SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2311.00136 · 2023

Neuroformer: Multimodal and Multitask Generative Pretraining for Brain Data

arXiv · PDF · Open in the Atlas

Code that ran

16 of 19 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; 4 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 12 failures of our instrument; 3 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/2311.00136.json · badge: /api/ran/2311.00136.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
a-antoniades/neuroformer canonical (classifier label)
classifier basis: census
16 ran · 3 no recorded run
FunctionStatusWhere it lives
calc_corr_psth Ran
ran, honoured the contract we drafted
a-antoniades/neuroformer/neuroformer/analysis.py
code served (permissive licence) · get_code("d1186178a7836808")
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
import numpy as np
from scipy import stats

# Create sample PSTH (Peri-Stimulus Time Histogram) rate data
# rates1 and rates2 are dictionaries mapping neuron IDs to firing rate arrays
rates1 = {
    'neuron_1': np.array([1.2, 2.3, 3.1, 2.8, 1.9]),
    'neuron_2': np.array([0.5, 1.2, 2.1, 1.8, 0.9]),
    'neuron_3': np.array([2.1, 3.2, 4.1, 3.5, 2.3]),
}

rates2 = {
    'neuron_1': np.array([1.1, 2.4, 3.0, 2.9, 1.8]),
    'neuron_2': np.array([0.6, 1.1, 2.2, 1.7, 1.0]),
    'neuron_3': np.array([2.0, 3.3, 4.0, 3.6, 2.2]),
}

# neurons parameter can filter which neurons to include
neurons = ['neuron_1', 'neuron_2']

ARGS = {
    'rates1': rates1,
    'rates2': rates2,
    'neurons': neurons,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, scipy — 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
combo3_V1AL_callback Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
a-antoniades/neuroformer/neuroformer/datasets.py
code served (permissive licence) · get_code("4c1a31f765ef032e")
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 4, dtype family float, shape (1, 32, 64, 112) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Construct frames tensor with shape [3, 640, 64, 112]
# 3 stimuli, each with 640 frames of 64x112 video
frames = torch.randn(3, 640, 64, 112)

# frame_idx: some index within the 640 frames
frame_idx = 100

# n_frames: number of frames to look back
n_frames = 32

# trial: determines which stimulus (n_stim) to use
trial = 25

ARGS = {
    'frames': frames,
    'frame_idx': frame_idx,
    'n_frames': n_frames,
    'trial': trial
}
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 Tensor, ndim 4, dtype family float, shape (1, 32, 64, 112) — shape (1, 32, 64, 112) 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
contrastive_loss Ran
ran, honoured the contract we drafted
a-antoniades/neuroformer/neuroformer/modules.py
code served (permissive licence) · get_code("d496ef53516a57ca")
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 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)

# The function expects:
# - features: a dict with modality names as keys and embedding tensors as values
# - temp: temperature parameter (scalar float)

# Build features dict with multiple modalities
# Each embedding should be shape (batch_size, embedding_dim)
batch_size = 2
embedding_dim = 4

features = {
    "modality_1": torch.randn(batch_size, embedding_dim),
    "modality_2": torch.randn(batch_size, embedding_dim),
    "modality_3": torch.randn(batch_size, embedding_dim),
}

ARGS = {
    "features": features,
    "temp": 0.1,
}
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 Tensor, 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
convolve_atts_3D Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
a-antoniades/neuroformer/neuroformer/attention_vis.py
code served (permissive licence) · get_code("ff4f6fdd29b64764")
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, 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
from scipy import signal

# Create a small 4D array with shape (ID, T, Y, X)
# Using distinct sizes: 2 IDs, 3 time steps, 4 Y positions, 5 X positions
stim_atts = np.random.randn(2, 3, 4, 5).astype(np.float32)

ARGS = {
    "stim_atts": stim_atts
}
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 4, dtype family float, shape (2, 3, 4, 5) — shape (2, 3, 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
get_attr Ran
ran; our contract draft was wrong, not the code
a-antoniades/neuroformer/neuroformer/model_neuroformer.py
code served (permissive licence) · get_code("c3503f3549cd9acf")
How it ran

