SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2307.13503 · ICML · 2023

Continuous Time Evidential Distributions for Irregular Time Series

Taylor Killian, Haoran Zhang, Thomas Hartvigsen, Ava Amini

arXiv · PDF · Open in the Atlas

Code that ran

12 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (5 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 7 have no contract verdict; 4 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/2307.13503.json · badge: /api/ran/2307.13503.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
twkillian/edict canonical (classifier label)
classifier basis: classifier_2025
11 ran · 3 no recorded run
salesforce/EDICT canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
GRUODECell Ran
ran on a synthesized input
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("e1c1df132395f35d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 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

# INIT_ARGS for GRUODECell.__init__
INIT_ARGS = {
    'hidden_size': 6,
    'bias': True
}

# ARGS for GRUODECell.forward
ARGS = {
    't': torch.tensor(0.5),  # scalar time value
    'h': torch.randn(2, 6)   # batch_size=2, hidden_size=6
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 2, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
GRUObsCell Ran
ran on a synthesized input · constructed an object; no result was computed
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("a1d575433e879380")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, GRUObsCell: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import math

torch.manual_seed(0)

# Model parameters
input_size = 3
hidden_size = 4
prep_hidden = 2
bias = True
dist_type = 'log_normal'
reweighting = False
rewt = 1.96
pop_mean = None
pop_std = None

# Create tensors for forward pass
batch_size = 2
num_obs = 1

# h: hidden state, shape (batch_size, hidden_size)
h = torch.randn(batch_size, hidden_size)

# p: parameters for distribution
# For 'log_normal', p is indexed by i_obs, so it should have shape (batch_size, 2*hidden_size)
# containing [mean, logvar] chunks
p = torch.randn(batch_size, 2 * hidden_size)

# X_obs: observed data, shape (num_obs, input_size)
X_obs = torch.randn(num_obs, input_size)

# M_obs: mask for observations, shape (num_obs, input_size)
M_obs = torch.ones(num_obs, input_size)

# i_obs: indices of observations, shape (num_obs,)
i_obs = torch.tensor([0], dtype=torch.long)

ARGS = {
    'input_size': input_size,
    'hidden_size': hidden_size,
    'prep_hidden': prep_hidden,
    'bias': bias,
    'dist_type': dist_type,
    'reweighting': reweighting,
    'rewt': rewt,
    'pop_mean': pop_mean,
    'pop_std': pop_std,
}
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 GRUObsCell — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GRUObsCell 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
MultVariateKLD Ran
ran on a synthesized input
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("733bcda5c09a4c19")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 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 torch

INIT_ARGS = {
    'reduction': 'mean',
}

ARGS = {
    'mu1': torch.randn(2, 4, dtype=torch.float32),
    'mu2': torch.randn(2, 4, dtype=torch.float32),
    'sigma1': torch.abs(torch.randn(2, 4, dtype=torch.float32)) + 0.1,
    'sigma2': torch.abs(torch.randn(2, 4, dtype=torch.float32)) + 0.1,
    'mask': None,
}

if __name__ == '__main__':
    model = torch.nn.Module()
    model.__class__ = type('MultVariateKLD', (torch.nn.Module,), {
        '__init__': lambda self, reduction: (
            super(self.__class__, self).__init__(),
            setattr(self, 'reduction', reduction)
        ),
        'forward': lambda self, mu1, mu2, sigma1, sigma2, mask=None, **kwargs: (
            (lambda mu1: (
                mu1.type(dtype=torch.float64),
                mu2.type(dtype=torch.float64),
                sigma1.type(dtype=torch.float64),
                sigma2.type(dtype=torch.float64),
            ) if not torch.any(mu1.isnan()) else (
                torch.nan_to_num(mu1, 0.0).type(dtype=torch.float64),
                mu2.type(dtype=torch.float64),
                sigma1.type(dtype=torch.float64),
                sigma2.type(dtype=torch.float64),
            )
            )(mu1)
        )[0] if False else (
            lambda: (
                mu1 := torch.nan_to_num(mu1, 0.0) if torch.any(mu1.isnan()) else mu1,
                mu1 := mu1.type(dtype=torch.float64),
                mu2 := mu2.type(dtype=torch.float64),
                sigma1 := sigma1.type(dtype=torch.float64),
                sigma2 := sigma2.type(dtype=torch.float64),
                sigma_diag_1 := torch.diag_embed(sigma1, offset=0, dim1=-2, dim2=-1),
                sigma_diag_2 := torch.diag_embed(sigma2, offset=0, dim1=-2, dim2=-1),
                sigma_diag_2_inv := sigma_diag_2.inverse(),
                mu_diff := (mask * (mu2 - mu1) if mask is not None else (mu2 - mu1)).unsqueeze(-1),
                term_1 := (sigma_diag_2.det() / sigma_diag_1.det()).log(),
                term_2 := torch.diagonal(torch.matmul(sigma_diag_2_inv, sigma_diag_1), dim1=-2, dim2=-1).sum(-1),
                term_3 := torch.matmul(torch.matmul(mu_diff.transpose(-2, -1), sigma_diag_2_inv), mu_diff).squeeze(),
                n := mu1.shape[-1],
                kl := 0.5 * (term_1 - n + term_2 + term_3),
                (torch.mean(kl) if self.reduction == 'mean' else torch.sum(kl, dim=0) if self.reduction == 'sum' else (_ for _ in ()).throw(NotImplementedError(f'Reduction type not implemented: {self.reduction}')))
            )[-1]
        )()
    })
    
