SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1609.03499 · 2016

WAVENET: A GENERATIVE MODEL FOR RAW AUDIO Aäron van den Oord

Sander Dieleman, Karen Simonyan, Oriol Vinyals, Nal Kalchbrenner, Heiga Zen, Andrew Senior, Koray Kavukcuoglu, Alex Graves, Aaron van den Oord

arXiv · PDF · Open in the Atlas

Code that ran

65 of 103 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (23 of them constructed an object rather than computing a result; 3 honoured the contract we drafted, 1 violated it and 47 have no contract verdict; 14 failures of our instrument; 38 no recorded run), measured 2026-11-21; 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/1609.03499.json · badge: /api/ran/1609.03499.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
otosense/slang community (classifier label)
classifier basis not recorded on the link
9 ran · 3 no recorded run
rampage644/wavenet community (classifier label)
classifier basis not recorded on the link
9 ran · 0 no recorded run
swasun/VQ-VAE-Speech community
not in the paper's declared repositories
7 ran · 4 no recorded run
LucaHermes/lightweight-motion-forecasting community
Papers with Code archive
5 ran · 2 no recorded run
outofculture/talk-like-me community (classifier label)
classifier basis not recorded on the link
5 ran · 1 no recorded run
isadrtdinov/wavenet community
not in the paper's declared repositories
3 ran · 1 no recorded run
vicky-hnk/time-flex community
not in the paper's declared repositories
3 ran · 1 no recorded run
ShichengChen/WaveNetSeparateAudio community
Papers with Code archive
2 ran · 6 no recorded run
ibab/tensorflow-wavenet community (classifier label)
classifier basis not recorded on the link
2 ran · 3 no recorded run
AI-Huang/WaveNet community
Papers with Code archive
2 ran · 1 no recorded run
benmoseley/simple-wavenet community
Papers with Code archive
2 ran · 1 no recorded run
glakshay/Generating-audio-DL community
not in the paper's declared repositories
2 ran · 0 no recorded run
Gal1eo/DT2119 community
Papers with Code archive
2 ran · 0 no recorded run
zhong110020/keras-tcn community
not in the paper's declared repositories
2 ran · 0 no recorded run
zhong110020/Tensorflow-TCN community
not in the paper's declared repositories
2 ran · 0 no recorded run
peustr/wavenet community
not in the paper's declared repositories
2 ran · 0 no recorded run
PhilippeNguyen/keras_wavenet community
Papers with Code archive
1 ran · 4 no recorded run
ShotDownDiane/tcn-master community
Papers with Code archive
1 ran · 2 no recorded run
anandharaju/Basic_TCN community
Papers with Code archive
1 ran · 1 no recorded run
DevonFulcher/CryptoPricePredictor community
Papers with Code archive
1 ran · 1 no recorded run
Salazar-99/Gravitational-WaveNet community
Papers with Code archive
1 ran · 0 no recorded run
stdereka/liverpool-ion-switching community
not in the paper's declared repositories
1 ran · 0 no recorded run
PeihaoChen/regnet community
Papers with Code archive
0 ran · 2 no recorded run
yebiny/DepthOfAnaesthesia_eeg community
not in the paper's declared repositories
0 ran · 1 no recorded run
MSRDL/Deep4Cast community
Papers with Code archive
0 ran · 1 no recorded run
FunctionStatusWhere it lives
AddEncoder Ran
ran on a synthesized input · constructed an object; no result was computed
PhilippeNguyen/keras_wavenet/keras_wavenet/models/wavenet.py
pointer only (licence: NONE) · get_code("9382ada555ac8131")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, PhilippeNguyen/keras_wavenet 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.

BlockLayer Ran
ran on a synthesized input
glakshay/Generating-audio-DL/wavenet_model.py
pointer only (licence: NONE) · get_code("0d5b595bd813f5d2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, glakshay/Generating-audio-DL 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.

CausalConv1d Ran
ran on a synthesized input · constructed an object; no result was computed
isadrtdinov/wavenet/wavenet/model/wavenet.py
code served (permissive licence) · get_code("5170661e509fdac5")
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 an instance of its own class, CausalConv1d: the run constructed the object and did not call it, so no result was computed. 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)

# CausalConv1d expects:
# - in_channels: int, number of input channels
# - out_channels: int, number of output channels
# - kernel: int, kernel size
# - dilation: int, dilation factor (optional, defaults to 1)

ARGS = {
    "in_channels": 4,
    "out_channels": 3,
    "kernel": 5,
    "dilation": 1
}
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 CausalConv1d — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a CausalConv1d 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
ComplexAdaptiveAvgPool1d Ran
ran on a synthesized input · constructed an object; no result was computed
vicky-hnk/time-flex/src/models/wave_net.py
pointer only (licence: NONE) · get_code("79f4a42767095c68")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, vicky-hnk/time-flex 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.

ComplexConv1d Ran
ran on a synthesized input · constructed an object; no result was computed
vicky-hnk/time-flex/src/models/wave_net.py
pointer only (licence: NONE) · get_code("aeed3dc3c228cb85")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, vicky-hnk/time-flex 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.

ComplexDropout Ran
ran on a synthesized input · constructed an object; no result was computed
vicky-hnk/time-flex/src/models/wave_net.py
pointer only (licence: NONE) · get_code("d48c7052e4342b63")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, vicky-hnk/time-flex 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.

