SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.12786 · ICCV · 2023

FeatureNeRF: Learning Generalizable NeRFs by Distilling Foundation Models

Xiaolong Wang, U San, Naiyan Wang, Jianglong Ye, Tusimple

arXiv · PDF · Open in the Atlas

Code that ran

5 of 7 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 3 have no contract verdict; 2 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/2303.12786.json · badge: /api/ran/2303.12786.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
jianglongye/featurenerf canonical (classifier label)
classifier basis not recorded on the link
5 ran · 2 no recorded run
FunctionStatusWhere it lives
extra_args Ran
ran on a synthesized input
jianglongye/featurenerf/eval/eval_2d_task.py
code served (permissive licence) · get_code("2efc16adaed41431")
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 ArgumentParser. 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 argparse
import torch

# Create a minimal argument parser
parser = argparse.ArgumentParser()

# The function extra_args only takes 'parser' as an argument
# and modifies it by adding arguments, then returns it
ARGS = {
    'parser': parser
}
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 ArgumentParser — a ArgumentParser output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ArgumentParser 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
get_alpha_loss Ran
ran; our contract draft was wrong, not the code
jianglongye/featurenerf/src/model/loss.py
pointer only (licence: MIT); the stored copy get_code serves, from dawning77/nerrf, is pointer only (licence: NONE) · get_code("bf28ce0de1a71cef")
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 dawning77/nerrf, 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.

get_rgb_loss Ran
ran; our contract draft was wrong, not the code
jianglongye/featurenerf/src/model/loss.py
code served (permissive licence) · get_code("2c5e306529ece6cf")
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 MSELoss. 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 Config:
    def __init__(self, use_l1=False, use_uncertainty=False):
        self.use_l1 = use_l1
        self.use_uncertainty = use_uncertainty
    
    def get_bool(self, key, default=None):
        if key == 'use_l1':
            return self.use_l1
        elif key == 'use_uncertainty':
            return self.use_uncertainty
        return default

conf = Config(use_l1=False, use_uncertainty=False)

ARGS = {
    'conf': conf,
    'coarse': True,
    'using_bg': False,
    'reduction': 'mean'
}
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 MSELoss — a MSELoss output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MSELoss 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
seg_map_to_vis_map Ran
ran on a synthesized input
jianglongye/featurenerf/eval/eval_2d_task.py
code served (permissive licence) · get_code("2cb93c8f9b1fad49")
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 3, dtype family int, shape (3, 4, 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 numpy as np

# Create a small segmentation map with multiple labels
# The function expects a 2D array of integer labels
seg_map = np.array([
    [0, 1, 1, 2],
    [0, 1, 2, 2],
    [3, 3, 2, 1]
], dtype=np.int32)

# Optional text parameter
text = "Test Label"

ARGS = {
    "seg_map": seg_map,
    "text": text
}
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 3, dtype family int, shape (3, 4, 3) — shape (3, 4, 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
seg_map_to_vis_map Ran
ran on a synthesized input
jianglongye/featurenerf/eval/eval_3d_task.py
code served (permissive licence) · get_code("fee53d3b28454688")
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 3, dtype family int, shape (10, 12, 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 numpy as np

# Create small test data for seg_map_to_vis_map
# img: RGB image array (height x width x channels)
img = np.random.randint(0, 256, size=(10, 12, 3), dtype=np.uint8)

# uvs: array of (x, y) coordinates, shape (num_points, 2)
num_points = 20
uvs = np.array([
    [np.random.randint(0, 12), np.random.randint(0, 10)] 
    for _ in range(num_points)
], dtype=np.int32)

# seg_labels: segmentation labels for each point, shape (num_points,)
# Labels can be 0 (background, skipped) or 1-4 (foreground segments)
seg_labels = np.array([
    0, 1, 1, 2, 0, 3, 1, 0, 2, 4,
    1, 2, 3, 0, 4, 1, 2, 3, 4, 0
], dtype=np.int32)

# text: optional text to overlay (None or string)
text = "Test Label"

ARGS = {
    "img": img,
    "uvs": uvs,
    "seg_labels": seg_labels,
    "text": text
}
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 3, dtype family int, shape (10, 12, 3) — shape (10, 12, 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
make_encoder Not yet run
unverified (harvested, no recorded run)
jianglongye/featurenerf/src/model/model_util.py
pointer only (licence: MIT); the stored copy get_code serves, from dawning77/nerrf, is pointer only (licence: NONE) · get_code("09578629f7267e4b")
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 dawning77/nerrf, 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.

make_mlp Not yet run
unverified (harvested, no recorded run)
jianglongye/featurenerf/src/model/model_util.py
code served (permissive licence) · get_code("ea00c4d267522342")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 4 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 4 later) and 3 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
get_alpha_loss 2 other papers
2 later
2309.13039 (2023-09)
NeRRF: 3D Reconstruction and View Synthesis for Transparent and Specular Objects with Neural Refractive-Reflective Fields
aaai_28229 (2024 (year only))
ColNeRF: Collaboration for Generalizable Sparse Input Neural Radiance Field
get_rgb_loss 2 other papers
2 later
2309.13039 (2023-09)
NeRRF: 3D Reconstruction and View Synthesis for Transparent and Specular Objects with Neural Refractive-Reflective Fields
aaai_28229 (2024 (year only))
ColNeRF: Collaboration for Generalizable Sparse Input Neural Radiance Field
make_encoder 2 other papers
2 later
2309.13039 (2023-09)
NeRRF: 3D Reconstruction and View Synthesis for Transparent and Specular Objects with Neural Refractive-Reflective Fields
aaai_28229 (2024 (year only))
ColNeRF: Collaboration for Generalizable Sparse Input Neural Radiance Field
make_mlp 1 other paper
1 later
aaai_28229 (2024 (year only))
ColNeRF: Collaboration for Generalizable Sparse Input Neural Radiance Field

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

Figure 1: While most generalizable NeRFs focus on novel-view synthesis, we propose a framework named FeatureNeRF to learn 3D semantic representations by distilling vision foundation models. After distillation, FeatureNeRF allows to render novel-view feature maps given a single input image (a), which can be leveraged to various downstream tasks. Here, we show how we propagate part segmentation labels and keypoints to different views and instances in both 2D and 3D domains (b).

For agents

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

get_harvested_code_for_paper("2303.12786")
get_code_for_paper("2303.12786")
have("2303.12786")

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

curl https://syntology.ai/api/ran/2303.12786.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.12786.svg)](https://syntology.ai/paper/2303.12786)

Connect an agent — have() is free.