SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2302.06359 · 2023

Fixing Overconfidence in Dynamic Neural Networks

arXiv · PDF · Open in the Atlas

Code that ran

7 of 9 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 4 have no contract verdict; 1 failure of our instrument; 2 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/2302.06359.json · badge: /api/ran/2302.06359.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
aaltoml/calibrated-dnn canonical (classifier label)
classifier basis not recorded on the link
7 ran · 2 no recorded run
FunctionStatusWhere it lives
Entropy Ran
ran on a synthesized input
aaltoml/calibrated-dnn/adaptive_inference.py
code served (permissive licence) · get_code("045fc85256f7d946")
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, 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

torch.manual_seed(0)

# p: m * n * c
# m: Exits (number of exit points)
# n: Samples (number of samples)
# c: Classes (number of classes)

# Create a valid probability tensor with shape (m, n, c)
# Using small distinct sizes: m=2, n=3, c=4
m, n, c = 2, 3, 4

# Create a tensor with valid probabilities (sum to 1 along dim 2)
p = torch.rand(m, n, c)
# Normalize along the class dimension to make valid probabilities
p = p / p.sum(dim=2, keepdim=True)

# Ensure no zero probabilities to avoid log(0) issues
p = torch.clamp(p, min=1e-10)

ARGS = {
    "p": p
}
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, 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: none by the harness; any seed is in the setup block
accuracy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
aaltoml/calibrated-dnn/adaptive_inference.py
code served (permissive licence) · get_code("e6844a563cce4871")
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
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors for output and target
batch_size = 2
num_classes = 4

# output: logits tensor of shape (batch_size, num_classes)
# This will be used with topk to get predictions
output = torch.randn(batch_size, num_classes)

# target: ground truth class indices of shape (batch_size,)
# Values should be in range [0, num_classes)
target = torch.tensor([1, 3])

# topk: tuple of k values to compute accuracy for
topk = (1, 2)

ARGS = {
    "output": output,
    "target": target,
    "topk": topk
}
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
calc_ensemble_logits Ran
ran on a synthesized input
aaltoml/calibrated-dnn/adaptive_inference.py
code served (permissive licence) · get_code("ad7b29a67bdb8524")
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 3, dtype family float, shape (2, 4, 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

torch.manual_seed(0)

# Create logits tensor: shape (batch=2, dim1=4, dim2=3)
logits = torch.randn(2, 4, 3)

# Create flop_weights: one weight per batch element
flop_weights = torch.tensor([0.3, 0.7])

ARGS = {
    "logits": logits,
    "flop_weights": flop_weights,
}
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 3, dtype family float, shape (2, 4, 3) — shape (2, 4, 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: none by the harness; any seed is in the setup block
get_layer_info Ran
ran on a synthesized input
aaltoml/calibrated-dnn/op_counter.py
code served (permissive licence) · get_code("35a5cf7271901d9c")
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

torch.manual_seed(0)

# The function takes a layer and calls str() on it, then finds the '(' character
# to extract the type name. So we need to pass a PyTorch layer/module.

class SimpleLayer(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

layer = SimpleLayer()

ARGS = {
    "layer": layer
}
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
get_num_gen Ran
ran, honoured the contract we drafted
aaltoml/calibrated-dnn/op_counter.py
code served (permissive licence) · get_code("76cc74efee42ba21")
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 int, ndim 0, dtype family int, 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 torch

torch.manual_seed(0)

def simple_generator():
    """A simple generator that yields a few items"""
    yield 1
    yield 2
    yield 3
    yield 4
    yield 5

ARGS = {
    "gen": simple_generator()
}
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 int, ndim 0, dtype family int, 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
is_leaf Ran
ran, violated the contract we drafted
aaltoml/calibrated-dnn/op_counter.py
code served (permissive licence) · get_code("26ff085b343fa39e")
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 bool, 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 torch
import torch.nn as nn

torch.manual_seed(0)

