SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2109.04463 · 2021

Neural Latents Benchmark '21: Evaluating latent variable models of neural population activity

arXiv · PDF · Open in the Atlas

Code that ran

17 of 24 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 15 have no contract verdict; 2 failures of our instrument; 7 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/2109.04463.json · badge: /api/ran/2109.04463.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
neurallatents/nlb_tools canonical (classifier label)
classifier basis: census
5 ran · 4 no recorded run
kabirdabholkar/nlb_tools_fewshot community (classifier label)
classifier basis not recorded on the link
12 ran · 3 no recorded run
FunctionStatusWhere it lives
bits_per_spike Ran
ran on a synthesized input
neurallatents/nlb_tools/nlb_tools/evaluation.py
code served (permissive licence) · get_code("0d3d3adb1e0b12f1")
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
from scipy.special import gammaln

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

# Create small test data for bits_per_spike
# rates: 3d numpy array of predicted firing rates
# Shape: (batch=2, neurons=3, time=4)
rates = np.random.rand(2, 3, 4) * 10 + 0.5  # Positive rates

# spikes: 3d numpy array of true spike counts
# Shape: (batch=2, neurons=3, time=4) - same shape as rates
spikes = np.random.poisson(lam=5, size=(2, 3, 4)).astype(float)

# Construct arguments dict
ARGS = {
    "rates": rates,
    "spikes": spikes,
}
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
chop_data Ran
ran on a synthesized input
neurallatents/nlb_tools/nlb_tools/chop.py
code served (permissive licence) · get_code("7be81192bfe31963")
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 (3, 8, 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 2D array of continuous data (T x N)
# T=20 time points, N=4 features
data = np.random.randn(20, 4).astype(np.float32)

# Define overlap and window parameters
overlap = 3
window = 8
offset = 0

# Construct ARGS dict with all parameters
ARGS = {
    "data": data,
    "overlap": overlap,
    "window": window,
    "offset": offset,
}
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 (3, 8, 4) — shape (3, 8, 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
extract_reallyheldout_by_id Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/make_fewshot.py
code served (permissive licence) · get_code("edfc3fd7ed9643f0")
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

# Create sample data_dict with heldin neurons and psth
data_dict = {
    'heldin_spikes': np.random.randn(10, 8, 12),  # (time, batch, neurons)
    'heldin_rates': np.random.randn(10, 6),  # (time, neurons)
    'psth': np.random.randn(15, 9),  # (time, neurons)
    'other_data': np.random.randn(5, 5),  # non-heldin data
}

# Define neuron IDs to extract for few-shot evaluation
neuron_ids_to_extract = [1, 3, 5]

ARGS = {
    'data_dict': data_dict,
    'neuron_ids_to_extract': neuron_ids_to_extract,
}
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
fewshot_from_train Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/make_fewshot.py
code served (permissive licence) · get_code("dd4a6f58dc498993")
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 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

# Create a simple train_dict with numpy arrays
# Each value should be a 2D array where the first dimension is the number of trials
train_array_1 = np.random.randn(12, 4)  # 12 trials, 4 features
train_array_2 = np.random.randn(12, 3)  # 12 trials, 3 features

train_dict = {
    'feature_set_1': train_array_1,
    'feature_set_2': train_array_2,
}

# Define K values for few-shot splits
Kvalues = [2, 3, 4, 6]

ARGS = {
    'train_dict': train_dict,
    'Kvalues': Kvalues,
}
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, ndim 1, dtype family float — 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
fit_poisson Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/sklearn_glm.py
code served (permissive licence) · get_code("2687687849f6dbf2")
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

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

# Create small dummy data for the function
# train_factors_s: input features for training (n_samples, n_features)
train_factors_s = np.random.randn(10, 4).astype(np.float32)

# eval_factors_s: input features for evaluation (m_samples, n_features)
eval_factors_s = np.random.randn(8, 4).astype(np.float32)

# train_spikes_s: output spike counts for training (n_samples, n_channels)
# Using non-negative values since Poisson regression expects count data
train_spikes_s = np.random.poisson(lam=2, size=(10, 3)).astype(np.float32)

# eval_spikes_s: output spike counts for evaluation (m_samples, n_channels)
# Can be None, but providing it to test the full path
eval_spikes_s = np.random.poisson(lam=2, size=(8, 3)).astype(np.float32)