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

import torch
import numpy as np

torch.manual_seed(0)

# Test case 1: obj is a dict with the name present
obj_dict_with_name = {"key1": "value1", "key2": 42}
name_present = "key1"
default_val_1 = None

# Test case 2: obj is a dict without the name, uses default
obj_dict_without_name = {"key1": "value1"}
name_absent = "missing_key"
default_val_2 = "default_value"

# Test case 3: obj is an object with the attribute
class SimpleObject:
    def __init__(self):
        self.attr1 = "object_attr_value"
        self.attr2 = 123

obj_with_attr = SimpleObject()
name_obj_present = "attr1"
default_val_3 = None

# Test case 4: obj is an object without the attribute, uses default
obj_without_attr = SimpleObject()
name_obj_absent = "missing_attr"
default_val_4 = "fallback"

# Use the first test case for ARGS
ARGS = {
    "obj": obj_dict_with_name,
    "name": name_present,
    "default": default_val_1
}
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 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
get_emb Ran
ran, honoured the contract we drafted
a-antoniades/neuroformer/neuroformer/model_neuroformer.py
code served (permissive licence) · get_code("33805df68cb6892e")
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, 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 torch

torch.manual_seed(0)

# The function expects sin_inp to be a tensor that has .sin() and .cos() methods
# It will stack the results and flatten them
sin_inp = torch.randn(2, 4)

ARGS = {
    "sin_inp": sin_inp
}
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, 8) — shape (2, 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
get_rates Ran
ran; our contract draft was wrong, not the code
a-antoniades/neuroformer/neuroformer/analysis.py
code served (permissive licence) · get_code("f2941cf251031ecc")
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
import pandas as pd

# Create a small dataframe with the required columns
data = {
    'True': [1, 1, 2, 2, 3, 3],
    'Predicted': [1, 2, 2, 3, 1, 3],
    'Interval': [0.5, 0.5, 0.75, 0.75, 0.25, 0.25]
}
df = pd.DataFrame(data)

# Create intervals array
intervals = np.array([0.25, 0.5, 0.75, 1.0])

ARGS = {
    'df': df,
    'intervals': intervals
}
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 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
get_rates Ran
ran; our contract draft was wrong, not the code
a-antoniades/neuroformer/neuroformer/analysis.py
code served (permissive licence) · get_code("2398af86975c9957")
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
import pandas as pd

# Create a sample DataFrame with the structure expected by the function
# The function expects a DataFrame with 'ID', 'Interval', and 'Time' columns
data = {
    'ID': [1, 1, 1, 2, 2, 2, 3, 3],
    'Interval': ['A', 'B', 'A', 'B', 'C', 'A', 'B', 'C'],
    'Time': [10, 20, 15, 25, 30, 12, 18, 22]
}
df = pd.DataFrame(data)

# List of IDs to process
ids = [1, 2, 3]

# Array of intervals
intervals = ['A', 'B', 'C']

ARGS = {
    'df': df,
    'ids': ids,
    'intervals': intervals,
    'interval': 'Interval'
}
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 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
model_ensemble Ran
ran; our contract draft was wrong, not the code
a-antoniades/neuroformer/neuroformer/simulation.py
code served (permissive licence) · get_code("30d823e13e046f0b")
How it ran

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

import torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        logits = {
            'id': self.linear(x),
            'dt': self.linear(x) * 2
        }
        return logits, None, None

# Create ensemble of models
model1 = SimpleModel()
model2 = SimpleModel()
models = [model1, model2]

# Create input tensor (batch_size=2, feature_dim=4)
x = torch.randn(2, 4)

ARGS = {
    'models': models,
    '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 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
rollout_attentions Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
a-antoniades/neuroformer/neuroformer/attention_vis.py
code served (permissive licence) · get_code("861a7367a2ed59cd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (4, 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)

