SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2306.06359 · ICML · 2023

NeRFool: Uncovering the Vulnerability of Generalizable Neural Radiance Fields against Adversarial Perturbations

Ye Yuan, Yonggan Fu, Shang Wu, Shunyao Zhang, Celine Lin, Souvik Kundu

arXiv · PDF · Open in the Atlas

Code that ran

14 of 19 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; 2 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 11 failures of our instrument; 5 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/2306.06359.json · badge: /api/ran/2306.06359.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
gatech-eic/nerfool canonical (classifier label)
classifier basis: classifier_2025
9 ran · 4 no recorded run
GATECH-EIC/NeRFool canonical (classifier label)
classifier basis: census
5 ran · 1 no recorded run
FunctionStatusWhere it lives
SL1Loss Ran
ran on a synthesized input
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("375ac5d275e351d0")
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, 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

INIT_ARGS = {}

ARGS = {
    'depth_pred': torch.randn(2, 4, 3),
    'depth_gt': torch.rand(2, 4, 3) * 10,
    'mask': None,
    'useMask': True
}
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, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
calc_depth_smooth_loss Ran
ran, honoured the contract we drafted
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("5f62ab16341c86e9")
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)

# The function expects:
# - ret: a dict with a 'depth' key containing a tensor
# - patch_size: an integer
# - loss_type: a string (default 'l2')

# The depth tensor needs to have shape that can be reshaped to [-1, patch_size, patch_size]
# For patch_size=4, we need a tensor with total elements divisible by 4*4=16
# Let's use batch=2, so total = 2*4*4 = 32 elements, flattened to shape [32]

patch_size = 4
# Create a depth tensor: shape [32] which will reshape to [2, 4, 4]
depth_tensor = torch.randn(2 * patch_size * patch_size)

ret = {
    'depth': depth_tensor
}

ARGS = {
    'ret': ret,
    'patch_size': patch_size,
    'loss_type': 'l2'
}
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
calc_depth_var Ran
ran, honoured the contract we drafted
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("abd301e7de6eb76a")
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 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 function
batch_size = 2
num_samples = 4

# depth: shape (batch_size,) - mean depth per ray
depth = torch.rand(batch_size)

# weights: shape (batch_size, num_samples) - weights along ray
weights = torch.rand(batch_size, num_samples)
weights = weights / weights.sum(dim=1, keepdim=True)  # normalize to sum to 1

# z_vals: shape (batch_size, num_samples) - depth sample positions
z_vals = torch.linspace(0, 1, num_samples).unsqueeze(0).expand(batch_size, -1)

ret = {
    'depth': depth,
    'weights': weights,
    'z_vals': z_vals
}

ARGS = {
    'ret': ret
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
class_for_name Ran
ran; our contract draft was wrong, not the code
GATECH-EIC/NeRFool/gnt/feature_network.py
code served (permissive licence) · get_code("22e670c8093b7c80")
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 type. 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 importlib

torch_imported = False
try:
    import torch
    torch.manual_seed(0)
    torch_imported = True
except ImportError:
    pass

# The function class_for_name takes two string parameters:
# - module_name: the name of a module to import
# - class_name: the name of a class to retrieve from that module

# We'll use a real module and class that exist in Python's standard library
# For example, we can use the 'collections' module and get 'OrderedDict'

ARGS = {
    "module_name": "collections",
    "class_name": "OrderedDict"
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
conv1x1 Ran
ran; our contract draft was wrong, not the code
GATECH-EIC/NeRFool/gnt/feature_network.py
pointer only (licence: MIT); the stored copy get_code serves, from KLMAV-CUC/GoLF-NRT, is pointer only (licence: NONE) · get_code("036782aebf2185a0")
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 KLMAV-CUC/GoLF-NRT, 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.

conv3x3 Ran
ran; our contract draft was wrong, not the code
GATECH-EIC/NeRFool/gnt/feature_network.py
pointer only (licence: MIT); the stored copy get_code serves, from KLMAV-CUC/GoLF-NRT, is pointer only (licence: NONE) · get_code("7e558f4229d053af")
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 KLMAV-CUC/GoLF-NRT, 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.