Conv Ran
ran on a synthesized input
Gal1eo/DT2119/WaveNet.py
code served (permissive licence) · get_code("2a41b9a5b5561707")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 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
import torch.nn as nn
import torch.nn.functional as F

class Conv(nn.Module):
    """
    A convolution with the option to be causal and use xavier initialization
    """

    def __init__(self, input_channels, output_channels, kernel_size=1, stride=1, dilation=1, bias=True, init_gain='linear', is_causal=False):
        super(Conv, self).__init__()
        self.is_causal = is_causal
        self.kernel_size = kernel_size
        self.dilation = dilation
        self.conv = nn.Conv1d(input_channels, output_channels, kernel_size=kernel_size, stride=stride, dilation=dilation, bias=bias)
        nn.init.xavier_uniform_(self.conv.weight, gain=torch.nn.init.calculate_gain(init_gain))

    def forward(self, signal):
        if self.is_causal:
            padding = (int((self.kernel_size - 1) * self.dilation), 0)
            signal = F.pad(signal, padding)
        return self.conv(signal)


INIT_ARGS = {
    'input_channels': 4,
    'output_channels': 3,
    'kernel_size': 3,
    'stride': 1,
    'dilation': 1,
    'bias': True,
    'init_gain': 'linear',
    'is_causal': False,
}

ARGS = {
    'signal': torch.randn(2, 4, 5),
}

if __name__ == '__main__':
    conv = Conv(**INIT_ARGS)
    output = conv(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {output}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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, 3, 3) — shape (2, 3, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
ConvTranspose2d Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT) · constructed an object; no result was computed
swasun/VQ-VAE-Speech/src/wavenet_vocoder/wavenet.py
code served (permissive licence) · get_code("3579d050a889047c")
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 an instance of its own class, ConvTranspose2d: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    'in_channels': 3,
    'out_channels': 16,
    'kernel_size': (4, 5),
    'weight_normalization': True,
}
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 ConvTranspose2d — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ConvTranspose2d 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
Embedding Ran
ran; our contract draft was wrong, not the code · constructed an object; no result was computed
swasun/VQ-VAE-Speech/src/wavenet_vocoder/wavenet.py
code served (permissive licence) · get_code("d792a0fba4201cf8")
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 an instance of its own class, Embedding: the run constructed the object and did not call it, so no result was computed. 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)

ARGS = {
    "num_embeddings": 10,
    "embedding_dim": 4,
    "padding_idx": 0,
    "std": 0.01,
}
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 Embedding — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Embedding 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
EncoderBase Ran
ran on a synthesized input · constructed an object; no result was computed
LucaHermes/lightweight-motion-forecasting/models/encoders.py
pointer only (licence: NONE) · get_code("e64b73bbdaad21c1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LucaHermes/lightweight-motion-forecasting 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.

GWN Ran
ran on a synthesized input · constructed an object; no result was computed
Salazar-99/Gravitational-WaveNet/models.py
pointer only (licence: NONE) · get_code("d11e91181dcc9a07")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Salazar-99/Gravitational-WaveNet 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.

GraphWavenetBlock Ran
ran on a synthesized input · constructed an object; no result was computed
LucaHermes/lightweight-motion-forecasting/models/encoders.py
pointer only (licence: NONE) · get_code("52e40ac0e567678c")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LucaHermes/lightweight-motion-forecasting 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.

GraphWavenetBlockOrig Ran
ran on a synthesized input · constructed an object; no result was computed
LucaHermes/lightweight-motion-forecasting/models/encoders.py
pointer only (licence: NONE) · get_code("177000aa16ed0914")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LucaHermes/lightweight-motion-forecasting 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.

ResidualBlock Ran
ran on a synthesized input · constructed an object; no result was computed
anandharaju/Basic_TCN/tcn/tcn.py
code served (permissive licence) · get_code("507c33200751b432")
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 an instance of its own class, ResidualBlock: the run constructed the object and did not call it, so no result was computed. 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 tensorflow as tf
from tensorflow.keras import backend as K
from tensorflow.keras.layers import Layer, Conv1D, Dense, BatchNormalization, LayerNormalization, Activation, SpatialDropout1D, Lambda
import numpy as np

ARGS = {
    'dilation_rate': 2,
    'nb_filters': 16,
    'kernel_size': 3,
    'padding': 'same',
    'activation': 'relu',
    'dropout_rate': 0.1,
    'kernel_initializer': 'he_normal',
    'use_batch_norm': False,
    'use_layer_norm': False,
    'last_block': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ResidualBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ResidualBlock 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
TCN Ran
ran on a synthesized input · constructed an object; no result was computed
DevonFulcher/CryptoPricePredictor/tcn/tcn.py
code served (permissive licence) · get_code("ab75326a9401df22")
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 an instance of its own class, TCN: the run constructed the object and did not call it, so no result was computed. 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 keras
from keras.layers import Input
import numpy as np

# Create a minimal Keras model input tensor to work with
input_tensor = Input(shape=(10, 4))  # (timesteps=10, input_dim=4)