# alpha: regularization parameter for PoissonRegressor
alpha = 0.01

ARGS = {
    "train_factors_s": train_factors_s,
    "eval_factors_s": eval_factors_s,
    "train_spikes_s": train_spikes_s,
    "eval_spikes_s": eval_spikes_s,
    "alpha": alpha,
}
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 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
fit_poisson_parallel Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/sklearn_glm.py
code served (permissive licence) · get_code("2477a50a3c586b76")
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

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

# Create small synthetic data
# train_factors_s: training input features (samples x features)
train_factors_s = np.random.randn(10, 4).astype(np.float64)

# eval_factors_s: evaluation input features (samples x features)
eval_factors_s = np.random.randn(6, 4).astype(np.float64)

# train_spikes_s: training output spike counts (samples x channels)
# Using small non-negative integers for spike counts
train_spikes_s = np.random.poisson(2, size=(10, 3)).astype(np.float64)

# eval_spikes_s: evaluation output spike counts (samples x channels)
eval_spikes_s = np.random.poisson(2, size=(6, 3)).astype(np.float64)

# Other parameters
alpha = 0.01
max_iter = 100

ARGS = {
    "train_factors_s": train_factors_s,
    "eval_factors_s": eval_factors_s,
    "train_spikes_s": train_spikes_s,
    "eval_spikes_s": eval_spikes_s,
    "alpha": alpha,
    "max_iter": max_iter,
}
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 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
generate_unique_prefix Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/collate_results.py
code served (permissive licence) · get_code("5cc10defe6e4c731")
How it ran

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

import hashlib

def generate_unique_prefix(path):
    return hashlib.md5(path.encode()).hexdigest()[:6]

ARGS = {
    "path": "/home/user/model/checkpoint.pt"
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
merge_chops Ran
ran on a synthesized input
neurallatents/nlb_tools/nlb_tools/chop.py
code served (permissive licence) · get_code("3d8a5bc8cf916fb0")
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 (26, 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 logging

logging.basicConfig(level=logging.WARNING)

# Create sample data for merge_chops
# data should be a 3D array: SxTxN (S segments, T time points, N features)
S = 3  # number of overlapping segments
T = 10  # time points per segment
N = 4  # number of features

# Create overlapping segments with some reasonable values
data = np.random.randn(S, T, N).astype(np.float32)

# overlap parameter - number of overlapping points between segments
overlap = 2

# orig_len - original length of continuous data (optional)
orig_len = 20

# smooth_pwr - power of smoothing
smooth_pwr = 2

ARGS = {
    "data": data,
    "overlap": overlap,
    "orig_len": orig_len,
    "smooth_pwr": smooth_pwr,
}
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 (26, 4) — shape (26, 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
neg_log_likelihood Ran
ran on a synthesized input
neurallatents/nlb_tools/nlb_tools/evaluation.py
code served (permissive licence) · get_code("dd5f9738be104eaf")
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 float32, 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 numpy as np
from scipy.special import gammaln

# Create sample rate predictions and spike counts
# rates: predicted firing rates (non-negative)
# spikes: observed spike counts (non-negative integers)

rates = np.array([
    [0.5, 1.2, 2.3, 0.1],
    [1.5, 0.8, 3.1, 2.0]
], dtype=np.float32)

spikes = np.array([
    [0, 1, 2, 0],
    [2, 1, 3, 1]
], dtype=np.float32)

ARGS = {
    "rates": rates,
    "spikes": spikes,
    "zero_warning": True
}
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 float32, 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
r2_score Ran
ran; our contract draft was wrong, not the code
kabirdabholkar/nlb_tools_fewshot/nlb_tools/metrics.py
pointer only (licence: MIT); the stored copy get_code serves, from arsedler9/lfads-torch, is pointer only (licence: NOASSERTION) · get_code("a27593d38113b6a5")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from arsedler9/lfads-torch, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

read_concat_results Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/collate_results.py
code served (permissive licence) · get_code("68e6cf9fdc15aae9")
How it ran

