    return U, s, Vt

class MockWWLayer:
    """Mock layer object with the attributes needed by identify_trap_mode_indices."""
    def __init__(self):
        # Create a small weight matrix (4x3)
        self.Wmats = [np.random.randn(4, 3)]
        self.N = 8  # number of samples
        self.M = 4  # dimension
        self.sigma_mp = 1.5  # Marchenko-Pastur parameter
        self.W_scale = 0.5  # weight scale

class MockWW:
    """Mock WW object (may not be used but included for completeness)."""
    pass

# Create instances
ww_layer = MockWWLayer()
ww = MockWW()

# Compute singular values for reference
W = ww_layer.Wmats[0]
_, svals, _ = svd_full(W)

# Create evals_desc from singular values
evals_desc = svals * svals

# Construct ARGS dictionary
ARGS = {
    "ww": ww,
    "ww_layer": ww_layer,
    "svals": svals,
    "evals_desc": evals_desc,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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
summarize_trace Ran
ran on a synthesized input
CalculatedContent/WeightWatcher/weightwatcher/compute_trace.py
code served (permissive licence) · get_code("5fab4af8f547e332")
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 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 numpy as np

torch.manual_seed(0)

# Construct a realistic trace events list
events = [
    {
        'event_type': 'svd_full_end',
        'phase': 'randomize_model',
        'layer_id': 0,
        'elapsed_ms': 10.5
    },
    {
        'event_type': 'svd_full_end',
        'phase': 'randomize_model',
        'layer_id': 1,
        'elapsed_ms': 12.3
    },
    {
        'event_type': 'svd_full_end',
        'phase': 'analyze_traps',
        'layer_id': 0,
        'elapsed_ms': 8.7
    },
    {
        'event_type': 'svd_full_end',
        'phase': 'remove_traps',
        'layer_id': 2,
        'elapsed_ms': 15.2
    },
    {
        'event_type': 'full_bulk_metrics',
        'sampled_bulk_modes': 0
    },
    {
        'event_type': 'full_bulk_metrics',
        'sampled_bulk_modes': 0
    },
    {
        'event_type': 'fast_bulk_metrics',
        'sampled_bulk_modes': 64
    },
    {
        'event_type': 'fast_bulk_metrics',
        'sampled_bulk_modes': 32
    },
    {
        'event_type': 'original_basis_metrics',
        'sampled_bulk_modes': 0
    },
    {
        'event_type': 'collect_trap_artifacts_start',
        'sampled_bulk_modes': 0
    },
    {
        'event_type': 'other_event',
        'data': 'some_value'
    }
]

ARGS = {
    'events': events
}
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 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
analyze_single_trap Not yet run
unverified (harvested, no recorded run)
CalculatedContent/WeightWatcher/weightwatcher/remove_traps.py
code served (permissive licence) · get_code("64443cf24b9525f2")
How it ran

no invocation recorded.

load_trap_bundle Not yet run
unverified (harvested, no recorded run)
CalculatedContent/WeightWatcher/weightwatcher/trap_bundles.py
code served (permissive licence) · get_code("b1fbf1fc91adc0a3")
How it ran

no invocation recorded.

pl_compare Not yet run
unverified (harvested, no recorded run)
CalculatedContent/WeightWatcher/weightwatcher/WW_powerlaw.py
code served (permissive licence) · get_code("cb6ec82893bf637a")
How it ran

no invocation recorded.

save_trap_bundle Not yet run
unverified (harvested, no recorded run)
CalculatedContent/WeightWatcher/weightwatcher/trap_bundles.py
code served (permissive licence) · get_code("60345cbbb29b8e97")
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 2026-11-22.

Repositories linked to this paper

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

Abstract

Random Matrix Theory (RMT) is applied to analyze the weight matrices of Deep Neural Networks (DNNs), including both production quality, pre-trained models such as AlexNet and Inception, and smaller models trained from scratch, such as LeNet5 and a miniature-AlexNet. Empirical and theoretical results clearly indicate that the empirical spectral density (ESD) of DNN layer matrices displays signatures of traditionally-regularized statistical models, even in the absence of exogenously specifying traditional forms of regularization, such as Dropout or Weight Norm constraints. Building on recent results in RMT, most notably its extension to Universality classes of Heavy-Tailed matrices, we develop a theory to identify \emph{5+1 Phases of Training}, corresponding to increasing amounts of \emph{Implicit Self-Regularization}. For smaller and/or older DNNs, this Implicit Self-Regularization is like traditional Tikhonov regularization, in that there is a `size scale' separating signal from noise. For state-of-the-art DNNs, however, we identify a novel form of \emph{Heavy-Tailed Self-Regularization}, similar to the self-organization seen in the statistical physics of disordered systems. This implicit Self-Regularization can depend strongly on the many knobs of the training process. By exploiting the generalization gap phenomena, we demonstrate that we can cause a small model to exhibit all 5+1 phases of training simply by changing the batch size.

For agents

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

get_harvested_code_for_paper("1901.08276")
get_code_for_paper("1901.08276")
have("1901.08276")

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

curl https://syntology.ai/api/ran/1901.08276.json

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

[![Syntology run record](https://syntology.ai/api/ran/1901.08276.svg)](https://syntology.ai/paper/1901.08276)

Connect an agent — have() is free.