SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2008.01307 · 2020

The Jazz Transformer on the Front Line: Exploring the Shortcomings of AI-composed Music through Quantitative Measures

arXiv · PDF · Open in the Atlas

Code that ran

13 of 14 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 13 have no contract verdict; 0 failures of our instrument; 1 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2008.01307.json · badge: /api/ran/2008.01307.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
slSeanWU/MusDr canonical (classifier label)
classifier basis: census
9 ran · 0 no recorded run
slSeanWU/jazz_transformer canonical (classifier label)
classifier basis: census
4 ran · 1 no recorded run
FunctionStatusWhere it lives
collect_segments Ran
ran on a synthesized input
slSeanWU/jazz_transformer/src/convert_to_remi.py
code served (permissive licence) · get_code("aa1a0fcc8b6dfe82")
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 pandas as pd

class MockBeat:
    """Mock beat object with patch_segment_tag method"""
    def __init__(self):
        self.segment_tags = {}
    
    def patch_segment_tag(self, start_seg=None, end_seg=None):
        """Mock method that stores segment tag information"""
        self.segment_tags['start_seg'] = start_seg
        self.segment_tags['end_seg'] = end_seg

# Create a sample beat_df DataFrame
beat_df = pd.DataFrame({
    'form': ['A', 'A', 'B', 'B', 'C'],
    'bar': [1, 2, 3, 4, 5],
    'beat': [1, 1, 1, 1, 1]
})

# Create a beat_dict with MockBeat instances for each (bar, beat) tuple
beat_dict = {
    (1, 1): MockBeat(),
    (2, 1): MockBeat(),
    (3, 1): MockBeat(),
    (4, 1): MockBeat(),
    (5, 1): MockBeat(),
}

ARGS = {
    'beat_df': beat_df,
    'beat_dict': beat_dict
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — 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
compute_SM_dot Ran
ran on a synthesized input
slSeanWU/MusDr/mueller_audio_tools/ssm_features.py
code served (permissive licence) · get_code("e0ab2086c54458d0")
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, 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 small feature sequences
# X shape: (feature_dim, sequence_length_1)
# Y shape: (feature_dim, sequence_length_2)
# Using distinct dimensions for clarity

X = np.random.randn(4, 3).astype(np.float32)  # 4-dim features, 3 time steps
Y = np.random.randn(4, 5).astype(np.float32)  # 4-dim features, 5 time steps

ARGS = {
    "X": X,
    "Y": Y,
}
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, 3) — shape (5, 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
compute_accumulated_score_matrix Ran
ran on a synthesized input
slSeanWU/MusDr/mueller_audio_tools/scapeplot.py
code served (permissive licence) · get_code("5d740a70b2bc6c53")
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
import math

# Create a small valid S_seg matrix
# S_seg represents a submatrix of an enhanced and normalized SSM
# It should satisfy: S(n,m) <= 1 and S(n,n) = 1
# Shape: (N, N) where N is the number of rows/columns

N = 5
S_seg = np.array([
    [1.0, 0.8, 0.6, 0.4, 0.2],
    [0.8, 1.0, 0.7, 0.5, 0.3],
    [0.6, 0.7, 1.0, 0.6, 0.4],
    [0.4, 0.5, 0.6, 1.0, 0.7],
    [0.2, 0.3, 0.4, 0.7, 1.0]
], dtype=np.float64)

ARGS = {
    "S_seg": S_seg
}
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
compute_histogram_entropy Ran
ran on a synthesized input
slSeanWU/MusDr/musdr/side_utils.py
code served (permissive licence) · get_code("1f26bd9136f2e20d")
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 float64, ndim 0, 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
import scipy.stats

# Create a normalized histogram (probabilities that sum to 1)
# Using a small array with distinct values to test the function
histogram = np.array([0.1, 0.2, 0.3, 0.25, 0.15])

# Ensure it's normalized (sums to 1)
histogram = histogram / histogram.sum()

ARGS = {
    "hist": histogram
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
compute_optimal_path_family Ran
ran on a synthesized input
slSeanWU/MusDr/mueller_audio_tools/scapeplot.py
code served (permissive licence) · get_code("be870c68b8b96085")
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)

# Create a small accumulated score matrix D
# The function expects D to have shape (N, M) where it accesses D[n, m]
# Looking at the function logic, it needs:
# - D.shape[0] for N (number of rows)
# - D.shape[1] for M (number of columns)
# - Numerical values accessible via D[i, j] indexing
# We'll use numpy array to match typical usage
D = np.array([
    [0.0, 1.0, 2.0, 3.0],
    [1.0, 2.0, 3.0, 4.0],
    [2.0, 3.0, 4.0, 5.0],
    [3.0, 4.0, 5.0, 6.0],
    [4.0, 5.0, 6.0, 7.0]
], dtype=np.float32)