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

import os
import pandas as pd
import tempfile
import shutil

# Create a temporary directory structure with CSV files
temp_root = tempfile.mkdtemp()

# Create subdirectories and CSV files
subdir1 = os.path.join(temp_root, "exp1")
subdir2 = os.path.join(temp_root, "exp2")
os.makedirs(subdir1, exist_ok=True)
os.makedirs(subdir2, exist_ok=True)

# Create sample CSV files
df1 = pd.DataFrame({
    'metric1': [0.5, 0.6],
    'metric2': [0.7, 0.8]
})
df1.index.name = 'index'
df1.to_csv(os.path.join(subdir1, "results.csv"))

df2 = pd.DataFrame({
    'metric1': [0.75, 0.85],
    'metric2': [0.65, 0.95]
})
df2.index.name = 'index'
df2.to_csv(os.path.join(subdir2, "results.csv"))

ARGS = {
    'root_dir': temp_root,
    'endswith': 'results.csv',
    'include_path': True
}
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 DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
regional_bits_per_spike Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
kabirdabholkar/nlb_tools_fewshot/nlb_tools/metrics.py
pointer only (licence: MIT); the stored copy get_code serves, from arsedler9/lfads-torch, is pointer only (licence: NOASSERTION) · get_code("b2719609b2d567a8")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from arsedler9/lfads-torch, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

result_dict_to_pandas Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/fewshot_utils.py
code served (permissive licence) · get_code("de16cf54220b6637")
How it ran

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

import pandas as pd

# Construct valid arguments for result_dict_to_pandas
# The function expects:
# - result_data: a list of dicts where each dict has a key mapping to another dict of column data
# - **kwargs: optional keyword arguments to add as columns

# Create a sample result_data structure
result_data = [
    {
        "model_metrics": {
            "[train] accuracy": 0.95,
            "[train] loss": 0.05,
            "[train] f1": 0.92
        }
    }
]

# Create kwargs
kwargs = {
    "experiment_id": "exp_001",
    "timestamp": "2024-01-01"
}

ARGS = {
    "result_data": result_data,
    "kwargs": kwargs
}
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 DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
run_model_on_numpy_dummy Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/load_and_save_latents.py
code served (permissive licence) · get_code("984d705b60df1c36")
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 4, 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 torch
from torch import nn

torch.manual_seed(0)

class DummyModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

model = DummyModel()

spikes_heldin = np.random.randn(2, 5, 6).astype(np.float32)

spikes_full_shape = (2, 5, 6)

ARGS = {
    'model': model,
    'spikes_heldin': spikes_heldin,
    'spikes_full_shape': spikes_full_shape
}
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 tuple, ndim 4, dtype family float — 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
select_best_cobps Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/cross_decoding.py
code served (permissive licence) · get_code("9c1716805b975c6e")
How it ran

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

import torch
import pandas as pd

torch.manual_seed(0)

# The function expects D to be a pandas DataFrame with a column named by the 'use' parameter
# It accesses D[use] to get max value and filters rows based on that column
# Create a small DataFrame with a 'co-bps' column
D = pd.DataFrame({
    'co-bps': [0.5, 0.75, 0.85, 0.9, 0.65],
    'other_col': [1, 2, 3, 4, 5]
})

ARGS = {
    'D': D,
    'use': 'co-bps',
    'margin': 0.02
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
smooth_column Ran
ran on a synthesized input
neurallatents/nlb_tools/nlb_tools/nwb_interface.py
code served (permissive licence) · get_code("67a8701afb5319da")
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 (8,) 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 scipy.signal as signal

# Create small test data
np.random.seed(0)

# Create a 1d array of data to smooth
x = np.array([1.0, 2.0, 3.0, 4.0, 5.0, np.nan, 7.0, 8.0], dtype=np.float64)

# Create a smoothing kernel (simple averaging window)
window = np.array([0.25, 0.5, 0.25], dtype=np.float64)

# Set ignore_nans to True to test the nan handling path
ignore_nans = True

# Set dtype for output
dtype = np.float64

# Construct the args tuple as expected by the function
args = (x, window, ignore_nans, dtype)

ARGS = {
    "args": args
}
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 ndarray, ndim 1, dtype family float, shape (8,) — shape (8,) 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_chunks_of_K Ran
ran on a synthesized input
kabirdabholkar/nlb_tools_fewshot/nlb_tools/make_fewshot.py
code served (permissive licence) · get_code("abd84c9759241332")
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 3, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a sample train_array with shape (20, 5)
# This will be split into chunks based on K
train_array = np.arange(100).reshape(20, 5).astype(np.float32)

