SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1804.09170 · 2018

Realistic Evaluation of Deep Semi-Supervised Learning Algorithms

arXiv · PDF · Open in the Atlas

Code that ran

9 of 15 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, 0 violated it and 5 have no contract verdict; 3 failures of our instrument; 6 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/1804.09170.json · badge: /api/ran/1804.09170.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
brain-research/realistic-ssl-evaluation canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
siit-vtt/semi-supervised-learning-pytorch community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
melkherj/puddle community (classifier label)
classifier basis not recorded on the link
3 ran · 2 no recorded run
perrying/realistic-ssl-evaluation-pytorch reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 1 no recorded run
DavidKo3/semi_supervised_mean_teacher reimplementation (classifier label)
classifier basis: classifier_2025
0 ran · 3 no recorded run
FunctionStatusWhere it lives
active_evaluate Ran
ran on a synthesized input
melkherj/puddle/puddle/active_evaluate.py
code served (permissive licence) · get_code("fefabd373ecadbc2")
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 DataFrame. 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
import random
from sklearn.linear_model import LogisticRegression

random.seed(0)
np.random.seed(0)

# Create synthetic data
n_samples = 50
n_features = 4

X = np.random.randn(n_samples, n_features)
Y = np.random.randint(0, 2, n_samples).astype(bool)

# Create a simple selector class that implements next_indices
class SimpleSelector:
    def next_indices(self, X, selected_ixs, Y_selected, model, n_ixs=5):
        """
        Select next indices to label.
        Returns indices not yet in selected_ixs
        """
        all_ixs = set(range(len(X)))
        available_ixs = list(all_ixs - set(selected_ixs))
        if len(available_ixs) < n_ixs:
            return available_ixs
        return random.sample(available_ixs, n_ixs)

selector = SimpleSelector()

# Create a simple sklearn model
model = LogisticRegression(random_state=0, max_iter=1000)

ARGS = {
    'X': X,
    'Y': Y,
    'model': model,
    'selector': selector,
    'epochs': 3,
    'labels_per_epoch': 2,
    'bootstrap_size': 10,
    'batch_run': 1,
    'batch_info': 'test_batch'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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
bert_featurize Ran
ran on a synthesized input
melkherj/puddle/puddle/datasets/amazon_reviews_bert.py
code served (permissive licence) · get_code("436b1dc12d590080")
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
import torch.nn as nn

torch.manual_seed(0)

# Stand-in tokenizer class
class MockTokenizer:
    def tokenize(self, txt):
        # Return a list of tokens
        return ["[CLS]", "hello", "world", "[SEP]"]
    
    def convert_tokens_to_ids(self, tokens):
        # Return a list of token IDs
        token_to_id = {
            "[CLS]": 101,
            "hello": 7592,
            "world": 2088,
            "[SEP]": 102
        }
        return [token_to_id.get(t, 0) for t in tokens]

# Stand-in BERT model class
class MockBertModel(nn.Module):
    def __init__(self):
        super().__init__()
        # Model needs to accept tokens_tensor and segments_tensors
        # and return (encoded_layers, _) where encoded_layers is a list
        self.embedding = nn.Embedding(30522, 4)
    
    def forward(self, tokens_tensor, segments_tensors):
        # tokens_tensor shape: [1, seq_len]
        # segments_tensors shape: [1, seq_len]
        # Return (encoded_layers, _) where encoded_layers is a list of tensors
        seq_len = tokens_tensor.shape[1]
        # Create a dummy output with shape [1, seq_len, 4] (batch=1, seq=seq_len, hidden=4)
        output = torch.randn(1, seq_len, 4)
        # encoded_layers should be a list; we return a list with one element
        return [output], None