ARGS = {
    'nb_filters': 64,
    'kernel_size': 2,
    'nb_stacks': 1,
    'dilations': [1, 2, 4, 8],
    'padding': 'causal',
    'use_skip_connections': True,
    'dropout_rate': 0.1,
    'return_sequences': False,
    'name': 'tcn_model'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
keras, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type TCN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a TCN 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
TCN Ran
ran on a synthesized input · constructed an object; no result was computed
zhong110020/keras-tcn/tcn/tcn.py
code served (permissive licence) · get_code("aa8c3981092c6563")
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 an instance of its own class, TCN: the run constructed the object and did not call it, so no result was computed. 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 keras
from keras.layers import Input
import numpy as np

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

# Create a minimal input tensor for testing
# Shape: (batch_size, timesteps, input_dim)
input_tensor = Input(shape=(10, 4))

# Define ARGS dict with all TCN parameters
ARGS = {
    'nb_filters': 64,
    'kernel_size': 2,
    'nb_stacks': 1,
    'dilations': [1, 2, 4, 8],
    'padding': 'causal',
    'use_skip_connections': True,
    'dropout_rate': 0.1,
    'return_sequences': False,
    'activation': 'linear',
    'name': 'tcn_test'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
keras, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type TCN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a TCN 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
TEConvolution Ran
ran on a synthesized input · constructed an object; no result was computed
LucaHermes/lightweight-motion-forecasting/models/encoders.py
pointer only (licence: NONE) · get_code("8816614f0f19222e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LucaHermes/lightweight-motion-forecasting 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.

TemporalBlock Ran
ran on a synthesized input · constructed an object; no result was computed
zhong110020/Tensorflow-TCN/tcn.py
code served (permissive licence) · get_code("98dbe4ae7dc18a75")
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 an instance of its own class, TemporalBlock: the run constructed the object and did not call it, so no result was computed. 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 tensorflow as tf

# Create the TemporalBlock class
layers = tf.keras.layers

class TemporalBlock(tf.keras.Model):
    def __init__(self, dilation_rate, nb_filters, kernel_size, padding, dropout_rate=0.0):
        super(TemporalBlock, self).__init__()
        init = tf.keras.initializers.RandomNormal(mean=0.0, stddev=0.01)
        assert padding in ['causal', 'same']
        self.conv1 = layers.Conv1D(filters=nb_filters, kernel_size=kernel_size, dilation_rate=dilation_rate, padding=padding, kernel_initializer=init)
        self.batch1 = layers.BatchNormalization(axis=-1)
        self.ac1 = layers.Activation('relu')
        self.drop1 = layers.Dropout(rate=dropout_rate)
        self.conv2 = layers.Conv1D(filters=nb_filters, kernel_size=kernel_size, dilation_rate=dilation_rate, padding=padding, kernel_initializer=init)
        self.batch2 = layers.BatchNormalization(axis=-1)
        self.ac2 = layers.Activation('relu')
        self.drop2 = layers.Dropout(rate=dropout_rate)
        self.downsample = layers.Conv1D(filters=nb_filters, kernel_size=1, padding='same', kernel_initializer=init)
        self.ac3 = layers.Activation('relu')

    def call(self, x, training):
        prev_x = x
        x = self.conv1(x)
        x = self.batch1(x)
        x = self.ac1(x)
        x = self.drop1(x) if training else x
        x = self.conv2(x)
        x = self.batch2(x)
        x = self.ac2(x)
        x = self.drop2(x) if training else x
        if prev_x.shape[-1] != x.shape[-1]:
            prev_x = self.downsample(prev_x)
        assert prev_x.shape == x.shape
        return self.ac3(prev_x + x)

# Build arguments
ARGS = {
    'dilation_rate': 1,
    'nb_filters': 16,
    'kernel_size': 3,
    'padding': 'causal',
    'dropout_rate': 0.2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type TemporalBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a TemporalBlock 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
TemporalConvNet Ran
ran on a synthesized input · constructed an object; no result was computed
zhong110020/Tensorflow-TCN/tcn.py
code served (permissive licence) · get_code("a082de3ab11812a2")
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 an instance of its own class, TemporalConvNet: the run constructed the object and did not call it, so no result was computed. 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 tensorflow as tf
import numpy as np

tf.keras.backend.set_floatx('float32')

# Create a small input tensor for testing
# Shape: (batch_size, sequence_length, input_channels)
input_tensor = tf.constant(np.random.randn(2, 8, 4).astype(np.float32))

# Define arguments for TemporalConvNet
ARGS = {
    'num_channels': [16, 32, 64],
    'kernel_size': 2,
    'dropout': 0.2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type TemporalConvNet — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a TemporalConvNet 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
WaveNet Ran
ran on a synthesized input · constructed an object; no result was computed
isadrtdinov/wavenet/wavenet/model/wavenet.py
code served (permissive licence) · get_code("4e167db5b9f6fc08")
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 an instance of its own class, WaveNet: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

# Build the WaveNet model with minimal parameters
ARGS = {
    "num_mels": 80,
    "num_quants": 256,
    "win_length": 1024,
    "hop_length": 256,
    "upsample_kernel": 800,
    "residual_channels": 120,
    "skip_channels": 240,
    "causal_kernel": 2,
    "num_blocks": 16,
    "dilation_cycle": 8,
}
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 WaveNet — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a WaveNet 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
WaveNet Ran
ran on a synthesized input · constructed an object; no result was computed
glakshay/Generating-audio-DL/wavenet_model.py
pointer only (licence: NONE) · get_code("09786ed59e7e9cf2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, glakshay/Generating-audio-DL 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.

