SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.13391 · NeurIPS · 2020

Learning Disentangled Representations of Video with Missing Data

Rose Yu, Chi Zhang, Armand Comas Massagué, Zlatan Feric

arXiv · PDF · Open in the Atlas

Code that ran

5 of 11 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 0 failures of our instrument; 6 no recorded run), measured 2026-11-22; 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/2006.13391.json · badge: /api/ran/2006.13391.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
Rose-STL-Lab/DIVE marked official in Papers with Code
Papers with Code archive
4 ran · 3 no recorded run
jthsieh/DDPAE-video-prediction community
Papers with Code archive
1 ran · 3 no recorded run
FunctionStatusWhere it lives
AppearanceEncoder Ran
ran on a synthesized input · constructed an object; no result was computed
Rose-STL-Lab/DIVE/models/DiveModel.py
code served (permissive licence) · get_code("65056688f9a348ae")
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, AppearanceEncoder: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Define the AppearanceEncoder class
class AppearanceEncoder(nn.Module):
    """
    Encode the appearance hidden vectors.
    """

    def __init__(self, input_size, hidden_size, stat_app_size, var_app_size, n_frames_output, num_layers=2):
        super(AppearanceEncoder, self).__init__()
        self.encoder = nn.LSTM(input_size=input_size + hidden_size, hidden_size=hidden_size, num_layers=num_layers, batch_first=True)
        self.predictor = nn.LSTM(input_size=hidden_size, hidden_size=hidden_size, num_layers=num_layers, batch_first=True)
        self.stat_app_layer = nn.Linear(hidden_size, stat_app_size)
        self.var_app_layer = nn.Linear(hidden_size, var_app_size)
        self.var_app_layer_0 = nn.Linear(hidden_size + stat_app_size, hidden_size)
        self.var_app_ini_layer = nn.Linear(hidden_size, var_app_size)
        self.var_app_ini_layer_0 = nn.Linear(input_size + stat_app_size, hidden_size)
        self.lrelu = nn.LeakyReLU()
        self.input_size = input_size
        self.hidden_size = hidden_size
        self.n_frames_output = n_frames_output

    def accumulate_varying_appearance(self, ini_cond, var_app):
        var_app_output = ini_cond
        var_app_outputs = []
        for t in range(var_app.shape[1]):
            var_app_output = var_app_output + var_app[:, t:t + 1]
            var_app_outputs.append(var_app_output)
        var_app_output = torch.cat(var_app_outputs, dim=1)
        return var_app_output

    def forward(self, input, prev_obj_hidden):
        """
        input: temporal factor of size                        (batch_size, n_frames, input_size)
        prev_obj_hidden: hidden state of the previous object  (batch_size, n_frames, hidden_size)
        stat_app_output: static appearance                    (batch_size, n_frames, output_size)
        var_app_output: varying appearance in time            (batch_size, n_frames, output_size)
        """
        input_rnn = torch.cat([input, prev_obj_hidden], dim=-1)
        encoder_output, hidden = self.encoder(input_rnn)
        last_output = encoder_output[:, -1, :]
        stat_app_output = self.stat_app_layer(last_output)
        prev_obj_hidden = encoder_output
        pred_outputs = []
        pred_rnn_input = encoder_output[:, -1:]
        for t in range(self.n_frames_output):
            pred_encoder_output, pred_hidden = self.predictor(pred_rnn_input)
            pred_rnn_input = pred_encoder_output
            pred_outputs.append(pred_encoder_output)
        pred_output = torch.cat(pred_outputs, dim=1)
        initial_conditions = self.var_app_ini_layer(self.lrelu(self.var_app_ini_layer_0(torch.cat([input[:, :1], stat_app_output.unsqueeze(1)], dim=-1))))
        encoder_output = torch.cat([encoder_output, pred_output], dim=1)
        var_app_output = self.var_app_layer(self.lrelu(self.var_app_layer_0(torch.cat([encoder_output, stat_app_output.unsqueeze(1).repeat(1, encoder_output.size(1), 1)], dim=-1))))
        var_app_output = self.accumulate_varying_appearance(initial_conditions, var_app_output)
        return (stat_app_output, var_app_output, prev_obj_hidden)


# Construct the arguments for the forward method
batch_size = 2
n_frames = 3
input_size = 4
hidden_size = 5
stat_app_size = 6
var_app_size = 7
n_frames_output = 2
num_layers = 2

# Create the model instance
model = AppearanceEncoder(
    input_size=input_size,
    hidden_size=hidden_size,
    stat_app_size=stat_app_size,
    var_app_size=var_app_size,
    n_frames_output=n_frames_output,
    num_layers=num_layers
)

# Create input tensors for the forward pass
input_tensor = torch.randn(batch_size, n_frames, input_size)
prev_obj_hidden_tensor = torch.randn(batch_size, n_frames, hidden_size)

# Define ARGS dictionary with all constructor parameters
ARGS = {
    'input_size': input_size,
    'hidden_size': hidden_size,
    'stat_app_size': stat_app_size,
    'var_app_size': var_app_size,
    'n_frames_output': n_frames_output,
    'num_layers': num_layers
}
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 AppearanceEncoder — 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 AppearanceEncoder 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
BaseModel Ran
ran on a synthesized input · constructed an object; no result was computed
Rose-STL-Lab/DIVE/models/DiveModel.py
code served (permissive licence) · get_code("c72a806093b6179a")
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, BaseModel: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import pyro
import pyro.distributions as dist
import tempfile
import os

torch.manual_seed(0)