dilate_img Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
GATECH-EIC/NeRFool/ibrnet/sample_ray.py
code served (permissive licence) · get_code("ffadd89803ecb27b")
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 ndarray, ndim 2, dtype family float, shape (10, 15) 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

# Create a small test image as uint8 numpy array
# The function expects a 2D or 3D uint8 array for cv2.dilate
test_img = np.random.randint(0, 256, size=(10, 15), dtype=np.uint8)

ARGS = {
    "img": test_img,
    "kernel_size": 20
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (10, 15) — shape (10, 15) 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
fused_mean_variance Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
GATECH-EIC/NeRFool/ibrnet/mlp_network.py
code served (permissive licence) · get_code("34ea5f2db980a0f7")
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, ndim 4, 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

torch.manual_seed(0)

# Construct tensors for fused_mean_variance
# The function operates on x and weight with dim=2
# x shape: (batch, features, sequence_length) = (2, 4, 3)
# weight shape: must be broadcastable, typically (2, 4, 3) or (1, 1, 3)

x = torch.randn(2, 4, 3)
weight = torch.ones(2, 4, 3)

ARGS = {
    "x": x,
    "weight": weight,
}
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, ndim 4, dtype family float — a tuple 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
img2mse Ran
ran; our contract draft was wrong, not the code
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("598359a17b0a3a7f")
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 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 two image tensors with shape [batch, height, width, 3]
# Using distinct sizes: batch=2, height=4, width=5, channels=3
x = torch.randn(2, 4, 5, 3)
y = torch.randn(2, 4, 5, 3)

# Create a mask tensor with shape [batch, height, width]
mask = torch.ones(2, 4, 5)

ARGS = {
    "x": x,
    "y": y,
    "mask": mask,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
parse_camera Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("3b75759486992d96")
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 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

torch.manual_seed(0)

# The function expects params to be a tensor with shape (batch_size, 34)
# where:
# - params[:, 0] is H (height)
# - params[:, 1] is W (width)
# - params[:, 2:18] are intrinsics (16 values, reshaped to 4x4)
# - params[:, 18:34] are c2w (16 values, reshaped to 4x4)

batch_size = 2
params = torch.randn(batch_size, 34)

ARGS = {
    "params": params
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
project_with_depth Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("a4e9aa55fe6d49cd")
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

torch.manual_seed(0)

# Parameters for project_with_depth
# The function expects:
# - depth_ref: tensor with shape (batchsize, height, width)
# - intrinsics_ref: tensor with shape (batchsize, 3, 3) for camera intrinsics
# - extrinsics_ref: tensor with shape (batchsize, 4, 4) for camera extrinsics
# - intrinsics_src: tensor with shape (batchsize, 3, 3) for source camera intrinsics
# - extrinsics_src: tensor with shape (batchsize, 4, 4) for source camera extrinsics

batchsize = 2
height = 4
width = 5

# Create depth map tensor
depth_ref = torch.ones(batchsize, height, width, dtype=torch.float32) * 2.0

# Create intrinsic matrices (3x3)
# Standard camera intrinsics with focal length and principal point
intrinsics_ref = torch.zeros(batchsize, 3, 3, dtype=torch.float32)
intrinsics_ref[:, 0, 0] = 500.0  # fx
intrinsics_ref[:, 1, 1] = 500.0  # fy
intrinsics_ref[:, 0, 2] = width / 2.0  # cx
intrinsics_ref[:, 1, 2] = height / 2.0  # cy
intrinsics_ref[:, 2, 2] = 1.0

intrinsics_src = torch.zeros(batchsize, 3, 3, dtype=torch.float32)
intrinsics_src[:, 0, 0] = 500.0
intrinsics_src[:, 1, 1] = 500.0
intrinsics_src[:, 0, 2] = width / 2.0
intrinsics_src[:, 1, 2] = height / 2.0
intrinsics_src[:, 2, 2] = 1.0

# Create extrinsic matrices (4x4) - identity transformation
extrinsics_ref = torch.eye(4, dtype=torch.float32).unsqueeze(0).repeat(batchsize, 1, 1)
extrinsics_src = torch.eye(4, dtype=torch.float32).unsqueeze(0).repeat(batchsize, 1, 1)

