SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2304.04321 · ICCV · 2023

ARNOLD: A Benchmark for Language-Grounded Task Learning With Continuous States in Realistic 3D Scenes

Song-Chun Zhu, Demetri Terzopoulos, Siyuan Huang, Xiaofeng Gao, Yizhou Zhao, Baoxiong Jia, Qingyang Wu, Jiangyong Huang, Haoran Geng, Ran Gong, Ziheng Zhou, Wensi Ai

arXiv · PDF · Open in the Atlas

Code that ran

9 of 11 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, 2 violated it and 3 have no contract verdict; 4 failures of our instrument; 2 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/2304.04321.json · badge: /api/ran/2304.04321.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
arnold-benchmark/arnold canonical (classifier label)
classifier basis not recorded on the link
9 ran · 2 no recorded run
FunctionStatusWhere it lives
act_layer Ran
ran; our contract draft was wrong, not the code
arnold-benchmark/arnold/bc_z/blocks.py
code served (permissive licence) · get_code("c99fb7b9654ea364")
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 ReLU. 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)

LRELU_SLOPE = 0.02

def act_layer(act):
    if act == 'relu':
        return nn.ReLU()
    elif act == 'lrelu':
        return nn.LeakyReLU(LRELU_SLOPE)
    elif act == 'elu':
        return nn.ELU()
    elif act == 'tanh':
        return nn.Tanh()
    elif act == 'prelu':
        return nn.PReLU()
    else:
        raise ValueError('%s not recognized.' % act)

ARGS = {
    'act': 'relu'
}
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 ReLU — a ReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ReLU 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
cache_fn Ran
ran; our contract draft was wrong, not the code
arnold-benchmark/arnold/peract/network.py
pointer only (licence: MIT); the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("8458f9ba0361ff43")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ThomasMrY/VCT, and GitHub detected no licence in that repository 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.

default Ran
ran, violated the contract we drafted
arnold-benchmark/arnold/peract/network.py
pointer only (licence: MIT); the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("60fff7c3c400d7ff")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ThomasMrY/VCT, and GitHub detected no licence in that repository 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.

exists Ran
ran, violated the contract we drafted
arnold-benchmark/arnold/peract/network.py
pointer only (licence: MIT); the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("aa5486a3650902d8")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ThomasMrY/VCT, and GitHub detected no licence in that repository 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.

load_data Ran
ran on a synthesized input
arnold-benchmark/arnold/ckpt_selection.py
code served (permissive licence) · get_code("e3f04ff9250f8f3f")
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 list, 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 tempfile
import numpy as np
from pathlib import Path

# Create a temporary directory with some dummy .npz files
temp_dir = tempfile.mkdtemp()

# Create sample npz files
npz_data_1 = {"array1": np.array([1, 2, 3]), "array2": np.array([[4, 5], [6, 7]])}
np.savez(Path(temp_dir) / "demo_001.npz", **npz_data_1)

npz_data_2 = {"array1": np.array([8, 9, 10]), "array2": np.array([[11, 12], [13, 14]])}
np.savez(Path(temp_dir) / "demo_002.npz", **npz_data_2)

npz_data_3 = {"array1": np.array([15, 16, 17]), "array2": np.array([[18, 19], [20, 21]])}
np.savez(Path(temp_dir) / "demo_003.npz", **npz_data_3)

# Create a subdirectory to ensure it's filtered out
subdir = Path(temp_dir) / "subdir"
subdir.mkdir()

ARGS = {
    "data_path": temp_dir
}
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 list, ndim 2, dtype family float — a list 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
norm_layer1d Ran
ran; our contract draft was wrong, not the code
arnold-benchmark/arnold/bc_z/blocks.py
code served (permissive licence) · get_code("9ed9933be23b28a5")
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 BatchNorm1d. 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 = {
    'norm': 'batch',
    'num_channels': 64
}
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 BatchNorm1d — a BatchNorm1d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a BatchNorm1d 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
norm_layer2d Ran
ran; our contract draft was wrong, not the code
arnold-benchmark/arnold/bc_z/blocks.py
code served (permissive licence) · get_code("d0264b5831943fb9")
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 BatchNorm2d. 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 = {
    'norm': 'batch',
    'channels': 16
}
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 BatchNorm2d — a BatchNorm2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a BatchNorm2d 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
sec_to_str Ran
ran on a synthesized input
arnold-benchmark/arnold/train_cliport6d.py
code served (permissive licence) · get_code("4e87462cc7d896d2")
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 datetime

def sec_to_str(delta):
    t = datetime.timedelta(seconds=delta)
    s = str(t)
    return s.split('.')[0] + 's'