ARGS = {
    "D": D
}
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
embedding_lookup Ran
ran on a synthesized input
slSeanWU/jazz_transformer/transformer_xl/modules.py
code served (permissive licence) · get_code("0a923f63ece46f08")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

# Create a lookup table (embedding matrix)
# Shape: [vocab_size=5, embedding_dim=4]
lookup_table = tf.constant(
    np.random.randn(5, 4).astype(np.float32),
    dtype=tf.float32
)

# Create indices to look up
# Shape: [batch_size=2]
x = tf.constant([0, 2], dtype=tf.int32)

ARGS = {
    "lookup_table": lookup_table,
    "x": x
}
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 EagerTensor, 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
get_chord_sequence Ran
ran on a synthesized input
slSeanWU/MusDr/musdr/side_utils.py
code served (permissive licence) · get_code("74c496bf558c63ae")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Construct a realistic event sequence
# The function expects events that are in chord_evs dict keys
# It looks for 'Chord-Slash' events specifically
# The logic groups events in triplets where the 3rd event is a 'Chord-Slash'

# Create sample chord-related events
chord_root_events = ['C', 'D', 'E', 'F', 'G', 'A', 'B']
chord_quality_events = ['maj', 'min', 'dim', 'aug']
chord_slash_events = ['C/E', 'G/B', 'D/F#', 'Am/E']

# Build chord_evs dictionary mapping event types to their possible encodings
chord_evs = {
    'Chord-Root': chord_root_events,
    'Chord-Quality': chord_quality_events,
    'Chord-Slash': chord_slash_events
}

# Construct an event sequence that will satisfy the function logic
# We need triplets where the 3rd element is in 'Chord-Slash'
# And we need non-chord events mixed in to test filtering

ev_seq = [
    'C', 'maj', 'C/E',           # triplet 1: valid (ends with Chord-Slash)
    'D', 'min', 'G/B',           # triplet 2: valid (ends with Chord-Slash)
    'E', 'dim', 'D/F#',          # triplet 3: valid (ends with Chord-Slash)
]

ARGS = {
    'ev_seq': ev_seq,
    'chord_evs': chord_evs
}
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 2, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_event_seq Ran
ran on a synthesized input
slSeanWU/MusDr/musdr/side_utils.py
code served (permissive licence) · get_code("c77e1599951d2579")
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 pandas as pd
import tempfile
import os

# Create a temporary CSV file with the required format
temp_dir = tempfile.mkdtemp()
csv_path = os.path.join(temp_dir, "piece.csv")

# Create a sample DataFrame with event encodings
df = pd.DataFrame({
    'ENCODING': [1, 2, 3, 4, 5, 6, 7, 8]
})
df.to_csv(csv_path, index=False, encoding='utf-8')

ARGS = {
    'piece_csv': csv_path,
    'seq_col_name': 'ENCODING'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — 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
get_note_duration_distr Ran
ran on a synthesized input
slSeanWU/jazz_transformer/src/explore_mcsv.py
code served (permissive licence) · get_code("28d4744d03114079")
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
import pandas as pd

# Create beat_df with required columns: 'bar' and 'duration'
beat_df = pd.DataFrame({
    'bar': [1, 1, 2, 2, 3, 3],
    'duration': [0.5, 0.5, 0.5, 0.5, 0.5, 0.5]
})

# Create melody_df with required columns: 'bar' and 'duration'
melody_df = pd.DataFrame({
    'bar': [1, 1, 2, 3, 3],
    'duration': [0.25, 0.5, 0.5, 0.25, 0.25]
})

ARGS = {
    'beat_df': beat_df,
    'melody_df': melody_df
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
normal_softmax Ran
ran on a synthesized input
slSeanWU/jazz_transformer/transformer_xl/modules.py
code served (permissive licence) · get_code("71f9313c35afe0e6")
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 tensorflow as tf
import numpy as np

tf.compat.v1.disable_eager_execution()

# Set random seed for reproducibility
np.random.seed(0)
tf.compat.v1.set_random_seed(0)

# Define dimensions - use distinct sizes for different meanings
batch_size = 2
hidden_dim = 4
n_token = 5
proj_dim = 3

# Create hidden tensor: shape (batch, seq_len, hidden_dim)
# Using ibd notation: i=batch, b=seq_len, d=hidden_dim
hidden = tf.constant(
    np.random.randn(batch_size, 2, hidden_dim).astype(np.float32)
)

