SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2605.14937 · 2026

Slot-MPC: Goal-Conditioned Model Predictive Control with Object-Centric Representations

Angel Villar-Corrales, Sven Behnke, Jonathan Spieler

arXiv · PDF · Open in the Atlas

Code that ran

4 of 11 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 1 failure of our instrument; 7 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/2605.14937.json · badge: /api/ran/2605.14937.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
angelvillar96/PlaySlot community
Papers with Code archive
2 ran · 6 no recorded run
gaoyuezhou/dino_wm community
Papers with Code archive
2 ran · 1 no recorded run
FunctionStatusWhere it lives
MLP Ran
ran on a synthesized input
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("0ea0cd54ea1ac540")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

MPCPlanner Ran
ran on a synthesized input · constructed an object; no result was computed
gaoyuezhou/dino_wm/planning/mpc.py
code served (permissive licence) · get_code("5cd5218937e3dfc4")
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, MPCPlanner: 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 numpy as np
from unittest.mock import Mock, MagicMock
import tempfile

torch.manual_seed(0)

# Mock World Model
class MockWorldModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x):
        return self.linear(x)

# Mock Objective Function
class MockObjectiveFn(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def forward(self, x):
        return self.linear(x)

# Mock Preprocessor
class MockPreprocessor:
    def normalize_actions(self, actions):
        return actions * 0.1

# Mock Evaluator
class MockEvaluator:
    def __init__(self):
        self.frameskip = 2
    
    def get_init_cond(self):
        obs_0 = {
            'visual': torch.randn(2, 1, 4),
        }
        state_0 = torch.randn(2, 4)
        return obs_0, state_0
    
    def assign_init_cond(self, obs_0, state_0):
        pass
    
    def eval_actions(self, actions, action_len, filename=None, save_video=False):
        batch_size = actions.shape[0]
        logs = {'reward': 0.5, 'loss': 0.1}
        successes = np.array([True, False])
        e_obses = {
            'visual': torch.randn(batch_size, 5, 4),
        }
        e_states = torch.randn(batch_size, 5, 4)
        return logs, successes, e_obses, e_states

# Mock Sub-Planner
class MockSubPlanner:
    def __init__(self, wm, action_dim, objective_fn, preprocessor, evaluator, wandb_run, log_filename, **kwargs):
        self.wm = wm
        self.action_dim = action_dim
        self.objective_fn = objective_fn
        self.preprocessor = preprocessor
        self.evaluator = evaluator
        self.wandb_run = wandb_run
        self.log_filename = log_filename
        self.logging_prefix = 'plan_0'
    
    def plan(self, obs_0, obs_g, actions=None):
        batch_size = obs_0['visual'].shape[0]
        planned_actions = torch.randn(batch_size, 8, 3)
        info = {}
        return planned_actions, info

# Mock Environment
class MockEnv:
    def reset(self):
        return {'visual': torch.randn(2, 4)}
    
    def step(self, action):
        return {'visual': torch.randn(2, 4)}, 0, False, {}

# Mock wandb_run
class MockWandBRun:
    def log(self, logs):
        pass

# Create instances
wm = MockWorldModel()
action_dim = 3
objective_fn = MockObjectiveFn()
preprocessor = MockPreprocessor()
evaluator = MockEvaluator()
wandb_run = MockWandBRun()
env = MockEnv()

# Create temp log file
log_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.json')
log_filename = log_file.name
log_file.close()