# Create instances
txt = "hello world"
bert_model = MockBertModel()
bert_tokenizer = MockTokenizer()

ARGS = {
    "txt": txt,
    "bert_model": bert_model,
    "bert_tokenizer": bert_tokenizer
}
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
check_size Ran
ran, honoured the contract we drafted
siit-vtt/semi-supervised-learning-pytorch/loader_cifar_zca.py
pointer only (licence: MIT); the stored copy get_code serves, from DebeshJha/ResUNetplusplus, is pointer only (licence: NONE) · get_code("9570e36a7f52ab5d")
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 DebeshJha/ResUNetplusplus, 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.

conv3x3 Ran
ran; our contract draft was wrong, not the code
siit-vtt/semi-supervised-learning-pytorch/preresnet_sd_cifar.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.

get_zca_normalization_param Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
perrying/realistic-ssl-evaluation-pytorch/build_dataset.py
code served (permissive licence) · get_code("1f390b46a93ced80")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) 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

np.random.seed(0)

# Create a small batch of images with shape (n_data, height, width, channels)
# Using distinct dimensions: 2 samples, 3x4 spatial, 5 channels
images = np.random.randn(2, 3, 4, 5).astype(np.float32)

ARGS = {
    'images': images,
    'scale': 0.1,
    'eps': 1e-10
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
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
horizontal_flip Ran
ran on a synthesized input
siit-vtt/semi-supervised-learning-pytorch/loader_cifar_zca.py
code served (permissive licence) · get_code("b96d4224eb3255fa")
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 3, dtype family int, shape (4, 3, 2) 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
import random

random.seed(0)

# Create a small image array with shape (height, width, channels)
# Using distinct dimensions: 4 height, 3 width, 2 channels
image = np.random.randint(0, 256, size=(4, 3, 2), dtype=np.uint8)

ARGS = {
    "image": image,
    "rate": 0.5
}
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 3, dtype family int, shape (4, 3, 2) — shape (4, 3, 2) 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
random_crop Ran
ran on a synthesized input
siit-vtt/semi-supervised-learning-pytorch/loader_cifar_zca.py
code served (permissive licence) · get_code("55f1499fcb95cc00")
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 3, dtype family int, shape (24, 24, 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 numpy as np
import random

# Set seed for reproducibility
random.seed(0)
np.random.seed(0)

# Create a sample image: height=32, width=32, channels=3 (RGB)
# Using distinct dimensions to allow later probing
sample_image = np.random.randint(0, 256, size=(32, 32, 3), dtype=np.uint8)

# Define ARGS dict with all parameters for random_crop
ARGS = {
    'image': sample_image,
    'crop_size': 24,
    'padding': 4
}
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 3, dtype family int, shape (24, 24, 3) — shape (24, 24, 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
run_experiment Ran
ran on a synthesized input
melkherj/puddle/puddle/active_evaluate.py
code served (permissive licence) · get_code("4f47977b925bb5e5")
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 DataFrame. 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
import random
from sklearn.linear_model import LogisticRegression

# Set seed for reproducibility
random.seed(0)
np.random.seed(0)

# Create a minimal selector class that exposes the required method
class MinimalSelector:
    def next_indices(self, X, selected_ixs, Y, model, n_ixs=5):
        """
        Return the next n_ixs indices to sample.
        Must avoid already selected indices.
        """
        available_ixs = [i for i in range(X.shape[0]) if i not in selected_ixs]
        return random.sample(available_ixs, min(n_ixs, len(available_ixs)))

# Create synthetic dataset
n_samples = 50
n_features = 4
X = np.random.randn(n_samples, n_features).astype(np.float32)
Y = np.random.randint(0, 2, n_samples).astype(bool)

# Ensure we have at least some positives and negatives for bootstrap
Y[0:10] = True
Y[10:20] = False

# Create dataset tuple
dataset = ('synthetic_data', (X, Y))

# Create a simple sklearn model
model = LogisticRegression(max_iter=100)
model_tuple = ('logistic_regression', model)

# Create selector
selector = MinimalSelector()
selector_tuple = ('random_selector', selector)

# Batch run identifier
batch_run = 1

# Create the experiment tuple
experiment = (dataset, model_tuple, selector_tuple, batch_run)

# Define ARGS dict with all parameters for run_experiment
ARGS = {
    'experiment': experiment,
    'epochs': 3,
    'labels_per_epoch': 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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
top_k Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
brain-research/realistic-ssl-evaluation/evaluate_model.py
code served (permissive licence) · get_code("4c46154099e4d408")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned ndarray, ndim 2, dtype family int, 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 numpy as np

