SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2005.05220 · 2020

iUNets: Fully invertible U-Nets with Learnable Up- and Downsampling

arXiv · PDF · Open in the Atlas

Code that ran

12 of 14 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 10 have no contract verdict; 2 failures 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/2005.05220.json · badge: /api/ran/2005.05220.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
cetmann/iunets canonical (classifier label)
classifier basis: census
5 ran · 2 no recorded run
silvandeleemput/memcnn community (classifier label)
classifier basis not recorded on the link
7 ran · 0 no recorded run
FunctionStatusWhere it lives
accuracy Ran
ran on a synthesized input
silvandeleemput/memcnn/memcnn/utils/stats.py
code served (permissive licence) · get_code("b76726eb649a2075")
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
import torch.nn as nn

torch.manual_seed(0)

# The function expects:
# - output: a tensor with shape (batch_size, num_classes) that supports .topk() and .size()
# - target: a tensor with shape (batch_size,) that supports .size() and .view()
# - topk: a tuple of integers (default (1,))

batch_size = 2
num_classes = 4

# Create output logits tensor (batch_size x num_classes)
output = torch.randn(batch_size, num_classes)

# Create target labels tensor (batch_size,) with values in [0, num_classes)
target = torch.tensor([0, 2], dtype=torch.long)

ARGS = {
    "output": output,
    "target": target,
    "topk": (1, 2)
}
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
batch_norm Ran
ran; our contract draft was wrong, not the code
silvandeleemput/memcnn/memcnn/models/resnet.py
code served (permissive licence) · get_code("06864fef7f15d70b")
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 BatchNorm2d. 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 batch_norm(x) expects x to be an integer (number of features/channels)
# It calls nn.BatchNorm2d(x, ...) which expects an integer as the first argument

ARGS = {
    "x": 4
}
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 BatchNorm2d — a BatchNorm2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a BatchNorm2d 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
build_dict Ran
ran on a synthesized input
silvandeleemput/memcnn/memcnn/experiment/factory.py
code served (permissive licence) · get_code("a07ee03ce1d4a9b5")
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 copy

# The function build_dict takes three parameters:
# 1. experiments_dict: a dict mapping experiment names to their config dicts
# 2. experiment_name: a string naming which experiment to build
# 3. classhist: optional list to track inheritance chain (defaults to None)

# Looking at the function logic:
# - It accesses experiments_dict[experiment_name]
# - It checks for 'base' key in the config dict
# - It recursively calls build_dict with the base experiment name
# - It uses _inject_items to merge configs
# - It checks for circular dependencies by looking in classhist

# Create a valid experiments_dict with a simple hierarchy
experiments_dict = {
    "base_config": {
        "learning_rate": 0.001,
        "batch_size": 32,
        "optimizer": {
            "type": "adam",
            "beta1": 0.9
        },
        "base": None
    },
    "child_config": {
        "learning_rate": 0.0001,
        "model": {
            "hidden_dim": 128
        },
        "base": "base_config"
    },
    "grandchild_config": {
        "batch_size": 64,
        "base": "child_config"
    }
}

ARGS = {
    "experiments_dict": experiments_dict,
    "experiment_name": "grandchild_config",
    "classhist": None
}
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 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
conv3x3 Ran
ran; our contract draft was wrong, not the code
silvandeleemput/memcnn/memcnn/models/resnet.py
pointer only (licence: MIT); the stored copy get_code serves, from lixiaotong97/DSU, is pointer only (licence: NONE) · get_code("fac5364e2f53c6db")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from lixiaotong97/DSU, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

eye_like Ran
ran on a synthesized input
cetmann/iunets/iunets/utils.py
code served (permissive licence) · get_code("580b5bba856a169e")
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 (4, 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)

# Test case 1: 2D matrix (n, n)
M_2d = torch.randn(4, 4)

# Test case 2: 3D batch of matrices (m, n, n)
M_3d = torch.randn(2, 4, 4)

# We'll use the 2D case as the main test
M = M_2d

# device and dtype parameters
device = None  # Will use M's device
dtype = None   # Will use M's dtype

ARGS = {
    "M": M,
    "device": device,
    "dtype": dtype,
}
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 (4, 4) — shape (4, 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
get_attr_from_module Ran
ran on a synthesized input
silvandeleemput/memcnn/memcnn/experiment/factory.py
code served (permissive licence) · get_code("5a1180a5d05a8e13")
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
import torch.nn as nn

# The function get_attr_from_module takes a string parameter 'pclass' 
# that represents a module path and class name in the format "module.path.ClassName"
# It splits on the last dot, imports the module, and returns the class/attribute

# To construct valid arguments, we need a string that:
# 1. Can be split into a valid module path and attribute name
# 2. The module must be importable
# 3. The attribute must exist in that module

# Example: "torch.nn.Linear" - torch.nn is importable, and Linear exists in it
# Another example: "torch.Tensor" - torch is importable, and Tensor exists in it

ARGS = {
    "pclass": "torch.nn.Linear"
}
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
get_model_parameters_count Ran
ran on a synthesized input
silvandeleemput/memcnn/memcnn/trainers/classification.py
code served (permissive licence) · get_code("8973f10202a02a78")
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 int64, 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
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

