SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2306.01667 · 2023

Towards In-context Scene Understanding

arXiv · PDF · Open in the Atlas

Code that ran

4 of 6 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 4 have no contract verdict; 0 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/2306.01667.json · badge: /api/ran/2306.01667.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
vpariza/open-hummingbird-eval community (classifier label)
classifier basis not recorded on the link
4 ran · 2 no recorded run
FunctionStatusWhere it lives
apply_horizontal_flip Ran
ran on a synthesized input
vpariza/open-hummingbird-eval/hbird/utils/image_transformations.py
code served (permissive licence) · get_code("a92852437d910218")
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
from PIL import Image
import torchvision.transforms.functional as F
from torchvision import transforms

torch.manual_seed(0)

# Create a small PIL Image (2x4 RGB)
pil_image = Image.new('RGB', (4, 2), color=(100, 150, 200))

# Create a small tensor target (can be a PIL Image or Tensor)
# Using a simple tensor of shape (3, 2, 4) to represent a small image-like target
tensor_target = torch.randn(3, 2, 4)

ARGS = {
    'image': pil_image,
    'target': tensor_target
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, torch, torchvision — 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
get_hbird_train_transforms_for_imgs Ran
ran on a synthesized input
vpariza/open-hummingbird-eval/hbird/utils/transforms.py
code served (permissive licence) · get_code("d379a871676e8fc0")
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 torchvision.transforms as trn

torch.manual_seed(0)

ARGS = {
    'input_size': 224,
    'min_scale_factor': 0.5,
    'max_scale_factor': 2.0,
    'brightness_jitter_range': 0.1,
    'contrast_jitter_range': 0.1,
    'saturation_jitter_range': 0.1,
    'hue_jitter_range': 0.1,
    'brightness_jitter_probability': 0.5,
    'contrast_jitter_probability': 0.5,
    'saturation_jitter_probability': 0.5,
    'hue_jitter_probability': 0.5,
    'img_mean': [0.485, 0.456, 0.406],
    'img_std': [0.229, 0.224, 0.255],
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — 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
list_files Ran
ran on a synthesized input
vpariza/open-hummingbird-eval/hbird/utils/io.py
code served (permissive licence) · get_code("974405d0f091a147")
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 1, 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 os
import tarfile
import tempfile
from pathlib import Path

# Create a temporary directory with some test files
temp_dir = tempfile.mkdtemp()
os.makedirs(os.path.join(temp_dir, 'subdir'), exist_ok=True)
Path(os.path.join(temp_dir, 'file1.txt')).touch()
Path(os.path.join(temp_dir, 'file2.txt')).touch()
Path(os.path.join(temp_dir, 'subdir', 'file3.txt')).touch()

# Create a tar archive with test files
tar_path = os.path.join(temp_dir, 'test_archive.tar')
with tarfile.open(tar_path, 'w') as tar:
    tar.add(os.path.join(temp_dir, 'file1.txt'), arcname='file1.txt')
    tar.add(os.path.join(temp_dir, 'file2.txt'), arcname='file2.txt')
    tar.add(os.path.join(temp_dir, 'subdir', 'file3.txt'), arcname='subdir/file3.txt')

# Test with regular directory path
ARGS = {
    'path': temp_dir
}
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 list, ndim 1, 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
read_file_set Ran
ran on a synthesized input
vpariza/open-hummingbird-eval/hbird/utils/io.py
code served (permissive licence) · get_code("eb2df6f81672e7cf")
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 1, 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 tarfile
import tempfile
from pathlib import Path

# Create a temporary directory for test files
tmpdir = tempfile.mkdtemp()

# Test case 1: Regular text file
regular_file_path = Path(tmpdir) / "test.txt"
regular_file_path.write_text("line1\nline2  \n  line3\n\n")

# Test case 2: Text file inside a tar archive
tar_path = Path(tmpdir) / "test.tar"
inner_file_path = "subdir/data.txt"

with tarfile.open(tar_path, "w") as tar:
    # Create a temporary file to add to the tar
    inner_file = Path(tmpdir) / "temp_data.txt"
    inner_file.write_text("data_line1\ndata_line2  \n  data_line3\n")
    tar.add(inner_file, arcname=inner_file_path)
    inner_file.unlink()

# Construct arguments for the function
ARGS = {
    "path": str(regular_file_path)
}
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 list, ndim 1, 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
random_resize_crop Not yet run
unverified (harvested, no recorded run)
vpariza/open-hummingbird-eval/hbird/utils/image_transformations.py
code served (permissive licence) · get_code("d9d0b09899aea20e")
How it ran

no invocation recorded.

resize Not yet run
unverified (harvested, no recorded run)
vpariza/open-hummingbird-eval/hbird/utils/image_transformations.py
code served (permissive licence) · get_code("f64df2cf48096a0c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 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 6 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

In-context learning$\unicode{x2013}$the ability to configure a model's behavior with different prompts$\unicode{x2013}$has revolutionized the field of natural language processing, alleviating the need for task-specific models and paving the way for generalist models capable of assisting with any query. Computer vision, in contrast, has largely stayed in the former regime: specialized decoders and finetuning protocols are generally required to perform dense tasks such as semantic segmentation and depth estimation. In this work we explore a simple mechanism for in-context learning of such scene understanding tasks: nearest neighbor retrieval from a prompt of annotated features. We propose a new pretraining protocol$\unicode{x2013}$leveraging attention within and across images$\unicode{x2013}$which yields representations particularly useful in this regime. The resulting Hummingbird model, suitably prompted, performs various scene understanding tasks without modification while approaching the performance of specialists that have been finetuned for each task. Moreover, Hummingbird can be configured to perform new tasks much more efficiently than finetuned models, raising the possibility of scene understanding in the interactive assistant regime.

For agents

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

get_harvested_code_for_paper("2306.01667")
get_code_for_paper("2306.01667")
have("2306.01667")

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

curl https://syntology.ai/api/ran/2306.01667.json

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

[![Syntology run record](https://syntology.ai/api/ran/2306.01667.svg)](https://syntology.ai/paper/2306.01667)

Connect an agent — have() is free.