ARGS = {
    'delta': 3661.5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — 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
val Ran
ran on a synthesized input
arnold-benchmark/arnold/train_cliport6d.py
code served (permissive licence) · get_code("e241a1dff2e417df")
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 float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
from scipy.spatial.transform import Rotation as R
import torch.nn as nn

torch.manual_seed(0)

class MockModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(10, 1)
    
    def forward(self, inp):
        # Return a dict of loss tensors
        return {
            'loss_1': torch.tensor(0.5),
            'loss_2': torch.tensor(0.3),
        }
    
    def eval(self):
        super().eval()
        return self

class MockConfig:
    def __init__(self):
        self.batch_size = 2
        self.epochs = 10

def create_sample_batch_data():
    """Create a single sample batch data dict"""
    batch_size = 2
    pixel_size = 0.01
    
    bounds = np.random.rand(3, 3)  # 3D bounds
    img = np.random.rand(64, 64, 3)
    language = "move the object"
    attention_points = np.array([0.5, 0.5, 0.5])
    target_points = np.array([0.6, 0.6, 0.6, 0.0, 0.0, 0.0, 1.0])  # position + quaternion
    
    return {
        'pixel_size': pixel_size,
        'bounds': bounds,
        'img': img,
        'language': language,
        'attention_points': attention_points,
        'target_points': target_points,
    }

def create_data_loader():
    """Create a simple data loader as a list of batches"""
    batch_data_1 = [create_sample_batch_data(), create_sample_batch_data()]
    batch_data_2 = [create_sample_batch_data(), create_sample_batch_data()]
    return [batch_data_1, batch_data_2]

data_loader = create_data_loader()
model = MockModel()
cfg = MockConfig()
epoch = 0

ARGS = {
    'data_loader': data_loader,
    'model': model,
    'cfg': cfg,
    'epoch': epoch,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
load_ckpt Not yet run
unverified (harvested, no recorded run)
arnold-benchmark/arnold/ckpt_selection.py
code served (permissive licence) · get_code("0800b4fef63dbf21")
How it ran

no invocation recorded.

parse_state Not yet run
unverified (harvested, no recorded run)
arnold-benchmark/arnold/dataset.py
code served (permissive licence) · get_code("bab389ab808463dc")
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, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 2 the same month, 6 later) and 5 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
exists 486 other papers
166 dated earlier, 7 the same month, 307 later, 4 same year, order not determinable, 2 undated · 78 only as an assembly brick
1603.08155 (2016-03)
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
1611.07004 (2016-11)
Image-to-Image Translation with Conditional Adversarial Networks
1804.04637 (2018-04)
EMBER: An Open Dataset for Training Static PE Malware Machine Learning Models
and 483 more not named here
default 163 other papers
80 dated earlier, 1 the same month, 80 later, 1 same year, order not determinable, 1 undated · 27 only as an assembly brick
1603.08155 (2016-03)
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
1810.00825 (2018-10)
Set Transformer: A Framework for Attention-based Permutation-Invariant Neural Networks
1907.01470 (2019-07)
Augmenting Self-attention with Persistent Memory
and 160 more not named here
cache_fn 9 other papers
6 dated earlier, 3 later · 2 only as an assembly brick
1810.00825 (2018-10)
Set Transformer: A Framework for Attention-based Permutation-Invariant Neural Networks
2103.03206 (2021-03)
Perceiver: General Perception with Iterative Attention
2107.14795 (2021-07)
Perceiver IO: A General Architecture for Structured Inputs & Outputs
and 6 more not named here
act_layer 5 other papers
1 dated earlier, 4 later
2207.11514 (2022-07)
Semantic Abstraction: Open-World 3D Scene Understanding from 2D Vision-Language Models
2306.10474 (2023-06)
A Universal Semantic-Geometric Representation for Robotic Manipulation
2308.16891 (2023-08)
GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fields
and 2 more not named here
norm_layer1d 5 other papers
1 dated earlier, 4 later
2207.11514 (2022-07)
Semantic Abstraction: Open-World 3D Scene Understanding from 2D Vision-Language Models
2306.10474 (2023-06)
A Universal Semantic-Geometric Representation for Robotic Manipulation
2308.16891 (2023-08)
GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fields
and 2 more not named here
norm_layer2d 5 other papers
1 dated earlier, 4 later
2207.11514 (2022-07)
Semantic Abstraction: Open-World 3D Scene Understanding from 2D Vision-Language Models
2306.10474 (2023-06)
A Universal Semantic-Geometric Representation for Robotic Manipulation
2308.16891 (2023-08)
GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fields
and 2 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

demonstrations with template-generated language descriptions. We assess task performance by utilizing the latest language-conditioned policy learning models. Our results indicate that current models for language-conditioned manipulations continue to experience significant challenges in novel goal-state generalizations, scene generalizations, and object generalizations. These findings highlight the need to develop new algorithms that address this gap and underscore the potential for further research in this area. Project website: https://arnold-benchmark.github.io.

For agents

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

get_harvested_code_for_paper("2304.04321")
get_code_for_paper("2304.04321")
have("2304.04321")

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

curl https://syntology.ai/api/ran/2304.04321.json

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

[![Syntology run record](https://syntology.ai/api/ran/2304.04321.svg)](https://syntology.ai/paper/2304.04321)

Connect an agent — have() is free.