SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2311.01644 · NeurIPS · 2023

Should Under-parameterized Student Networks Copy or Average Teacher Weights?

Wulfram Gerstner, Johanni Brea, Nyu, Şimşek Berfin, Amire Bendjeddou

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 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; 1 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 1 failure of our instrument; 0 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/2311.01644.json · badge: /api/ran/2311.01644.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
berfinsimsek/neural-net-regression canonical (classifier label)
classifier basis: census
3 ran · 0 no recorded run
FunctionStatusWhere it lives
optim_loss_erf Ran
ran, violated the contract we drafted
berfinsimsek/neural-net-regression/helper.py
code served (permissive licence) · get_code("686217491193858e")
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 ndarray, 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 numpy as np

def optim_loss_erf(x):
    return (x * np.arcsin(1 / 2) - x ** 2 * np.arcsin(1 / (2 * x))) * 2 / np.pi

# The function takes a single parameter 'x' which is used in arithmetic operations
# with np.arcsin, so it needs to be a numeric type that numpy can handle.
# Looking at the function body, x is used in:
# - x * np.arcsin(1 / 2)
# - x ** 2 * np.arcsin(1 / (2 * x))
# So x should be a numpy array or scalar numeric value

x_value = np.array([2.0, 3.0, 4.0, 5.0])

ARGS = {
    "x": x_value
}
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 ndarray, 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
permute_CA Ran
ran, honoured the contract we drafted
berfinsimsek/neural-net-regression/helper.py
code served (permissive licence) · get_code("887080d31a20e660")
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 permute_CA function
# The function operates on:
# - angles: shape (d, n) - 2D array of angles
# - norms: shape (n,) - 1D array of norms
# - outgoing_weights: shape (n,) - 1D array of outgoing weights

d = 4  # number of rows (distinct from n)
n = 3  # number of columns

# Create angles array (d x n)
angles = np.random.randn(d, n) * 0.5  # Keep values small, mostly < 0.95 in abs value

# Create norms array (n,)
norms = np.random.randn(n) * 0.5

# Create outgoing_weights array (n,)
outgoing_weights = np.random.randn(n) * 0.5

ARGS = {
    "angles": angles,
    "norms": norms,
    "outgoing_weights": outgoing_weights
}
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
res_to_param Ran
ran; our contract draft was wrong, not the code
berfinsimsek/neural-net-regression/helper.py
code served (permissive licence) · get_code("0824b9a920b418b8")
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 a valid 'res' dictionary with the structure expected by res_to_param
# The function expects res['x']['w1'] and res['x']['w2']

num_teacher = 3
num_neurons = 4

# w1 has shape (num_neurons, num_teacher)
w1 = np.random.randn(num_neurons, num_teacher)

# w2 has shape (1, num_neurons)
w2 = np.random.randn(1, num_neurons)

res = {
    'x': {
        'w1': w1,
        'w2': w2
    }
}

ARGS = {
    'res': res,
    'num_teacher': num_teacher
}
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

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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

Any continuous function f * can be approximated arbitrarily well by a neural network with sufficiently many neurons k. We consider the case when f * itself is a neural network with one hidden layer and k neurons. Approximating f * with a neural network with n < k neurons can thus be seen as fitting an under-parameterized "student" network with n neurons to a "teacher" network with k neurons. As the student has fewer neurons than the teacher, it is unclear, whether each of the n student neurons should copy one of the teacher neurons or rather average a group of teacher neurons. For shallow neural networks with erf activation function and for the standard Gaussian input distribution, we prove that "copy-average" configurations are critical points if the teacher's incoming vectors are orthonormal and its outgoing weights are unitary. Moreover, the optimum among such configurations is reached when n -1 student neurons each copy one teacher neuron and the n-th student neuron averages the remaining kn + 1 teacher neurons. For the student network with n = 1 neuron, we provide additionally a closed-form solution of the non-trivial critical point(s) for commonly used activation functions through solving an equivalent constrained optimization problem. Empirically, we find for the erf activation function that gradient flow converges either to the optimal copy-average critical point or to another point where each student neuron approximately copies a different teacher neuron. Finally, we find similar results for the ReLU activation function, suggesting that the optimal solution of underparameterized networks has a universal structure.

For agents

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

get_harvested_code_for_paper("2311.01644")
get_code_for_paper("2311.01644")
have("2311.01644")

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

curl https://syntology.ai/api/ran/2311.01644.json

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

[![Syntology run record](https://syntology.ai/api/ran/2311.01644.svg)](https://syntology.ai/paper/2311.01644)

Connect an agent — have() is free.