# K is the number of chunks to split into
K = 4

ARGS = {
    "train_array": train_array,
    "K": K,
}
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 3, 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
bits_per_spike Not yet run
unverified (harvested, no recorded run)
kabirdabholkar/nlb_tools_fewshot/nlb_tools/metrics.py
pointer only (licence: MIT); the stored copy get_code serves, from arsedler9/lfads-torch, is pointer only (licence: NOASSERTION) · get_code("e44e57b951a765b2")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from arsedler9/lfads-torch, and GitHub could not classify the licence of that repository (NOASSERTION). 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. No invocation recorded.

evaluate Not yet run
unverified (harvested, no recorded run)
neurallatents/nlb_tools/nlb_tools/evaluation.py
code served (permissive licence) · get_code("bd7c0d0bf0c8c24c")
How it ran

no invocation recorded.

evaluate Not yet run
unverified (harvested, no recorded run)
kabirdabholkar/nlb_tools_fewshot/nlb_tools/evaluation.py
code served (permissive licence) · get_code("14fc654bea6d290a")
How it ran

no invocation recorded.

load_result_files Not yet run
unverified (harvested, no recorded run)
kabirdabholkar/nlb_tools_fewshot/nlb_tools/cross_decoding.py
code served (permissive licence) · get_code("861c3e54aa2c4e05")
How it ran

no invocation recorded.

make_eval_input_tensors Not yet run
unverified (harvested, no recorded run)
neurallatents/nlb_tools/nlb_tools/make_tensors.py
code served (permissive licence) · get_code("eff7a97dd8cb0604")
How it ran

no invocation recorded.

make_eval_target_tensors Not yet run
unverified (harvested, no recorded run)
neurallatents/nlb_tools/nlb_tools/make_tensors.py
code served (permissive licence) · get_code("bcfca1cb80e661c1")
How it ran

no invocation recorded.

make_train_input_tensors Not yet run
unverified (harvested, no recorded run)
neurallatents/nlb_tools/nlb_tools/make_tensors.py
code served (permissive licence) · get_code("16f126abd979f82f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 24 samples lifted from this paper's linked repositories, 24 carry a function identity and 0 do not. Of the 24 distinct functions with an identity, 3 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 3 later) and 21 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
bits_per_spike 2 other papers
2 later
2403.01371 (2024-03)
eXponential FAmily Dynamical Systems (XFADS): Large-scale nonlinear Gaussian state-space modeling
2512.00239 (2025-12)
Self-Supervised Dynamical System Representations for Physiological Time-Series
r2_score 1 other paper
1 later
2512.00239 (2025-12)
Self-Supervised Dynamical System Representations for Physiological Time-Series
regional_bits_per_spike 1 other paper
1 later
2512.00239 (2025-12)
Self-Supervised Dynamical System Representations for Physiological Time-Series

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

Advances in neural recording present increasing opportunities to study neural activity in unprecedented detail. Latent variable models (LVMs) are promising tools for analyzing this rich activity across diverse neural systems and behaviors, as LVMs do not depend on known relationships between the activity and external experimental variables. However, progress with LVMs for neuronal population activity is currently impeded by a lack of standardization, resulting in methods being developed and compared in an ad hoc manner. To coordinate these modeling efforts, we introduce a benchmark suite for latent variable modeling of neural population activity. We curate four datasets of neural spiking activity from cognitive, sensory, and motor areas to promote models that apply to the wide variety of activity seen across these areas. We identify unsupervised evaluation as a common framework for evaluating models across datasets, and apply several baselines that demonstrate benchmark diversity. We release this benchmark through EvalAI. http://neurallatents.github.io

For agents

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

get_harvested_code_for_paper("2109.04463")
get_code_for_paper("2109.04463")
have("2109.04463")

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

curl https://syntology.ai/api/ran/2109.04463.json

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

[![Syntology run record](https://syntology.ai/api/ran/2109.04463.svg)](https://syntology.ai/paper/2109.04463)

Connect an agent — have() is free.