# The function is_leaf checks if a model has no children by calling model.children()
# We need to create a valid nn.Module instance

# Create a leaf model (no children)
leaf_model = nn.Linear(4, 3)

# Create a non-leaf model (has children) for comparison if needed
parent_model = nn.Sequential(
    nn.Linear(4, 3),
    nn.ReLU()
)

# The function only uses model.children(), so any nn.Module works
# Let's use a leaf model as the test case
ARGS = {
    "model": leaf_model
}
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 bool, 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
norm Ran
ran on a synthesized input
aaltoml/calibrated-dnn/laplace.py
code served (permissive licence) · get_code("cac21f2c1e3022d0")
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 (6,) 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

# Create a small numpy row vector for testing
x = np.array([3.0, 1.0, 4.0, 1.5, 9.0, 2.0])

ARGS = {
    "x": x
}
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 (6,) — shape (6,) 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
estimate_variance_efficient Not yet run
unverified (harvested, no recorded run)
aaltoml/calibrated-dnn/laplace.py
code served (permissive licence) · get_code("5af8b8b09515bd4b")
How it ran

no invocation recorded.

get_dataloaders Not yet run
unverified (harvested, no recorded run)
aaltoml/calibrated-dnn/dataloader.py
code served (permissive licence) · get_code("6ab6ac3cf72849e7")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 1 later) and 5 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
is_leaf 7 other papers
7 dated earlier
1703.09844 (2017-03)
Multi-Scale Dense Networks for Resource Efficient Image Classification
1711.02613 (2017-11)
Moonshine: Distilling with Cheap Convolutions
1904.12368 (2019-04)
Towards Efficient Model Compression via Learned Global Ranking
and 4 more not named here
get_num_gen 6 other papers
6 dated earlier
1703.09844 (2017-03)
Multi-Scale Dense Networks for Resource Efficient Image Classification
1711.02613 (2017-11)
Moonshine: Distilling with Cheap Convolutions
1904.12368 (2019-04)
Towards Efficient Model Compression via Learned Global Ranking
and 3 more not named here
accuracy 5 other papers
2 dated earlier, 3 later
1802.08831 (2018-02)
Convolutional Neural Networks combined with Runge-Kutta Methods
2012.10071 (2020-12)
TDN: Temporal Difference Networks for Efficient Action Recognition
2305.08661 (2023-05)
Global and Local Mixture Consistency Cumulative Learning for Long-tailed Visual Recognitions
and 2 more not named here
get_layer_info 3 other papers
3 dated earlier
1703.09844 (2017-03)
Multi-Scale Dense Networks for Resource Efficient Image Classification
136830480 (2022 (year only))
Bayesian Optimization with Clustering and Rollback for CNN Auto Pruning
2211.16726 (2022-11)
Boosted Dynamic Neural Networks

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

Dynamic neural networks are a recent technique that promises a remedy for the increasing size of modern deep learning models by dynamically adapting their computational cost to the difficulty of the inputs. In this way, the model can adjust to a limited computational budget. However, the poor quality of uncertainty estimates in deep learning models makes it difficult to distinguish between hard and easy samples. To address this challenge, we present a computationally efficient approach for post-hoc uncertainty quantification in dynamic neural networks. We show that adequately quantifying and accounting for both aleatoric and epistemic uncertainty through a probabilistic treatment of the last layers improves the predictive performance and aids decision-making when determining the computational budget. In the experiments, we show improvements on CIFAR-100, ImageNet, and Caltech-256 in terms of accuracy, capturing uncertainty, and calibration error.

For agents

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

get_harvested_code_for_paper("2302.06359")
get_code_for_paper("2302.06359")
have("2302.06359")

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

curl https://syntology.ai/api/ran/2302.06359.json

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

[![Syntology run record](https://syntology.ai/api/ran/2302.06359.svg)](https://syntology.ai/paper/2302.06359)

Connect an agent — have() is free.