SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2108.00981 · 2021

PSA-GAN: Progressive Self Attention GANs for Synthetic Time Series

arXiv · PDF · Open in the Atlas

Code that ran

9 of 11 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; 3 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/2108.00981.json · badge: /api/ran/2108.00981.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
awslabs/gluon-ts canonical (classifier label)
classifier basis: census
9 ran · 2 no recorded run
FunctionStatusWhere it lives
batcher Ran
ran; our contract draft was wrong, not the code
awslabs/gluon-ts/src/gluonts/itertools.py
code served (permissive licence) · get_code("9441ca1a63b924c3")
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 callable_iterator. 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 itertools
from typing import Callable, Collection, Dict, Iterable, Iterator, List, Optional, TypeVar, Sequence, Tuple, NamedTuple

torch.manual_seed(0)

# The batcher function takes an iterable and a batch_size
# It groups elements from the iterable into batches
# The iterable can be any iterable (list, string, generator, etc.)
# batch_size must be an integer

# Simple test with a list of integers
test_iterable = list(range(10))
test_batch_size = 3

ARGS = {
    "iterable": test_iterable,
    "batch_size": test_batch_size,
}
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 callable_iterator — a callable_iterator output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a callable_iterator 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
copy_with Ran
ran on a synthesized input
awslabs/gluon-ts/src/gluonts/util.py
code served (permissive licence) · get_code("b6d31d8f080218a4")
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 MyClass. 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
from dataclasses import dataclass

@dataclass
class MyClass:
    value: int
    name: str = "default"

# Create an instance to copy
obj = MyClass(value=42, name="original")

# Define kwargs to update the copy
kwargs = {
    "value": 100,
    "name": "updated"
}

ARGS = {
    "obj": obj,
    "kwargs": kwargs
}
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 MyClass — a MyClass output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MyClass 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
equals_default_impl Ran
ran on a synthesized input
awslabs/gluon-ts/src/gluonts/core/component.py
code served (permissive licence) · get_code("900b826876f774fb")
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 numpy as np

# Test case 1: Two objects with matching types and __init_args__
class ValidatedClass:
    def __init__(self, value):
        self.__init_args__ = (value,)
        self.value = value

# Test case 2: Two objects with matching types and __init_passed_kwargs__
class ValidatedClassKwargs:
    def __init__(self, x=1, y=2):
        self.__init_passed_kwargs__ = {'x': x, 'y': y}
        self.x = x
        self.y = y

# Test case 3: Simple objects without validation attributes
class SimpleClass:
    def __init__(self, val):
        self.val = val
    
    def __eq__(self, other):
        if not isinstance(other, SimpleClass):
            return False
        return self.val == other.val

# Create test objects
obj1 = ValidatedClass(42)
obj2 = ValidatedClass(42)

obj3 = ValidatedClassKwargs(x=10, y=20)
obj4 = ValidatedClassKwargs(x=10, y=20)

simple1 = SimpleClass(100)
simple2 = SimpleClass(100)

ARGS = {
    'this': obj1,
    'that': obj2,
}
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 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
flatten Ran
ran on a synthesized input
awslabs/gluon-ts/src/gluonts/maybe.py
code served (permissive licence) · get_code("4cd8c833a319daa5")
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
from typing import Optional

torch.manual_seed(0)

# The flatten function takes a single parameter 'val' which is Optional[Optional[T]]
# The function simply returns the value as-is, changing only the type annotation
# So we need to provide a value that could be Optional[Optional[T]]

# Test with a simple value (int wrapped in two levels of Optional)
val = 42

ARGS = {
    "val": val
}
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
fqname_for Ran
ran on a synthesized input
awslabs/gluon-ts/src/gluonts/core/_base.py
code served (permissive licence) · get_code("f5f4848418aecf04")
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

class SimpleModule(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)

# Test with a simple built-in class
obj = SimpleModule

ARGS = {
    'obj': obj
}
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
inject Ran
ran; our contract draft was wrong, not the code
awslabs/gluon-ts/src/gluonts/core/settings.py
code served (permissive licence) · get_code("2ce2e172f2056955")
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 torch.nn as nn

torch.manual_seed(0)

class MockSettings:
    """Stand-in settings class with _inject method"""
    def _inject(self, *args, **kwargs):
        """Mock inject method that returns args and kwargs"""
        return {
            'args': args,
            'kwargs': kwargs,
            'message': 'injection successful'
        }

# Create a settings object
settings = MockSettings()

# Define arguments for the inject function
ARGS = {
    'settings': settings,
}
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 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
let Ran
ran; our contract draft was wrong, not the code
awslabs/gluon-ts/src/gluonts/core/settings.py
code served (permissive licence) · get_code("d627f142e468c86e")
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 MockSettings. 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)

class MockSettings:
    """Mock settings object with _let method"""
    def __init__(self):
        self.config = {}
        self.model = None
        self.data = []
    
    def _let(self, **kwargs):
        """Simple implementation that updates internal state"""
        for key, value in kwargs.items():
            setattr(self, key, value)
        return self

# Create a settings object
settings = MockSettings()

ARGS = {
    "settings": settings,
}
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 MockSettings — a MockSettings output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MockSettings 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
maybe_len Ran
ran, honoured the contract we drafted
awslabs/gluon-ts/src/gluonts/itertools.py
code served (permissive licence) · get_code("2a1a76b9217ce9a7")
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
import numpy as np

torch.manual_seed(0)

# The function maybe_len attempts to call len() on an object
# and returns the length if successful, or None if it raises
# NotImplementedError or AttributeError