# Small perturbation to make src different from ref
extrinsics_src[:, 0, 3] = 0.1  # small translation in x

ARGS = {
    'depth_ref': depth_ref,
    'intrinsics_ref': intrinsics_ref,
    'extrinsics_ref': extrinsics_ref,
    'intrinsics_src': intrinsics_src,
    'extrinsics_src': extrinsics_src,
}
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
raw2outputs Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("1bbb09fcee81b67f")
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 OrderedDict. 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 collections import OrderedDict

torch.manual_seed(0)

# Set up tensor dimensions
N_rays = 2
N_samples = 4

# raw: [N_rays, N_samples, 4] - contains RGB (3 channels) + sigma (1 channel)
raw = torch.randn(N_rays, N_samples, 4)

# z_vals: [N_rays, N_samples] - depth values along rays
z_vals = torch.linspace(0.1, 10.0, N_samples).unsqueeze(0).expand(N_rays, -1)

# mask: [N_rays, N_samples] - binary mask for ray samples
mask = torch.randint(0, 2, (N_rays, N_samples))

# white_bkgd: boolean flag for white background
white_bkgd = False

# geo_noise: optional noise parameter for sigma
geo_noise = None

ARGS = {
    'raw': raw,
    'z_vals': z_vals,
    'mask': mask,
    'white_bkgd': white_bkgd,
    'geo_noise': geo_noise
}
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 OrderedDict — a OrderedDict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a OrderedDict 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
sample_along_camera_ray Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("1771162d12b6e77f")
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 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

torch.manual_seed(0)

# ray_o: origin of the ray in scene coordinate system; tensor of shape [N_rays, 3]
ray_o = torch.randn(2, 3)

# ray_d: homogeneous ray direction vectors in scene coordinate system; tensor of shape [N_rays, 3]
ray_d = torch.randn(2, 3)

# depth_range: [near_depth, far_depth]
depth_range = torch.tensor([[0.5, 5.0]])

# N_samples: number of samples along the ray
N_samples = 4

# inv_uniform: if True, uniformly sampling inverse depth
inv_uniform = False

# det: if True, will perform deterministic sampling
det = False

ARGS = {
    "ray_o": ray_o,
    "ray_d": ray_d,
    "depth_range": depth_range,
    "N_samples": N_samples,
    "inv_uniform": inv_uniform,
    "det": det,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
sample_pdf Ran
ran; our contract draft was wrong, not the code
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("ca8853bb68d69984")
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 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)

# Parameters for sample_pdf
N_rays = 2
M = 4  # number of bins
N_samples = 3

# bins: tensor of shape [N_rays, M+1]
bins = torch.linspace(0, 1, M + 1).unsqueeze(0).repeat(N_rays, 1)

# weights: tensor of shape [N_rays, M]
weights = torch.rand(N_rays, M)

# N_samples: number of samples along each ray
N_samples_val = N_samples

# det: deterministic sampling flag
det = False

ARGS = {
    "bins": bins,
    "weights": weights,
    "N_samples": N_samples_val,
    "det": det,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
RaySamplerSingleImage Not yet run
unverified (harvested, no recorded run)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("80acb005a349e795")
How it ran

no invocation recorded.

de_parallel Not yet run
unverified (harvested, no recorded run)
GATECH-EIC/NeRFool/gnt/model.py
pointer only (licence: MIT); the stored copy get_code serves, from KLMAV-CUC/GoLF-NRT, is pointer only (licence: NONE) · get_code("b7b60558eb5ef908")
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 KLMAV-CUC/GoLF-NRT, 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. No invocation recorded.

forward_warp Not yet run
unverified (harvested, no recorded run)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("ed70bc3a7a4fe74a")
How it ran

no invocation recorded.

optimize_adv_perturb Not yet run
unverified (harvested, no recorded run)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("d861a166b8b9beb1")
How it ran

no invocation recorded.