WaveNetBlock Ran
ran on a synthesized input · constructed an object; no result was computed
isadrtdinov/wavenet/wavenet/model/wavenet.py
code served (permissive licence) · get_code("65c6a6bb03ab096a")
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 an instance of its own class, WaveNetBlock: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

# The function signature is:
# def __init__(self, num_mels=80, causal_kernel=2, causal_dilation=1, residual_channels=120, skip_channels=240)

# Looking at the forward method, we need:
# - inputs: tensor of shape (batch, residual_channels, time_steps)
# - conditions: tensor of shape (batch, num_mels, time_steps)

# We'll use small dimensions for testing
batch_size = 2
num_mels = 5
residual_channels = 4
skip_channels = 6
causal_kernel = 2
causal_dilation = 1
time_steps = 8

ARGS = {
    "num_mels": num_mels,
    "causal_kernel": causal_kernel,
    "causal_dilation": causal_dilation,
    "residual_channels": residual_channels,
    "skip_channels": skip_channels,
}
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 WaveNetBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a WaveNetBlock 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
WaveNetBlock Ran
ran on a synthesized input · constructed an object; no result was computed
AI-Huang/WaveNet/wavenet/torch_fn/wavenet.py
code served (permissive licence) · get_code("f89017533ce3657d")
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 an instance of its own class, WaveNetBlock: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "in_channels": 4,
    "out_channels": 8,
    "kernel_size": 3,
    "n": 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 WaveNetBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a WaveNetBlock 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
WaveNetLayer Ran
ran on a synthesized input · constructed an object; no result was computed
AI-Huang/WaveNet/wavenet/torch_fn/wavenet.py
code served (permissive licence) · get_code("89203853d074962e")
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 an instance of its own class, WaveNetLayer: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "in_channels": 4,
    "out_channels": 8,
    "kernel_size": 3,
    "dilation": 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 WaveNetLayer — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a WaveNetLayer 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
WaveNetModel Ran
ran on a synthesized input
Gal1eo/DT2119/WaveNet.py
code served (permissive licence) · get_code("c45b9649408ac9fe")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import math

INIT_ARGS = {
    'input_channels': 8,
    'layers': 3,
    'max_dilation': 4,
    'residual_channels': 6,
    'skip_channels': 5,
    'output_channels': 8,
    'cond_channels': 7,
    'upsample_window': 4,
    'upsample_stride': 2
}

# Create a batch of input indices for embedding
# Shape: (batch_size, sequence_length)
batch_size = 2
sequence_length = 10
input_tensor = torch.randint(0, INIT_ARGS['input_channels'], (batch_size, sequence_length))

ARGS = {
    'forward_input': [input_tensor]
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
WaveNetResidualConv1D Ran
ran; our contract draft was wrong, not the code
peustr/wavenet/wavenet/model.py
pointer only (licence: NONE) · get_code("279056e98d5e0375")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, peustr/wavenet 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.

Wavenet1D Ran
ran on a synthesized input · constructed an object; no result was computed
benmoseley/simple-wavenet/wavenet.py
code served (permissive licence) · get_code("6b4b9533fe39cda6")
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 an instance of its own class, Wavenet1D: the run constructed the object and did not call it, so no result was computed. 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 tensorflow as tf
import numpy as np

ARGS = {
    'in_channels': 3,
    'filter_width': 2,
    'num_blocks': 2,
    'num_layers': 2,
    'hidden_channels': 4,
    'rates': None,
    'activation': tf.nn.relu,
    'biases': True,
    'verbose': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Wavenet1D — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Wavenet1D 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
_conv1x1_forward Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
swasun/VQ-VAE-Speech/src/wavenet_vocoder/wavenet.py
code served (permissive licence) · get_code("72c1ea6847fe785b")
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, 3, 5) 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)

class Conv1x1Module(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv = nn.Conv1d(4, 3, kernel_size=1)
    
    def forward(self, x):
        return self.conv(x)
    
    def incremental_forward(self, x):
        return self.conv(x)

conv = Conv1x1Module()
x = torch.randn(2, 4, 5)

ARGS = {
    "conv": conv,
    "x": x,
    "is_incremental": False
}
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, 3, 5) — shape (2, 3, 5) 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
_expand_global_features Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
swasun/VQ-VAE-Speech/src/wavenet_vocoder/wavenet.py
code served (permissive licence) · get_code("3f851150ab8a5983")
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, 5) 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)

# Construct arguments for _expand_global_features
B = 2  # batch size
T = 5  # time length
C = 4  # channels

# Create a global features tensor with shape (B x C)
g = torch.randn(B, C)

# Test with bct=True (returns B x C x T format)
bct = True

ARGS = {
    "B": B,
    "T": T,
    "g": g,
    "bct": bct,
}
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, 5) — shape (2, 4, 5) 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
audio_data_to_windows_of_normalized_ffts Ran
ran on a synthesized input
outofculture/talk-like-me/audio.py
code served (permissive licence) · get_code("7b74db674fb72fdf")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

from math import ceil
import numpy as np
from scipy.signal import stft, istft

sample_rate = 16000
lowest_freq = 20
window_size = 1.0 / lowest_freq
samples_per_window = int(ceil(sample_rate * window_size))
samples_per_step = int(ceil(sample_rate * 0.01))