# Create attention tensor with shape (L, H, ID, F)
# L=3 (layers), H=2 (heads), ID=4 (input dimension), F=5 (feature dimension)
L, H, ID, F = 3, 2, 4, 5

att = torch.randn(L, H, ID, F)

ARGS = {
    "att": att
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (4, 5) — shape (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
separate_eos_tokens Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
a-antoniades/neuroformer/neuroformer/modules.py
code served (permissive licence) · get_code("2e9f1a5673215030")
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 torch

torch.manual_seed(0)

# Define parameters
eos_id_tok = 5  # EOS token index for 'id'
eos_dt_tok = 6  # EOS token index for 'dt'

batch_size = 2
seq_length = 4
num_classes = 3

# Create targets dictionary with 'id' and 'dt' tensors
# Shape: (batch_size, seq_length)
targets = {
    'id': torch.tensor([[1, 5, 2, 3], [5, 2, 5, 4]]),  # Contains some EOS tokens (5)
    'dt': torch.tensor([[2, 3, 6, 1], [3, 6, 2, 1]])   # Contains some EOS tokens (6)
}

# Create id_logits: Shape (batch_size, seq_length, num_classes)
id_logits = torch.randn(batch_size, seq_length, num_classes)

# Create dt_logits: Shape (batch_size, seq_length, num_classes)
dt_logits = torch.randn(batch_size, seq_length, num_classes)

ARGS = {
    'eos_id_tok': eos_id_tok,
    'eos_dt_tok': eos_dt_tok,
    'targets': targets,
    'id_logits': id_logits,
    'dt_logits': dt_logits
}
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 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
split_data_by_interval Ran
ran; our contract draft was wrong, not the code
a-antoniades/neuroformer/neuroformer/datasets.py
code served (permissive licence) · get_code("015233a9d8a82fc1")
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 seed for reproducibility
np.random.seed(0)

# Create a small array of intervals for testing
# intervals should be a numpy array that can be indexed and compared
intervals = np.array([
    [0.0, 1.0],
    [1.0, 2.0],
    [2.0, 3.0],
    [3.0, 4.0],
    [4.0, 5.0],
    [5.0, 6.0],
    [6.0, 7.0],
    [7.0, 8.0],
    [8.0, 9.0],
    [9.0, 10.0]
])

ARGS = {
    'intervals': intervals,
    'r_split': 0.8,
    'r_split_ft': 0.1
}
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
trial_df Ran
ran, honoured the contract we drafted
a-antoniades/neuroformer/neuroformer/data_utils.py
code served (permissive licence) · get_code("a9af76e30c2a8e46")
How it ran

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

import numpy as np
import pandas as pd

# Create test data: a list of 2D arrays where each array represents neuron data
# Shape: (num_rows, num_trials) for each neuron
data = [
    np.array([[1.5, 2.3], [3.1, 4.2], [0.5, 1.1]]),  # ID 0: 3 neurons x 2 trials
    np.array([[2.1, 3.4], [0.0, 5.5], [1.2, 2.8], [-0.5, 3.1]])  # ID 1: 4 neurons x 2 trials
]

ARGS = {
    'data': data
}
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 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
trial_df_combo3 Ran
ran; our contract draft was wrong, not the code
a-antoniades/neuroformer/neuroformer/data_utils.py
code served (permissive licence) · get_code("2afb4903a5098278")
How it ran

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

import numpy as np
import pandas as pd
import collections

# Create sample data structure
# The function expects data to be iterable (len(data))
# Each element can be either a numpy array or np.void
# The shape should be compatible with indexing [trial_no] and .flatten()

# Create two neurons' worth of data
neuron_1 = np.random.randn(5, 3)  # 5 timepoints, 3 trials
neuron_2 = np.random.randn(5, 3)  # 5 timepoints, 3 trials

data = [neuron_1, neuron_2]

ARGS = {
    "data": data,
    "n_stim": 0,
}
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 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
trial_df_real Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
a-antoniades/neuroformer/neuroformer/data_utils.py
code served (permissive licence) · get_code("8ad3185a7b0c4d25")
How it ran