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

model = SimpleModel()

ARGS = {
    "model": model
}
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 int64, 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
get_num_channels Ran
ran on a synthesized input
cetmann/iunets/iunets/utils.py
code served (permissive licence) · get_code("97380980111ac54e")
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)

# Test case 1: passing a tuple (iterable) representing input shape
input_shape = (3, 32, 32)

# Test case 2: passing an integer representing number of channels
num_channels = 3

ARGS = {
    'input_shape_or_channels': input_shape
}
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
householder_transform Ran
ran on a synthesized input
cetmann/iunets/iunets/householder.py
code served (permissive licence) · get_code("5ea008d866ea3892")
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; 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)

def eye_like(tensor):
    """Helper function that the code uses but doesn't define."""
    return torch.eye(tensor.shape[-2], dtype=tensor.dtype, device=tensor.device)

# Monkey-patch the eye_like function into the module scope for the function to use
import sys
current_module = sys.modules[__name__]
setattr(current_module, 'eye_like', eye_like)

# Create a small test matrix: batch_size=2, n_rows=4, n_cols=3
matrix = torch.randn(2, 4, 3, dtype=torch.float32)

ARGS = {
    "matrix": matrix,
    "n_reflections": -1,
    "eps": 1e-06
}
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 — no output shape is stated: observe_signature.py could not observe it (fixture:KeyError)
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
load_experiment_config Ran
ran on a synthesized input
silvandeleemput/memcnn/memcnn/experiment/factory.py
code served (permissive licence) · get_code("7507bee5398cddc3")
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 json
import tempfile
import os

# Create a temporary JSON file with experiment configurations
experiments_data = {
    "base_exp": {
        "param1": "value1",
        "param2": 42,
        "nested": {
            "key1": "nested_value1"
        }
    },
    "child_exp": {
        "base": "base_exp",
        "param2": 100,
        "param3": "new_value"
    },
    "another_exp": {
        "param_a": "a_value",
        "param_b": [1, 2, 3]
    }
}

# Write the experiments data to a temporary JSON file
temp_file = tempfile.NamedTemporaryFile(mode='w', suffix='.json', delete=False)
json.dump(experiments_data, temp_file)
temp_file.close()
experiments_file_path = temp_file.name

# Define the arguments for load_experiment_config
ARGS = {
    "experiments_file": experiments_file_path,
    "experiment_tags": ["base_exp", "child_exp"]
}
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 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
matrix_1_norm Ran
ran on a synthesized input
cetmann/iunets/iunets/expm.py
code served (permissive licence) · get_code("f4e413b9d7da26a2")
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)

# Test case 1: single matrix (2, 3)
A_single = torch.randn(2, 3)

# Test case 2: batch of matrices (4, 2, 3)
A_batch = torch.randn(4, 2, 3)

ARGS = {
    "A": A_batch
}
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
normalize_matrix_rows Ran
ran on a synthesized input
cetmann/iunets/iunets/householder.py
code served (permissive licence) · get_code("d5d68c1071441b28")
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 a small matrix tensor with shape (2, 4, 3)
# This represents 2 batches of 4x3 matrices
# dim=-2 is the second-to-last dimension (size 4)
matrix = torch.randn(2, 4, 3)

ARGS = {
    "matrix": matrix,
    "eps": 1e-06,
}
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
create_standard_module Not yet run
unverified (harvested, no recorded run)
cetmann/iunets/iunets/layers.py
code served (permissive licence) · get_code("3f2f6049524c9cb1")
How it ran

no invocation recorded.

householder_matrix Not yet run
unverified (harvested, no recorded run)
cetmann/iunets/iunets/householder.py
code served (permissive licence) · get_code("ea4b1e4a5d5e867e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 2 later) and 12 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
conv3x3 686 other papers
221 dated earlier, 4 the same month, 460 later, 1 same year, order not determinable · 31 only as an assembly brick
1406.2661 (2014-06)
Generative Adversarial Networks
1409.0575 (2014-09)
ImageNet Large Scale Visual Recognition Challenge
1412.6572 (2014-12)
Explaining and Harnessing Adversarial Examples
and 683 more not named here
batch_norm 1 other paper
1 later
2406.04312 (2024-06)
ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization

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

U-Nets have been established as a standard architecture for image-to-image learning problems such as segmentation and inverse problems in imaging. For large-scale data, as it for example appears in 3D medical imaging, the U-Net however has prohibitive memory requirements. Here, we present a new fully-invertible U-Net-based architecture called the iUNet, which employs novel learnable and invertible up- and downsampling operations, thereby making the use of memory-efficient backpropagation possible. This allows us to train deeper and larger networks in practice, under the same GPU memory restrictions. Due to its invertibility, the iUNet can furthermore be used for constructing normalizing flows.

For agents

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

get_harvested_code_for_paper("2005.05220")
get_code_for_paper("2005.05220")
have("2005.05220")

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

curl https://syntology.ai/api/ran/2005.05220.json

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

[![Syntology run record](https://syntology.ai/api/ran/2005.05220.svg)](https://syntology.ai/paper/2005.05220)

Connect an agent — have() is free.