# Create a small synthetic audio signal
# Duration: ~0.5 seconds to keep it manageable
duration = 0.5
num_samples = int(sample_rate * duration)
data = np.sin(2 * np.pi * 440 * np.linspace(0, duration, num_samples)).astype(np.float32)

ARGS = {
    "data": data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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
b Ran
ran; our contract draft was wrong, not the code
benmoseley/simple-wavenet/wavenet.py
code served (permissive licence) · get_code("4881eb48ec733168")
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 ResourceVariable, ndim 2, dtype family float, shape (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 tensorflow as tf

# The function b() takes:
# - shape: a tuple/list defining the shape of the bias variable
# - const: a float constant to initialize the bias with (default 0.1)
# - name: optional string name for the variable (default None)

# Construct valid arguments
shape = (4, 3)  # Small 2D shape for the bias
const = 0.1
name = "bias_layer"

ARGS = {
    "shape": shape,
    "const": const,
    "name": name
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ResourceVariable, ndim 2, dtype family float, shape (4, 3) — shape (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
binarize Ran
ran on a synthesized input
rampage644/wavenet/wavenet/utils.py
code served (permissive licence) · get_code("3a12c3886a1fc338")
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 2, dtype family int, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small numpy array with values in [0, 1]
# Using shape (2, 4) - batch size 2, 4 features
images = np.array([
    [0.1, 0.5, 0.9, 0.3],
    [0.7, 0.2, 0.8, 0.4]
], dtype=np.float32)

ARGS = {
    'images': images,
    'xp': np
}
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 2, dtype family int, shape (2, 4) — shape (2, 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
build_wavenet_model Ran
ran; our contract draft was wrong, not the code
peustr/wavenet/wavenet/model.py
pointer only (licence: NONE) · get_code("143bb37d58864fbb")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, peustr/wavenet 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.

complex_to_polar Ran
ran on a synthesized input
outofculture/talk-like-me/torchit.py
code served (permissive licence) · get_code("64f12982e6a585b9")
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 float, shape (2, 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

# Create a small complex-valued numpy array
# Using shape (2, 3) - batch size 2, feature dimension 3
x = np.array([
    [1+2j, 3+4j, 5+6j],
    [7+8j, 9+10j, 11+12j]
], dtype=np.complex128)

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 3, dtype family float, shape (2, 3, 2) — shape (2, 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
convert_to_patch Ran
ran on a synthesized input
rampage644/wavenet/server.py
code served (permissive licence) · get_code("9c3ef54612737736")
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 numpy as np

# Analyze the function:
# - xt: used with np.append(xt, xt[::-1]), so needs to be array-like and reversible
# - y_lower: used with np.append(y_lower, y_upper), so needs to be array-like
# - y_upper: used with np.append(y_lower, y_upper), so needs to be array-like

# Create small numpy arrays with distinct sizes for clarity
xt = np.array([1.0, 2.0, 3.0, 4.0])  # 1D array of size 4
y_lower = np.array([0.5, 1.0, 1.5, 2.0])  # 1D array of size 4
y_upper = np.array([1.5, 2.0, 2.5, 3.0])  # 1D array of size 4

ARGS = {
    'xt': xt,
    'y_lower': y_lower,
    'y_upper': y_upper
}
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 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
convert_to_rgb Ran
ran on a synthesized input
rampage644/wavenet/wavenet/utils.py
code served (permissive licence) · get_code("d128a002ad6e4cf7")
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 4, dtype family float, shape (2, 3, 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 numpy as np

# Create a small batch of grayscale images (batch_size=2, channels=1, height=4, width=3)
images = np.random.rand(2, 1, 4, 3).astype(np.float32)

ARGS = {
    "images": images,
    "xp": np,
}
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 4, dtype family float, shape (2, 3, 4, 3) — shape (2, 3, 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_embedding_table Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ShichengChen/WaveNetSeparateAudio/archives/wavenet/model.py
code served (permissive licence) · get_code("98a4f58cb03b805d")
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 ResourceVariable, 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 numpy as np
import tensorflow as tf

# Case 1: Square matrix (identity initialization)
name_1 = "embedding_table_square"
shape_1 = (4, 4)

# Case 2: Non-square matrix (Xavier initialization via create_variable)
name_2 = "embedding_table_rect"
shape_2 = (4, 3)

# Test with square shape first
ARGS = {
    "name": name_1,
    "shape": shape_1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ResourceVariable, 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
decreasing_integer_geometric_sequence Ran
ran on a synthesized input
otosense/slang/slang/spectrop.py
code served (permissive licence) · get_code("44d903b3f04aa30a")
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 int; 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 numpy as np

# Call torch.manual_seed for reproducibility
torch.manual_seed(0)

# Construct arguments for decreasing_integer_geometric_sequence
ARGS = {
    'start': 1025,  # DFLT_INPUT_SIZE = int(1 + 2048/2) = 1025
    'scale_factor': 0.5
}
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 list, ndim 1, dtype family int — 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
find_files Ran
ran on a synthesized input
ibab/tensorflow-wavenet/wavenet/audio_reader.py
code served (permissive licence) · get_code("605a3067a4dbc087")
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 tempfile
import os