# Create a simple neural network module for testing
class SimpleNet(nn.Module):
    def __init__(self):
        super(SimpleNet, self).__init__()
        self.conv1 = nn.Conv2d(3, 16, kernel_size=3, padding=1)
        self.linear = nn.Linear(16, 8)
        self.bn = nn.BatchNorm2d(16)
    
    def forward(self, x):
        return x

# Since BaseModel.__init__ takes no parameters, we just need to verify
# the setup can be called and all methods work
ARGS = {}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pyro, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type BaseModel — 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 BaseModel 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
ImageDecoder Ran
ran on a synthesized input · constructed an object; no result was computed
Rose-STL-Lab/DIVE/models/DiveModel.py
code served (permissive licence) · get_code("183ce4a6d6e9884e")
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, ImageDecoder: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'input_size': 128,
    'n_channels': 3,
    'ngf': 64,
    'n_layers': 4,
    'activation': 'tanh'
}
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 ImageDecoder — 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 ImageDecoder 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
ImageEncoder Ran
ran on a synthesized input · constructed an object; no result was computed
Rose-STL-Lab/DIVE/models/DiveModel.py
code served (permissive licence) · get_code("10f313642e413ed5")
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, ImageEncoder: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "n_channels": 3,
    "output_size": 128,
    "ngf": 64,
    "n_layers": 4,
}
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 ImageEncoder — 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 ImageEncoder 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
SequenceEncoder Ran
ran on a synthesized input
jthsieh/DDPAE-video-prediction/models/DDPAE.py
code served (permissive licence) · get_code("db946948e9cb7afc")
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, shape (2, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

class SequenceEncoder(nn.Module):
    """
  Encode a sequence of input vectors into a single vector.
  """

    def __init__(self, input_size, hidden_size, output_size, num_layers=1):
        super(SequenceEncoder, self).__init__()
        self.encoder = nn.LSTM(input_size=input_size, hidden_size=hidden_size, num_layers=num_layers, batch_first=True)
        self.out_layer = nn.Linear(hidden_size, output_size)
        self.input_size = input_size
        self.hidden_size = hidden_size

    def forward(self, input):
        """
    input: temporal factor of size (batch_size, n_frames, input_size)
    output: prediction of size (batch_size, n_frames, output_size)
    """
        encoder_output, hidden = self.encoder(input)
        last_output = encoder_output[:, -1, :]
        output = self.out_layer(last_output)
        return output


INIT_ARGS = {
    'input_size': 4,
    'hidden_size': 8,
    'output_size': 3,
    'num_layers': 2,
}

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

if __name__ == '__main__':
    model = SequenceEncoder(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 3)")
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 2, dtype family float, shape (2, 3) — shape (2, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
BaseModel Not yet run
unverified (harvested, no recorded run)
jthsieh/DDPAE-video-prediction/models/DDPAE.py
code served (permissive licence) · get_code("50fa86d9ffb2c790")
How it ran

no invocation recorded.

DDPAE Not yet run
unverified (harvested, no recorded run)
jthsieh/DDPAE-video-prediction/models/DDPAE.py
code served (permissive licence) · get_code("dc0a8ab81a302f82")
How it ran

no invocation recorded.

DiveModel Not yet run
unverified (harvested, no recorded run)
Rose-STL-Lab/DIVE/models/DiveModel.py
code served (permissive licence) · get_code("5ec9e41854970f62")
How it ran

no invocation recorded.

PoseEncoder Not yet run
unverified (harvested, no recorded run)
Rose-STL-Lab/DIVE/models/DiveModel.py
code served (permissive licence) · get_code("0654f8275a9e6fb2")
How it ran

no invocation recorded.

PoseRNN Not yet run
unverified (harvested, no recorded run)
jthsieh/DDPAE-video-prediction/models/DDPAE.py
code served (permissive licence) · get_code("07b1bf020bd68c41")
How it ran

no invocation recorded.

init_weights Not yet run
unverified (harvested, no recorded run)
Rose-STL-Lab/DIVE/models/DiveModel.py
code served (permissive licence) · get_code("1da6f0594ca2dc30")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 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 11 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

Missing data poses significant challenges while learning representations of video sequences. We present Disentangled Imputed Video autoEncoder (DIVE), a deep generative model that imputes and predicts future video frames in the presence of missing data. Specifically, DIVE introduces a missingness latent variable, disentangles the hidden video representations into static and dynamic appearance, pose, and missingness factors for each object. DIVE imputes each object's trajectory where the data is missing. On a moving MNIST dataset with various missing scenarios, DIVE outperforms the state of the art baselines by a substantial margin. We also present comparisons on a real-world MOTSChallenge pedestrian dataset, which demonstrates the practical value of our method in a more realistic setting. Our code and data can be found at https://github.com/Rose-STL-Lab/DIVE.

For agents

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

get_harvested_code_for_paper("2006.13391")
get_code_for_paper("2006.13391")
have("2006.13391")

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

curl https://syntology.ai/api/ran/2006.13391.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.13391.svg)](https://syntology.ai/paper/2006.13391)

Connect an agent — have() is free.