# Create a 2D array of logits with shape (2, 6) - batch size 2, 6 classes
# Using distinct dimensions: 2 samples, 6 logits per sample
logits = np.array([
    [0.1, 0.5, 0.3, 0.9, 0.2, 0.7],
    [0.4, 0.8, 0.1, 0.6, 0.3, 0.5]
], dtype=np.float32)

# k is the number of top elements to retrieve
k = 3

ARGS = {
    'logits': logits,
    'k': k
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
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 2, dtype family int, 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
binary_class_metrics Not yet run
unverified (harvested, no recorded run)
melkherj/puddle/puddle/active_evaluate.py
code served (permissive licence) · get_code("05c1950f5c932d08")
How it ran

no invocation recorded.

gcn Not yet run
unverified (harvested, no recorded run)
perrying/realistic-ssl-evaluation-pytorch/build_dataset.py
code served (permissive licence) · get_code("20ec0d4caa9c361d")
How it ran

no invocation recorded.

sklearn_binarize_by_percentile Not yet run
unverified (harvested, no recorded run)
melkherj/puddle/puddle/datasets/sklearn_datasets.py
code served (permissive licence) · get_code("d035fc4df55fe056")
How it ran

no invocation recorded.

train_mt Not yet run
unverified (harvested, no recorded run)
DavidKo3/semi_supervised_mean_teacher/methods.py
code served (permissive licence) · get_code("ae84d93c214dfa6a")
How it ran

no invocation recorded.

train_pi Not yet run
unverified (harvested, no recorded run)
DavidKo3/semi_supervised_mean_teacher/methods.py
code served (permissive licence) · get_code("2250115852053551")
How it ran

no invocation recorded.

train_sup Not yet run
unverified (harvested, no recorded run)
DavidKo3/semi_supervised_mean_teacher/methods.py
code served (permissive licence) · get_code("915404fc2dd8ffd9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 4 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 4 later) and 11 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
46 dated earlier, 1 the same month, 639 later · 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
check_size 1 other paper
1 later
1911.07067 (2019-11)
ResUNet++: An Advanced Architecture for Medical Image Segmentation
gcn 1 other paper
1 later
1903.03825 (2019-03)
Interpolation Consistency Training for Semi-Supervised Learning
get_zca_normalization_param 1 other paper
1 later
1903.03825 (2019-03)
Interpolation Consistency Training for Semi-Supervised Learning

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

Semi-supervised learning (SSL) provides a powerful framework for leveraging unlabeled data when labels are limited or expensive to obtain. SSL algorithms based on deep neural networks have recently proven successful on standard benchmark tasks. However, we argue that these benchmarks fail to address many issues that these algorithms would face in real-world applications. After creating a unified reimplementation of various widely-used SSL techniques, we test them in a suite of experiments designed to address these issues. We find that the performance of simple baselines which do not use unlabeled data is often underreported, that SSL methods differ in sensitivity to the amount of labeled and unlabeled data, and that performance can degrade substantially when the unlabeled dataset contains out-of-class examples. To help guide SSL research towards real-world applicability, we make our unified reimplemention and evaluation platform publicly available.

For agents

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

get_harvested_code_for_paper("1804.09170")
get_code_for_paper("1804.09170")
have("1804.09170")

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

curl https://syntology.ai/api/ran/1804.09170.json

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

[![Syntology run record](https://syntology.ai/api/ran/1804.09170.svg)](https://syntology.ai/paper/1804.09170)

Connect an agent — have() is free.