# Create a temporary directory with some test files
temp_dir = tempfile.mkdtemp()

# Create some subdirectories and files for testing
os.makedirs(os.path.join(temp_dir, "subdir1"), exist_ok=True)
os.makedirs(os.path.join(temp_dir, "subdir2"), exist_ok=True)

# Create some .wav files
open(os.path.join(temp_dir, "file1.wav"), "w").close()
open(os.path.join(temp_dir, "subdir1", "file2.wav"), "w").close()
open(os.path.join(temp_dir, "subdir2", "file3.wav"), "w").close()

# Create some non-.wav files that should be ignored
open(os.path.join(temp_dir, "file.txt"), "w").close()
open(os.path.join(temp_dir, "subdir1", "file.mp3"), "w").close()

ARGS = {
    "directory": temp_dir,
    "pattern": "*.wav"
}
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 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
gated_activation Ran
ran, honoured the contract we drafted
LucaHermes/lightweight-motion-forecasting/models/encoders.py
pointer only (licence: NONE) · get_code("c178be7285ad41cc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, LucaHermes/lightweight-motion-forecasting 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 Ran
ran on a synthesized input
rampage644/wavenet/server.py
code served (permissive licence) · get_code("1b7121b9a228766d")
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 2, dtype family float, shape (1, 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

# Initialize the dataseries dict with some test data
dataseries = {
    'key1': np.array([[1.0, 2.0], [3.0, 4.0]]),
    'key2': np.array([[5.0, 6.0]]),
}

# Define arguments for the get function
ARGS = {
    'key': 'key1'
}
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 2, dtype family float, shape (1, 2) — shape (1, 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
get_category_cardinality Ran
ran on a synthesized input
ibab/tensorflow-wavenet/wavenet/audio_reader.py
code served (permissive licence) · get_code("931b3cb69e4b92b0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family int; 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 re

FILE_PATTERN = 'p([0-9]+)_([0-9]+)\\.wav'

def get_category_cardinality(files):
    id_reg_expression = re.compile(FILE_PATTERN)
    min_id = None
    max_id = None
    for filename in files:
        matches = id_reg_expression.findall(filename)[0]
        id, recording_id = [int(id_) for id_ in matches]
        if min_id is None or id < min_id:
            min_id = id
        if max_id is None or id > max_id:
            max_id = id
    return (min_id, max_id)


ARGS = {
    'files': [
        'p5_10.wav',
        'p12_3.wav',
        'p8_7.wav',
        'p2_15.wav'
    ]
}
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 tuple, ndim 1, dtype family int — a tuple 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
identity_func Ran
ran, honoured the contract we drafted
otosense/slang/slang/featurizers.py
code served (permissive licence) · get_code("2fae45e513f08b93")
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, 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)

# The identity_func simply returns its input as-is
# It takes one parameter 'x' which can be any value

# Create a small tensor as the input
x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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, 4) — shape (2, 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
is_iterable Ran
ran on a synthesized input
otosense/slang/slang/snippers.py
code served (permissive licence) · get_code("d0e871c0709173b8")
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
from collections.abc import Iterable

torch.manual_seed(0)

# Test cases for is_iterable function
# The function checks if an object is an instance of Iterable

ARGS = {
    "x": [1, 2, 3]  # A list is iterable
}

# Additional test cases can be constructed like:
# ARGS = {"x": (1, 2, 3)}  # tuple is iterable
# ARGS = {"x": "hello"}     # string is iterable
# ARGS = {"x": 42}          # int is not iterable
# ARGS = {"x": torch.tensor([1, 2, 3])}  # tensor is iterable
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
is_power_of_two Ran
ran, violated the contract we drafted
ShotDownDiane/tcn-master/tcn/tcn.py
code served (permissive licence) · get_code("8a1ea8ba7390e065")
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) 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

torch.manual_seed(0)

ARGS = {
    "num": 8
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
layers_params Ran
ran on a synthesized input
rampage644/wavenet/wavenet/monitor.py
code served (permissive licence) · get_code("6bd837a8733616b3")
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 (20,) 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 SimpleParam:
    """Stand-in for a Chainer parameter object."""
    def __init__(self, name, data):
        self.name = name
        self.data = data
        self.grad = np.zeros_like(data)

class SimpleChain:
    """Stand-in for a Chainer Chain model."""
    def __init__(self):
        self._params = [
            SimpleParam('W', np.random.randn(4, 3).astype(np.float32)),
            SimpleParam('b', np.random.randn(3).astype(np.float32)),
            SimpleParam('W', np.random.randn(2, 4).astype(np.float32)),
        ]
        self.xp = np
    
    def params(self):
        return self._params

model = SimpleChain()
param_name = 'W'
attr_name = 'data'

ARGS = {
    'model': model,
    'param_name': param_name,
    'attr_name': attr_name,
}
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 ndarray, ndim 1, dtype family float, shape (20,) — shape (20,) 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
load_log Ran
ran on a synthesized input
rampage644/wavenet/wavenet/plot_percentiles.py
code served (permissive licence) · get_code("8ce9ce5e4bfcc589")
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 defaultdict. 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 collections
import json
import tempfile
import os