render_rays Not yet run
unverified (harvested, no recorded run)
gatech-eic/nerfool/eval/ibrnet/eval_adv.py
code served (permissive licence) · get_code("be1957ada4055c7a")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 19 distinct functions with an identity, 10 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 10 later) and 9 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
class_for_name 5 other papers
2 dated earlier, 3 later
1505.04597 (2015-05)
U-Net: Convolutional Networks for Biomedical Image Segmentation
2004.13324 (2020-04)
Learning Feature Descriptors using Camera Pose Supervision
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
and 2 more not named here
conv1x1 5 other papers
5 later
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
2308.13897 (2023-08)
InsertNeRF: Instilling Generalizability into NeRF with HyperNet Modules
2310.15712 (2023-10)
GNeSF: Generalizable Neural Semantic Fields
and 2 more not named here
conv3x3 5 other papers
5 later
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
2308.13897 (2023-08)
InsertNeRF: Instilling Generalizability into NeRF with HyperNet Modules
2310.15712 (2023-10)
GNeSF: Generalizable Neural Semantic Fields
and 2 more not named here
de_parallel 4 other papers
4 later
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
2310.15712 (2023-10)
GNeSF: Generalizable Neural Semantic Fields
2311.11845 (2023-11)
Entangled View-Epipolar Information Aggregation for Generalizable Neural Radiance Fields
and 1 more not named here
fused_mean_variance 4 other papers
2 dated earlier, 2 later · 1 only as an assembly brick
2102.13090 (2021-02)
IBRNet: Learning Multi-View Image-Based Rendering
2303.13014 (2023-03)
Semantic Ray: Learning a Generalizable Semantic Field with Cross-Reprojection Attention
2308.13897 (2023-08)
InsertNeRF: Instilling Generalizability into NeRF with HyperNet Modules
and 1 more not named here
dilate_img 2 other papers
2 later
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
2311.11845 (2023-11)
Entangled View-Epipolar Information Aggregation for Generalizable Neural Radiance Fields
img2mse
brick of this paper's assembly: helper
2 other papers
1 later, 1 same year, order not determinable
Zhan_NeRFrac_Neural_Radiance_Fields_through_Refractive_Surface_ICCV_2023_paper (2023 (year only))
NeRFrac: Neural Radiance Fields through Refractive Surface
2310.15712 (2023-10)
GNeSF: Generalizable Neural Semantic Fields
parse_camera
brick of this paper's assembly: helper
2 other papers
2 later
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
2311.11845 (2023-11)
Entangled View-Epipolar Information Aggregation for Generalizable Neural Radiance Fields
sample_along_camera_ray
brick of this paper's assembly: helper
2 other papers
2 later
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
2311.11845 (2023-11)
Entangled View-Epipolar Information Aggregation for Generalizable Neural Radiance Fields
sample_pdf
brick of this paper's assembly: helper
2 other papers
2 later
2308.11793 (2023-08)
Enhancing NeRF akin to Enhancing LLMs: Generalizable NeRF Transformer with Mixture-of-View-Experts
2311.11845 (2023-11)
Entangled View-Epipolar Information Aggregation for Generalizable Neural Radiance Fields

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

Generalizable Neural Radiance Fields (GNeRF) are one of the most promising real-world solutions for novel view synthesis, thanks to their cross-scene generalization capability and thus the possibility of instant rendering on new scenes. While adversarial robustness is essential for realworld applications, little study has been devoted to understanding its implication on GNeRF. We hypothesize that because GNeRF is implemented by conditioning on the source views from new scenes, which are often acquired from the Internet or third-party providers, there are potential new security concerns regarding its real-world applications. Meanwhile, existing understanding and solutions for neural networks' adversarial robustness may not be applicable to GNeRF, due to its 3D nature and uniquely diverse operations. To this end, we present NeRFool, which to the best of our knowledge is the first work that sets out to understand the adversarial robustness of GNeRF. Specifically, NeRFool unveils the vulnerability patterns and important insights regarding GNeRF's adversarial robustness. Built upon the above insights gained from NeRFool, we further develop NeRFool + , which integrates two techniques capable of effectively attacking GNeRF across a wide range of target views, and provide guidelines for defending against our proposed attacks. We believe that our NeRFool/NeRFool + lays the initial foundation for future innovations in developing robust realworld GNeRF solutions. Our codes are available at: https://github.com/GATECH-EIC/NeRFool.

For agents

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

get_harvested_code_for_paper("2306.06359")
get_code_for_paper("2306.06359")
have("2306.06359")

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

curl https://syntology.ai/api/ran/2306.06359.json

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

[![Syntology run record](https://syntology.ai/api/ran/2306.06359.svg)](https://syntology.ai/paper/2306.06359)

Connect an agent — have() is free.