SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2404.01220 · ICLR · 2024

Entity-Centric Reinforcement Learning for Object Manipulation from Pixels

Tal Daniel, Aviv Tamar, Dan Haramati

arXiv · PDF · Open in the Atlas

Code that ran

7 of 9 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 7 failures of our instrument; 2 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/2404.01220.json · badge: /api/ran/2404.01220.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
danhrmti/ecrl canonical (classifier label)
classifier basis: census
7 ran · 2 no recorded run
FunctionStatusWhere it lives
axisangle2quat Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
danhrmti/ecrl/isaac_panda_push_env.py
code served (permissive licence) · get_code("c6779ef568a575ea")
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 3, dtype family float, shape (2, 4, 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 torch

torch.manual_seed(0)

# Create a small axis-angle tensor with shape (..., 3)
# Using batch size 2 with 4 different shapes to test broadcasting
vec = torch.randn(2, 4, 3)

# eps parameter - default stability threshold
eps = 1e-06

ARGS = {
    "vec": vec,
    "eps": eps,
}
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 3, dtype family float, shape (2, 4, 4) — shape (2, 4, 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
conv2d Ran
ran; our contract draft was wrong, not the code
danhrmti/ecrl/slot_attention/slate_model.py
code served (permissive licence) · get_code("9330641b45f958fd")
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 Conv2d. 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 = {
    'in_channels': 3,
    'out_channels': 64,
    'kernel_size': 3,
    'stride': 1,
    'padding': 1,
    'dilation': 1,
    'groups': 1,
    'bias': True,
    'padding_mode': 'zeros',
    'weight_init': 'xavier'
}
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 Conv2d — a Conv2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv2d 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
convert_obs_to_dlp_rep Ran
ran; our contract draft was wrong, not the code
danhrmti/ecrl/chamfer_reward.py
code served (permissive licence) · get_code("62001b2163460d68")
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 3, dtype family float, shape (2, 3, 11) 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)

class MockDLPModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(12, 20)
    
    def encode_all(self, normalized_obs, deterministic=True):
        batch_size = normalized_obs.shape[0]
        z = torch.randn(batch_size, 2)
        mu_scale = torch.randn(batch_size, 2)
        mu_depth = torch.randn(batch_size, 1)
        mu_features = torch.randn(batch_size, 5)
        obj_on = torch.randint(0, 2, (batch_size,)).float()
        
        return {
            'z': z,
            'mu_scale': mu_scale,
            'mu_depth': mu_depth,
            'mu_features': mu_features,
            'obj_on': obj_on
        }

obs = torch.randint(0, 256, (2, 3, 4, 4), dtype=torch.uint8)
dlp_model = MockDLPModel()

ARGS = {
    'obs': obs,
    'dlp_model': dlp_model
}
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 3, dtype family float, shape (2, 3, 11) — shape (2, 3, 11) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
extract_dlp_features Ran
ran; our contract draft was wrong, not the code
danhrmti/ecrl/chamfer_reward.py
code served (permissive licence) · get_code("03234d185899f504")
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 3, dtype family float, shape (2, 2, 19) 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)

class MockDLPModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(12, 1)
    
    def encode_all(self, normalized_obs, deterministic=False):
        batch_size = normalized_obs.shape[0]
        output = {
            'z': torch.randn(batch_size, 2, 4),
            'mu_scale': torch.randn(batch_size, 2, 3),
            'mu_depth': torch.randn(batch_size, 2, 5),
            'mu_features': torch.randn(batch_size, 2, 6),
            'obj_on': torch.randn(batch_size, 2)
        }
        return output

dlp_model = MockDLPModel()
obs = torch.randint(0, 256, (2, 12), dtype=torch.uint8)