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

import numpy as np
import pandas as pd
import collections

# Create sample data matching the function's expectations
# data: Number of neurons (N,) - a list/array where each element contains trial data

# Create 3 neurons worth of data
neuron_1_data = np.random.rand(4, 2)  # 4 time points, 2 trials
neuron_2_data = np.random.rand(4, 2)  # 4 time points, 2 trials
neuron_3_data = np.random.rand(4, 2)  # 4 time points, 2 trials

# Stack them into a structured array format that the function expects
# The function iterates through data[ID] and expects df.shape[-1] to be trials
data = [neuron_1_data, neuron_2_data, neuron_3_data]

ARGS = {
    "data": data
}
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 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
visnav_callback Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
a-antoniades/neuroformer/neuroformer/datasets.py
code served (permissive licence) · get_code("dc18b01aaf085b2c")
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 5, dtype family float, shape (1, 2, 1, 64, 112) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a small frames tensor: [n_frames, 1, 64, 112]
# Use distinct sizes for each dimension to make shapes unambiguous
frames = torch.randn(2, 1, 64, 112)

# Define frame_idx - should be within reasonable range of n_frames
frame_idx = 1

# Define n_frames - number of frames to extract
n_frames = 2

ARGS = {
    "frames": frames,
    "frame_idx": frame_idx,
    "n_frames": n_frames,
}
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 Tensor, ndim 5, dtype family float, shape (1, 2, 1, 64, 112) — shape (1, 2, 1, 64, 112) 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
nearest Not yet run
unverified (harvested, no recorded run)
a-antoniades/neuroformer/neuroformer/gather_atts.py
code served (permissive licence) · get_code("62b7dc6a0fcb3055")
How it ran

no invocation recorded.

reshape_features Not yet run
unverified (harvested, no recorded run)
a-antoniades/neuroformer/neuroformer/attention_vis.py
code served (permissive licence) · get_code("99fdd80b9ca7f208")
How it ran

no invocation recorded.

top_k_top_p_filtering Not yet run
unverified (harvested, no recorded run)
a-antoniades/neuroformer/neuroformer/simulation.py
code served (permissive licence) · get_code("120a5300accb00db")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 19 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 1 later) and 18 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
get_emb 11 other papers
4 dated earlier, 1 the same month, 6 later · 2 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need
2211.13220 (2022-11)
TetraDiffusion: Tetrahedral Diffusion Models for 3D Shape Generation
2303.15269 (2023-03)
Handwritten Text Generation from Visual Archetypes
and 8 more not named here

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

State-of-the-art systems neuroscience experiments yield large-scale multimodal data, and these data sets require new tools for analysis. Inspired by the success of large pretrained models in vision and language domains, we reframe the analysis of large-scale, cellular-resolution neuronal spiking data into an autoregressive spatiotemporal generation problem. Neuroformer is a multimodal, multitask generative pretrained transformer (GPT) model that is specifically designed to handle the intricacies of data in systems neuroscience. It scales linearly with feature size, can process an arbitrary number of modalities, and is adaptable to downstream tasks, such as predicting behavior. We first trained Neuroformer on simulated datasets, and found that it both accurately predicted simulated neuronal circuit activity, and also intrinsically inferred the underlying neural circuit connectivity, including direction. When pretrained to decode neural responses, the model predicted the behavior of a mouse with only few-shot fine-tuning, suggesting that the model begins learning how to do so directly from the neural representations themselves, without any explicit supervision. We used an ablation study to show that joint training on neuronal responses and behavior boosted performance, highlighting the model's ability to associate behavioral and neural representations in an unsupervised manner. These findings show that Neuroformer can analyze neural datasets and their emergent properties, informing the development of models and hypotheses associated with the brain.

For agents

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

get_harvested_code_for_paper("2311.00136")
get_code_for_paper("2311.00136")
have("2311.00136")

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

curl https://syntology.ai/api/ran/2311.00136.json

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

[![Syntology run record](https://syntology.ai/api/ran/2311.00136.svg)](https://syntology.ai/paper/2311.00136)

Connect an agent — have() is free.