    instance = torch.nn.Module()
    instance.__class__ = type('MultVariateKLD', (torch.nn.Module,), {
        '__init__': lambda self, reduction: (super(self.__class__, self).__init__(), setattr(self, 'reduction', reduction)),
        'forward': lambda self, mu1, mu2, sigma1, sigma2, mask=None, **kwargs: (
            mu1 := torch.nan_to_num(mu1, 0.0) if torch.any(mu1.isnan()) else mu1,
            mu1 := mu1.type(dtype=torch.float64),
            mu2 := mu2.type(dtype=torch.float64),
            sigma1 := sigma1.type(dtype=torch.float64),
            sigma2 := sigma2.type(dtype=torch.float64),
            sigma_diag_1 := torch.diag_embed(sigma1, offset=0, dim1=-2, dim2=-1),
            sigma_diag_2 := torch.diag_embed(sigma2, offset=0, dim1=-2, dim2=-1),
            sigma_diag_2_inv := sigma_diag_2.inverse(),
            mu_diff := (mask * (mu2 - mu1) if mask is not None else (mu2 - mu1)).unsqueeze(-1),
            term_1 := (sigma_diag_2.det() / sigma_diag_1.det()).log(),
            term_2 := torch.diagonal(torch.matmul(sigma_diag_2_inv, sigma_diag_1), dim1=-2, dim2=-1).sum(-1),
            term_3 := torch.matmul(torch.matmul(mu_diff.transpose(-2, -1), sigma_diag_2_inv), mu_diff).squeeze(),
            n := mu1.shape[-1],
            kl := 0.5 * (term_1 - n + term_2 + term_3),
            torch.mean(kl) if self.reduction == 'mean' else torch.sum(kl, dim=0) if self.reduction == 'sum' else (_ for _ in ()).throw(NotImplementedError(f'Reduction type not implemented: {self.reduction}'))
        )[-1]
    })(**INIT_ARGS)
    
    output = instance(**ARGS)
    print(output)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
clf_model Ran
ran on a synthesized input · constructed an object; no result was computed
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("fb4488a5ca614765")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, clf_model: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

torch.manual_seed(0)

ARGS = {
    "hidden_size": 8,
    "clf_hidden": 16,
    "output_dims": 3,
    "bias": True,
    "dropout_rate": 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 clf_model — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a clf_model 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_KL_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("d40a55d3b49160ad")
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
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors for testing
batch_size = 2
feature_dim = 4

# p_obs as a tuple (mean, lmbda, logvar, nu) for the MultVariateKLD path
mean = torch.randn(batch_size, feature_dim)
lmbda = torch.ones(batch_size, feature_dim)
logvar = torch.randn(batch_size, feature_dim)
nu = torch.tensor(feature_dim + 5.0)  # nu > mean.shape[-1] + 1
p_obs = (mean, lmbda, logvar, nu)

# Observed data
X_obs = torch.randn(batch_size, feature_dim)

# Mask
M_obs = torch.ones(batch_size, feature_dim)

# Noise standard deviation
obs_noise_std = 0.01

# logvar flag
logvar = True

ARGS = {
    'p_obs': p_obs,
    'X_obs': X_obs,
    'M_obs': M_obs,
    'obs_noise_std': obs_noise_std,
    'logvar': logvar
}
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
covariates_mapping Ran
ran on a synthesized input · constructed an object; no result was computed
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("0a36bfc002a6f50f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, covariates_mapping: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

torch.manual_seed(0)

ARGS = {
    'cov_size': 4,
    'cov_hidden': 8,
    'hidden_size': 6,
    'bias': True,
    'dropout_rate': 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 covariates_mapping — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a covariates_mapping 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
gaussian_KL Ran
ran; our contract draft was wrong, not the code
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("2d704d92dcce4865")
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)

# Create small tensors for the Gaussian KL divergence function
# The function expects 4 tensors: mu1, mu2, sigma1, sigma2
# All should be the same shape (batch_size, feature_dim)
batch_size = 2
feature_dim = 4

mu1 = torch.randn(batch_size, feature_dim)
mu2 = torch.randn(batch_size, feature_dim)
sigma1 = torch.abs(torch.randn(batch_size, feature_dim)) + 0.1  # Ensure positive
sigma2 = torch.abs(torch.randn(batch_size, feature_dim)) + 0.1  # Ensure positive