# Create a temporary JSON file with test data
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test_log.json")

test_data = [
    {"loss": 0.5, "accuracy": 0.85, "epoch": 1},
    {"loss": 0.3, "accuracy": 0.90, "epoch": 2},
    {"loss": 0.2, "accuracy": 0.93, "epoch": 3},
]

with open(temp_file, 'w') as f:
    json.dump(test_data, f)

ARGS = {
    "filename": temp_file,
    "keys": ["loss", "accuracy"]
}
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 defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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
make_band_matrix Ran
ran on a synthesized input
otosense/slang/slang/spectrop.py
code served (permissive licence) · get_code("1b1b873049a95522")
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 2, dtype family float, shape (5, 1025) 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

# First, let's understand what make_band_matrix needs:
# - buckets: a list of lists containing indices of non-zero entries
# - n_freq: number of columns (default is DFLT_INPUT_SIZE = 1025)

# Create a simple buckets structure - a list of lists with frequency band indices
buckets = [
    [200, 201, 202],           # bucket 0: indices 200-202
    [203, 204, 205, 206],      # bucket 1: indices 203-206
    [207, 208],                # bucket 2: indices 207-208
    [209, 210, 211, 212, 213], # bucket 3: indices 209-213
    [214, 215],                # bucket 4: indices 214-215
]

# Define n_freq - number of frequency bins
n_freq = 1025

ARGS = {
    'buckets': buckets,
    'n_freq': n_freq,
}
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 2, dtype family float, shape (5, 1025) — shape (5, 1025) 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
make_band_matrix_row Ran
ran on a synthesized input
otosense/slang/slang/spectrop.py
code served (permissive licence) · get_code("9d6af79974e65de0")
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 (5,) 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

ARGS = {
    "list_entries": [0, 3],
    "row_len": 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 1, dtype family float, shape (5,) — shape (5,) 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
mk_callable Ran
ran on a synthesized input
otosense/slang/slang/util.py
code served (permissive licence) · get_code("9c5ee587758b40ea")
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 function. 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
from sklearn.decomposition import PCA
from sklearn.neighbors import NearestNeighbors

# Test case 1: Using a string identifier for a single_* method
call_func_str = 'single_transform'

# Test case 2: Using a callable function directly
def custom_call_func(self, x):
    """Custom callable for a class"""
    return self.transform([x])[0]

ARGS = {
    'call_func': call_func_str
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
mk_model_caller Ran
ran on a synthesized input
otosense/slang/slang/snip_stats.py
code served (permissive licence) · get_code("5d62167ff5d0c13e")
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 function. 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 numpy as np
from sklearn.ensemble import RandomForestClassifier

torch.manual_seed(0)

# Create a minimal model with the required methods and attributes
class MinimalModel:
    def __init__(self):
        # Train a simple sklearn classifier
        X = np.random.rand(10, 4)
        y = np.array([0, 1, 0, 1, 0, 1, 0, 1, 0, 1])
        self.clf = RandomForestClassifier(n_estimators=2, random_state=0)
        self.clf.fit(X, y)
        self.classes_ = self.clf.classes_
    
    def predict_proba(self, X):
        return self.clf.predict_proba(X)
    
    def predict(self, X):
        return self.clf.predict(X)

ARGS = {
    'kind': 'predict_proba'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
normalize Ran
ran on a synthesized input
outofculture/talk-like-me/torchit.py
code served (permissive licence) · get_code("85002c58267da006")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a sample input array with shape (2, 4, 2)
# This is a batch of 2 samples, each with 4 features, and 2 channels
x = np.random.randn(2, 4, 2).astype(np.float32)

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 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
plot_stats Ran
ran on a synthesized input
rampage644/wavenet/server.py
code served (permissive licence) · get_code("693668a628d421fa")
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 numpy as np

class MockPlotter:
    """Mock plotter object that has line() and patch() methods."""
    def line(self, x, y, line_dash=None):
        return {'type': 'line', 'x': x, 'y': y, 'line_dash': line_dash}
    
    def patch(self, x, y, alpha=None):
        return {'type': 'patch', 'x': x, 'y': y, 'alpha': alpha}

# Set up the dataseries dict that get() and get_for() will use
dataseries = {}

p = MockPlotter()
key = "test_metric"

ARGS = {
    "p": p,
    "key": key
}
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 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
polar_to_complex Ran
ran on a synthesized input
outofculture/talk-like-me/torchit.py
code served (permissive licence) · get_code("f86589e53de0342d")
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 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small polar coordinate array
# Shape: (..., 2) where last dimension is [magnitude, phase]
# Using shape (2, 4, 2) to have distinct dimensions
polar_data = np.array([
    [[1.0, 0.0], [2.0, np.pi/4], [0.5, np.pi/2], [1.5, np.pi]],
    [[0.8, np.pi/6], [1.2, np.pi/3], [0.3, 3*np.pi/4], [2.0, 2*np.pi]]
], dtype=np.float32)