ARGS = {
    'obs': obs,
    'dlp_model': dlp_model
}
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 3, dtype family float, shape (2, 2, 19) — shape (2, 2, 19) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
extract_dlp_features Ran
ran; our contract draft was wrong, not the code
danhrmti/ecrl/latent_classifier.py
code served (permissive licence) · get_code("62186577f208bf25")
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
import torch.nn as nn

torch.manual_seed(0)

class MockDLPModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(12, 32)
    
    def encode_all(self, observations, deterministic=False):
        batch_size = observations.shape[0]
        z = torch.randn(batch_size, 2, 2)
        mu_scale = torch.randn(batch_size, 2, 3)
        mu_depth = torch.randn(batch_size, 2, 4)
        mu_features = torch.randn(batch_size, 2, 5)
        obj_on = torch.randn(batch_size, 2)
        cropped_objects = torch.randn(batch_size, 2, 3, 8, 8)
        
        return {
            'z': z,
            'mu_scale': mu_scale,
            'mu_depth': mu_depth,
            'mu_features': mu_features,
            'obj_on': obj_on,
            'cropped_objects': cropped_objects
        }

dlp_model = MockDLPModel()
obs = torch.randint(0, 256, (2, 3, 32, 32), dtype=torch.uint8)

ARGS = {
    'obs': obs,
    'dlp_model': dlp_model
}
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
gumbel_softmax Ran
ran; our contract draft was wrong, not the code
danhrmti/ecrl/slot_attention/slate_model.py
code served (permissive licence) · get_code("94c911439007d22c")
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 torch
import torch.nn.functional as F

torch.manual_seed(0)

# Create logits tensor with shape (batch=2, classes=4)
logits = torch.randn(2, 4)

# Define arguments for gumbel_softmax
ARGS = {
    "logits": logits,
    "tau": 1.0,
    "hard": False,
    "dim": -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 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
quat2axisangle Ran
ran; our contract draft was wrong, not the code
danhrmti/ecrl/isaac_panda_push_env.py
code served (permissive licence) · get_code("0a0911522ce533d0")
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, 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

torch.manual_seed(0)

# Create a small quaternion tensor with shape (2, 4)
# representing 2 quaternions in (x, y, z, w) format
quat = torch.tensor([
    [0.0, 0.0, 0.7071068, 0.7071068],  # 90 degree rotation around z-axis
    [0.0, 0.0, 0.0, 1.0]                # identity quaternion
], dtype=torch.float32)

ARGS = {
    "quat": quat
}
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, 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: none by the harness; any seed is in the setup block
get_dlp_rep Not yet run
unverified (harvested, no recorded run)
danhrmti/ecrl/chamfer_reward.py
code served (permissive licence) · get_code("649850da42fe3886")
How it ran

no invocation recorded.

get_dlp_rep Not yet run
unverified (harvested, no recorded run)
danhrmti/ecrl/latent_classifier.py
code served (permissive licence) · get_code("3219c7505517e845")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 9 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 9 later) and 0 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
gumbel_softmax 5 other papers
4 dated earlier, 1 later
2110.11405 (2021-10)
Illiterate DALL-E Learns to Compose
2205.14065 (2022-05)
Simple Unsupervised Object-Centric Learning for Complex and Naturalistic Videos
2211.01177 (2022-11)
Neural Systematic Binder
and 2 more not named here
axisangle2quat 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion
conv2d 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion
convert_obs_to_dlp_rep 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion
extract_dlp_features 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion
extract_dlp_features 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion
get_dlp_rep 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion
get_dlp_rep 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion
quat2axisangle 1 other paper
1 later
2602.02722 (2026-02)
Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion

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

Manipulating objects is a hallmark of human intelligence, and an important task in domains such as robotics. In principle, Reinforcement Learning (RL) offers a general approach to learn object manipulation. In practice, however, domains with more than a few objects are difficult for RL agents due to the curse of dimensionality, especially when learning from raw image observations. In this work we propose a structured approach for visual RL that is suitable for representing multiple objects and their interaction, and use it to learn goal-conditioned manipulation of several objects. Key to our method is the ability to handle goals with dependencies between the objects (e.g., moving objects in a certain order). We further relate our architecture to the generalization capability of the trained agent, based on a theoretical result for compositional generalization, and demonstrate agents that learn with 3 objects but generalize to similar tasks with over 10 objects. Videos and code are available on the project website: https://sites.google.com/view/entity-centric-rl

For agents

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

get_harvested_code_for_paper("2404.01220")
get_code_for_paper("2404.01220")
have("2404.01220")

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

curl https://syntology.ai/api/ran/2404.01220.json

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

[![Syntology run record](https://syntology.ai/api/ran/2404.01220.svg)](https://syntology.ai/paper/2404.01220)

Connect an agent — have() is free.