# mask is optional; create a binary mask of the same shape
mask = torch.ones(batch_size, feature_dim)

ARGS = {
    "mu1": mu1,
    "mu2": mu2,
    "sigma1": sigma1,
    "sigma2": sigma2,
    "mask": mask
}
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
im_to_np Ran
ran, violated the contract we drafted
salesforce/EDICT/edict_functions.py
code served (permissive licence) · get_code("4985a054efea36d0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned ndarray, ndim 3, dtype family float, shape (3, 4, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
from PIL import Image

# Create a small test image
# Using PIL Image as a common image type that can be converted to numpy array
test_image = Image.new('RGB', (4, 3), color=(128, 64, 192))

ARGS = {
    'im': test_image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
PIL, 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, 4, 3) — shape (3, 4, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
log_normal_dist Ran
ran on a synthesized input · constructed an object; no result was computed
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("94e7492c3892a050")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, log_normal_dist: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

torch.manual_seed(0)

ARGS = {
    "hidden_size": 8,
    "p_hidden": 16,
    "input_size": 4,
    "bias": True,
    "dropout_rate": 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 log_normal_dist — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a log_normal_dist 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
log_normal_nll Ran
ran; our contract draft was wrong, not the code
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("2218dbc19892f617")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch

torch.manual_seed(0)

# Create small tensors for the function
batch_size = 2
dim = 4

# X: observations tensor
X = torch.randn(batch_size, dim)

# mean: predicted mean
mean = torch.randn(batch_size, dim)

# logvar: log variance
logvar = torch.randn(batch_size, dim)

# masks: binary mask tensor (same shape as X)
masks = torch.ones(batch_size, dim)

ARGS = {
    "X": X,
    "mean": mean,
    "logvar": logvar,
    "masks": masks
}
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 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
niw_nll Ran
ran; our contract draft was wrong, not the code
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("ef8997b67ce66d1a")
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, 1) 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)

# Define tensor dimensions
batch_size = 2
dim = 4

# X: shape (batch_size, dim) - data points
X = torch.randn(batch_size, dim)

# mu: shape (batch_size, dim) - mean parameters
mu = torch.randn(batch_size, dim)

# logvar: shape (batch_size, dim) - log variance parameters
logvar = torch.randn(batch_size, dim)

# lmbda: scalar or tensor - precision parameter
lmbda = torch.tensor(2.0)

# nu: scalar or tensor - degrees of freedom parameter
nu = torch.tensor(10.0)

# masks: optional, shape (batch_size, dim) - binary masks
masks = torch.ones(batch_size, dim)

ARGS = {
    "X": X,
    "mu": mu,
    "logvar": logvar,
    "lmbda": lmbda,
    "nu": nu,
    "masks": masks,
}
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 2, dtype family float, shape (2, 1) — shape (2, 1) 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
normal_inverse_wishart Ran
ran on a synthesized input · constructed an object; no result was computed
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("4cd174ef548a9bad")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, normal_inverse_wishart: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

torch.manual_seed(0)

ARGS = {
    "hidden_size": 8,
    "p_hidden": 6,
    "input_size": 4,
    "bias": True,
    "dropout_rate": 0.1,
    "eps": 1e-05,
}
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 normal_inverse_wishart — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a normal_inverse_wishart 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
GRUODEBayes Not yet run
unverified (harvested, no recorded run)
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("65250a77127a6505")
How it ran

no invocation recorded.

init_weights Not yet run
unverified (harvested, no recorded run)
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("9f7fe82102a75b79")
How it ran

no invocation recorded.

niw_reg_error Not yet run
unverified (harvested, no recorded run)
twkillian/edict/model_utils.py
code served (permissive licence) · get_code("1ee3d0ea5c3d3968")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 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 15 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

Prevalent in many real-world settings such as healthcare, irregular time series are challenging to formulate predictions from. It is difficult to infer the value of a feature at any given time when observations are sporadic, as it could take on a range of values depending on when it was last observed. To characterize this uncertainty we present EDICT, a strategy that learns an evidential distribution over irregular time series in continuous time. This distribution enables well-calibrated and flexible inference of partially observed features at any time of interest, while expanding uncertainty temporally for sparse, irregular observations. We demonstrate that EDICT attains competitive performance on challenging time series classification tasks and enabling uncertaintyguided inference when encountering noisy data. Code is available at https://github.com/ twkillian/EDICT.

For agents

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

get_harvested_code_for_paper("2307.13503")
get_code_for_paper("2307.13503")
have("2307.13503")

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

curl https://syntology.ai/api/ran/2307.13503.json

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

[![Syntology run record](https://syntology.ai/api/ran/2307.13503.svg)](https://syntology.ai/paper/2307.13503)

Connect an agent — have() is free.