ARGS = {
    'x': polar_data
}
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 2, dtype family float, shape (2, 4) — shape (2, 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
quantisize Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
rampage644/wavenet/wavenet/utils.py
code served (permissive licence) · get_code("0007bb74d458f72c")
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 2, dtype family int, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create small image array for quantization
images = np.random.rand(2, 4).astype('f')  # 2 images, 4 pixels each

# Set number of quantization levels
levels = 5

ARGS = {
    'images': images,
    'levels': levels
}
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 2, dtype family int, shape (2, 4) — shape (2, 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
receptive_field_size Ran
ran, honoured the contract we drafted
swasun/VQ-VAE-Speech/src/wavenet_vocoder/wavenet.py
code served (permissive licence) · get_code("204025577056e168")
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

# Define ARGS dict with valid arguments for receptive_field_size
ARGS = {
    "total_layers": 8,
    "num_cycles": 2,
    "kernel_size": 3,
    "dilation": lambda x: 2 ** x
}
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
residual_block Ran
ran; our contract draft was wrong, not the code
zhong110020/keras-tcn/tcn/tcn.py
code served (permissive licence) · get_code("4cc58ece006978a9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import keras
import keras.layers
from keras.layers import Input, Activation, Lambda, Conv1D, SpatialDropout1D
import numpy as np

# Create a small input tensor for the residual block
# Shape: (batch_size=2, timesteps=8, channels=4)
input_tensor = Input(shape=(8, 4))

ARGS = {
    'x': input_tensor,
    'dilation_rate': 2,
    'nb_filters': 16,
    'kernel_size': 3,
    'padding': 'same',
    'activation': 'relu',
    'dropout_rate': 0.1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
keras, 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
sample_from_discretized_mix_logistic Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
swasun/VQ-VAE-Speech/src/wavenet_vocoder/wavenet.py
code served (permissive licence) · get_code("5dede1b4064f5b66")
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, 4) as a later run recorded it. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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)

# The function expects y to be a Tensor of shape (B, C, T)
# where C % 3 == 0 (since nr_mix = C // 3)
# Let's use B=2 (batch), C=6 (so nr_mix=2), T=4 (time steps)
B = 2
C = 6
T = 4
nr_mix = C // 3

# Create y tensor with appropriate shape and values
# The function will transpose it to (B, T, C) internally
y = torch.randn(B, C, T)

# log_scale_min is a float parameter with a default value
log_scale_min = -7.0

ARGS = {
    "y": y,
    "log_scale_min": log_scale_min,
}
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, 4) — shape (2, 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
sliding_window Ran
ran on a synthesized input
outofculture/talk-like-me/utils.py
code served (permissive licence) · get_code("e2e958ef5467a988")
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 2, dtype family int, shape (8, 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

numpy.random.seed(0)

# Create a simple 1D array for sliding window
data = numpy.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])

# Define arguments
ARGS = {
    'data': data,
    'size': 3,
    'stepsize': 1,
    'axis': -1,
    'copy': True
}
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 2, dtype family int, shape (8, 3) — shape (8, 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
split_into Ran
ran on a synthesized input
rampage644/wavenet/preprocess.py
code served (permissive licence) · get_code("d082485663961b2a")
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. 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 numpy as np

torch.manual_seed(0)

# The function split_into takes data and n as parameters
# It indexes into data using data[i::n] for i in range(n)
# So data needs to support slicing with step argument
# This works with lists, tuples, tensors, numpy arrays, etc.

# Create a simple list as data
data = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

# Create n as a small integer
n = 3

ARGS = {
    "data": data,
    "n": n,
}
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 list — a list 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

Showing the first 60. An agent gets every one from get_harvested_code_for_paper("1609.03499").

Also present in other papers' code

Of 103 samples lifted from this paper's linked repositories, 103 carry a function identity and 0 do not. Of the 103 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 2 later) and 101 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
quantisize 2 other papers
1 dated earlier, 1 later
1606.05328 (2016-06)
Conditional Image Generation with PixelCNN Decoders
2310.09604 (2023-10)
Learning Hierarchical Features with Joint Latent Space Energy-Based Prior
identity_func 1 other paper
1 later
1708.07469 (2017-08)
DGM: A deep learning algorithm for solving partial differential equations

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 2026-11-21.

Repositories linked to this paper

More are linked than shown here.

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

This paper introduces WaveNet, a deep neural network for generating raw audio waveforms. The model is fully probabilistic and autoregressive, with the predictive distribution for each audio sample conditioned on all previous ones; nonetheless we show that it can be efficiently trained on data with tens of thousands of samples per second of audio. When applied to text-to-speech, it yields state-ofthe-art performance, with human listeners rating it as significantly more natural sounding than the best parametric and concatenative systems for both English and Mandarin. A single WaveNet can capture the characteristics of many different speakers with equal fidelity, and can switch between them by conditioning on the speaker identity. When trained to model music, we find that it generates novel and often highly realistic musical fragments. We also show that it can be employed as a discriminative model, returning promising results for phoneme recognition.

For agents

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

get_harvested_code_for_paper("1609.03499")
get_code_for_paper("1609.03499")
have("1609.03499")

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

curl https://syntology.ai/api/ran/1609.03499.json

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

[![Syntology run record](https://syntology.ai/api/ran/1609.03499.svg)](https://syntology.ai/paper/1609.03499)

Connect an agent — have() is free.