# Test with various objects that have different len() behaviors

# Object with a valid __len__ method
list_obj = [1, 2, 3, 4, 5]

# Object without __len__ (will return None)
class NoLen:
    pass

no_len_obj = NoLen()

# Tensor (has __len__)
tensor_obj = torch.randn(2, 4)

# String (has __len__)
string_obj = "hello"

# Dict (has __len__)
dict_obj = {"a": 1, "b": 2}

# Numpy array (has __len__)
numpy_obj = np.array([1, 2, 3])

# We need to test the function with an object
# The function takes a single parameter: obj
# We'll use a list as a basic valid test case

ARGS = {
    "obj": [1, 2, 3, 4, 5]
}
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 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
prod Ran
ran, violated the contract we drafted
awslabs/gluon-ts/src/gluonts/itertools.py
code served (permissive licence) · get_code("67059eb6565db923")
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
import numpy as np

torch.manual_seed(0)

# The function prod() takes an iterable and computes the product of its elements
# It supports any iterable where elements can be multiplied together

# Test with different types of iterables:

# Option 1: List of integers
xs_int = [2, 3, 4, 5]

# Option 2: List of floats
xs_float = [1.5, 2.0, 3.5]

# Option 3: NumPy array
xs_numpy = np.array([2, 3, 4])

# Option 4: Torch tensor (1D)
xs_tensor = torch.tensor([2, 3, 4, 5])

# Option 5: Mixed numeric types in a list
xs_mixed = [2, 3.0, 4]

# Using a simple list of integers as the primary test case
ARGS = {
    "xs": [2, 3, 4, 5]
}
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 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
equals Not yet run
unverified (harvested, no recorded run)
awslabs/gluon-ts/src/gluonts/core/component.py
code served (permissive licence) · get_code("2f227eb18785b23e")
How it ran

no invocation recorded.

tqdm Not yet run
unverified (harvested, no recorded run)
awslabs/gluon-ts/src/gluonts/gluonts_tqdm.py
code served (permissive licence) · get_code("03c6c7d9c6ace85c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 11 are also present in other papers whose code we hold (8 with at least one paper dated earlier, 0 the same month, 11 later) and 0 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
batcher 4 other papers
2 dated earlier, 2 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
1910.03002 (2019-10)
High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes
2112.14436 (2021-12)
Monte Carlo EM for Deep Time Series Anomaly Detection Franc ¸ois
and 1 more not named here
equals 4 other papers
2 dated earlier, 2 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
1910.03002 (2019-10)
High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes
2112.14436 (2021-12)
Monte Carlo EM for Deep Time Series Anomaly Detection Franc ¸ois
and 1 more not named here
fqname_for 4 other papers
2 dated earlier, 2 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
1910.03002 (2019-10)
High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes
2112.14436 (2021-12)
Monte Carlo EM for Deep Time Series Anomaly Detection Franc ¸ois
and 1 more not named here
inject 4 other papers
2 dated earlier, 2 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
1910.03002 (2019-10)
High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes
2112.14436 (2021-12)
Monte Carlo EM for Deep Time Series Anomaly Detection Franc ¸ois
and 1 more not named here
let 4 other papers
2 dated earlier, 2 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
1910.03002 (2019-10)
High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes
2112.14436 (2021-12)
Monte Carlo EM for Deep Time Series Anomaly Detection Franc ¸ois
and 1 more not named here
tqdm 4 other papers
2 dated earlier, 2 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
1910.03002 (2019-10)
High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes
2112.14436 (2021-12)
Monte Carlo EM for Deep Time Series Anomaly Detection Franc ¸ois
and 1 more not named here
maybe_len 3 other papers
1 dated earlier, 2 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
2112.14436 (2021-12)
Monte Carlo EM for Deep Time Series Anomaly Detection Franc ¸ois
2207.09572 (2022-07)
Robust Multivariate Time-Series Forecasting: Adversarial Attacks and Defense Mechanisms
prod 3 other papers
2 dated earlier, 1 later
1906.05264 (2019-06)
GluonTS: Probabilistic Time Series Models in Python
1911.10416 (2019-11)
Intermittent Demand Forecasting with Deep Renewal Processes
2207.09572 (2022-07)
Robust Multivariate Time-Series Forecasting: Adversarial Attacks and Defense Mechanisms
copy_with 1 other paper
1 later
2207.09572 (2022-07)
Robust Multivariate Time-Series Forecasting: Adversarial Attacks and Defense Mechanisms
equals_default_impl 1 other paper
1 later
2207.09572 (2022-07)
Robust Multivariate Time-Series Forecasting: Adversarial Attacks and Defense Mechanisms
flatten 1 other paper
1 later
2207.09572 (2022-07)
Robust Multivariate Time-Series Forecasting: Adversarial Attacks and Defense Mechanisms

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

Realistic synthetic time series data of sufficient length enables practical applications in time series modeling tasks, such as forecasting, but remains a challenge. In this paper we present PSA-GAN, a generative adversarial network (GAN) that generates long time series samples of high quality using progressive growing of GANs and self-attention. We show that PSA-GAN can be used to reduce the error in two downstream forecasting tasks over baselines that only use real data. We also introduce a Frechet-Inception Distance-like score, Context-FID, assessing the quality of synthetic time series samples. In our downstream tasks, we find that the lowest scoring models correspond to the best-performing ones. Therefore, Context-FID could be a useful tool to develop time series GAN models.

For agents

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

get_harvested_code_for_paper("2108.00981")
get_code_for_paper("2108.00981")
have("2108.00981")

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

curl https://syntology.ai/api/ran/2108.00981.json

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

[![Syntology run record](https://syntology.ai/api/ran/2108.00981.svg)](https://syntology.ai/paper/2108.00981)

Connect an agent — have() is free.