# Sub-planner config
sub_planner_config = {
    'target': 'MockSubPlanner',
    '_target_': 'MockSubPlanner',
}

# Mock hydra.utils.instantiate to return MockSubPlanner
import hydra.utils
original_instantiate = hydra.utils.instantiate
def mock_instantiate(config, **kwargs):
    return MockSubPlanner(**kwargs)

hydra.utils.instantiate = mock_instantiate

ARGS = {
    'max_iter': 2,
    'n_taken_actions': 2,
    'sub_planner': sub_planner_config,
    'wm': wm,
    'env': env,
    'action_dim': action_dim,
    'objective_fn': objective_fn,
    'preprocessor': preprocessor,
    'evaluator': evaluator,
    'wandb_run': wandb_run,
    'logging_prefix': 'mpc',
    'log_filename': log_filename,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, hydra, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MPCPlanner — 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 MPCPlanner 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
MultiHeadSelfAttention Ran
ran on a synthesized input
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("0987ef4fdb7f10f2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

slice_trajdict_with_t Ran
ran; our contract draft was wrong, not the code
gaoyuezhou/dino_wm/planning/mpc.py
code served (permissive licence) · get_code("f6fdf774c83a40fc")
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 numpy as np

torch.manual_seed(0)

# Construct a data_dict with trajectory data
# The function expects a dict where values are arrays/tensors with shape (batch, time, ...)
# It slices along the time dimension (index 1)

data_dict = {
    'positions': torch.randn(2, 5, 3),      # batch=2, time=5, features=3
    'velocities': torch.randn(2, 5, 3),     # batch=2, time=5, features=3
    'actions': torch.randn(2, 5, 4),        # batch=2, time=5, action_dim=4
}

ARGS = {
    'data_dict': data_dict,
    'start_idx': 1,
    'end_idx': 4,
    'step': 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 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
BasePlanner Not yet run
unverified (harvested, no recorded run)
gaoyuezhou/dino_wm/planning/mpc.py
code served (permissive licence) · get_code("8fb3f995975d71d9")
How it ran

no invocation recorded.

BaseSlotDynamicsModel Not yet run
unverified (harvested, no recorded run)
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("121c7e805ff2052e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

MarkovTransformerDynamicsModel Not yet run
unverified (harvested, no recorded run)
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("1778f2dd486a95f5")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

MetaAttention Not yet run
unverified (harvested, no recorded run)
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("413506a9977868cb")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

MultiHeadCrossAttention Not yet run
unverified (harvested, no recorded run)
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("e46866d14747394f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

TransformerBlock Not yet run
unverified (harvested, no recorded run)
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("39f80a1d11dadcea")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

TransformerDecoder Not yet run
unverified (harvested, no recorded run)
angelvillar96/PlaySlot/src/models/Predictors/DynamicsModels.py
pointer only (licence: NONE) · get_code("53b95fb58c84e27e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, angelvillar96/PlaySlot when it was checked. 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.

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

Predictive world models enable agents to model scene dynamics and reason about the consequences of their actions. Inspired by human perception, object-centric world models capture scene dynamics using object-level representations, which can be used for downstream applications such as action planning. However, most objectcentric world models and reinforcement learning (RL) approaches learn reactive policies that are fixed at inference time, limiting generalization to novel situations. We propose Slot-MPC, an object-centric world modeling framework that enables planning through Model Predictive Control (MPC). Slot-MPC leverages vision encoders to learn slot-based representations, which encode individual objects in the scene, and uses these structured representations to learn an action-conditioned object-centric dynamics model. At inference time, the learned dynamics model enables action planning via MPC, allowing agents to adapt to previously unseen situations. Since the learned world model is differentiable, we can use gradient-based MPC to directly optimize actions, which is computationally more efficient than relying on gradient-free, sampling-based MPC methods. Experiments on simulated robotic manipulation tasks show that Slot-MPC improves both task performance and planning efficiency compared to non-object-centric world model baselines. In the considered offline setting with limited state-action coverage, we find that gradient-based MPC performs better than gradient-free, sampling-based MPC. Our results demonstrate that explicitly structured, object-centric representations provide a strong inductive bias for controllable and generalizable decision-making. Code and additional results are available at https://slot-mpc.github.io.

For agents

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

get_harvested_code_for_paper("2605.14937")
get_code_for_paper("2605.14937")
have("2605.14937")

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

curl https://syntology.ai/api/ran/2605.14937.json

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

[![Syntology run record](https://syntology.ai/api/ran/2605.14937.svg)](https://syntology.ai/paper/2605.14937)

Connect an agent — have() is free.