# Create target tensor: shape (batch, seq_len) with token indices
# Labels must be integers in range [0, n_token)
target = tf.constant(
    np.random.randint(0, n_token, size=(batch_size, 2)),
    dtype=tf.int64
)

# params[0] is params_W: shape (n_token, hidden_dim) or (n_token, proj_dim) if proj exists
# Used in einsum 'ibd,nd->ibn', so W is (n_token, d)
params_W = tf.constant(
    np.random.randn(n_token, proj_dim).astype(np.float32)
)

# params[1] is params_projs: projection matrix or None
# Used in einsum 'ibd,ed->ibe', so proj is (proj_dim, hidden_dim)
# where e=proj_dim, d=hidden_dim
params_projs = tf.constant(
    np.random.randn(proj_dim, hidden_dim).astype(np.float32)
)

# params is a tuple/list of [params_W, params_projs]
params = [params_W, params_projs]

ARGS = {
    'hidden': hidden,
    'target': target,
    'n_token': n_token,
    'params': params,
    'scope': 'normal_softmax',
}
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 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
normalization_properties_SSM Ran
ran on a synthesized input
slSeanWU/MusDr/mueller_audio_tools/scapeplot.py
code served (permissive licence) · get_code("970a6eb890d8917e")
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 (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 numpy as np

# Create a small self-similarity matrix (SSM)
# The function modifies it in-place and expects it to be a 2D numpy array
# with shape (N, N) where N is the number of frames
S = np.array([
    [0.8, 0.5, 0.3],
    [0.5, 0.7, 0.4],
    [0.3, 0.4, 0.6]
], dtype=np.float32)

ARGS = {
    "S": S
}
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 (3, 3) — shape (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: none by the harness; any seed is in the setup block
normalize_feature_sequence Ran
ran on a synthesized input
slSeanWU/MusDr/mueller_audio_tools/ssm_features.py
code served (permissive licence) · get_code("ed88321a270dd815")
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, 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 feature sequence with K=5 features and N=3 time steps
np.random.seed(0)
X = np.random.randn(5, 3) * 2 + 1

# Create a custom vector for threshold cases
v_custom = np.array([0.1, 0.2, 0.15, 0.25, 0.3])

ARGS = {
    'X': X,
    'norm': '2',
    'threshold': 0.0001,
    'v': v_custom
}
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, 3) — shape (5, 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
smooth_downsample_feature_sequence Ran
ran on a synthesized input
slSeanWU/MusDr/mueller_audio_tools/ssm_features.py
code served (permissive licence) · get_code("ce3d8f5244b4f7ee")
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
from scipy import signal

# Create a small feature sequence
# Shape: (n_features, n_frames)
# Using distinct dimensions: 4 features, 100 frames
X = np.random.randn(4, 100)

# Frame rate (arbitrary positive value)
Fs = 10

# Filter length (must be odd for typical usage)
filt_len = 41

# Downsampling factor
down_sampling = 10

# Window type for the smoothing filter
w_type = 'boxcar'

ARGS = {
    'X': X,
    'Fs': Fs,
    'filt_len': filt_len,
    'down_sampling': down_sampling,
    'w_type': w_type
}
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
normal_embedding_lookup Not yet run
unverified (harvested, no recorded run)
slSeanWU/jazz_transformer/transformer_xl/modules.py
code served (permissive licence) · get_code("43ce03d635bf63b4")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 14 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.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

This paper presents the Jazz Transformer, a generative model that utilizes a neural sequence model called the Transformer-XL for modeling lead sheets of Jazz music. Moreover, the model endeavors to incorporate structural events present in the Weimar Jazz Database (WJazzD) for inducing structures in the generated music. While we are able to reduce the training loss to a low value, our listening test suggests however a clear gap between the average ratings of the generated and real compositions. We therefore go one step further and conduct a series of computational analysis of the generated compositions from different perspectives. This includes analyzing the statistics of the pitch class, grooving, and chord progression, assessing the structureness of the music with the help of the fitness scape plot, and evaluating the model's understanding of Jazz music through a MIREX-like continuation prediction task. Our work presents in an analytical manner why machine-generated music to date still falls short of the artwork of humanity, and sets some goals for future work on automatic composition to further pursue.

For agents

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

get_harvested_code_for_paper("2008.01307")
get_code_for_paper("2008.01307")
have("2008.01307")

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

curl https://syntology.ai/api/ran/2008.01307.json

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

[![Syntology run record](https://syntology.ai/api/ran/2008.01307.svg)](https://syntology.ai/paper/2008.01307)

Connect an agent — have() is free.