SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2409.17424 · NeurIPS · 2025

Results of the Big ANN: NeurIPS'23 competition

Harsha Simhadri, Matthijs Douze, Dmitry Baranchuk, Zihao Wang, Kai Zhang, Jie Yin, Laxman Dhulipala, Yizhuo Chen, Rui Ma, Martin Aumüller, Amir Ingber, George Williams, and 11 more

arXiv · PDF · Open in the Atlas

Code that ran

7 of 10 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; 3 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/2409.17424.json · badge: /api/ran/2409.17424.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
harsha-simhadri/big-ann-benchmarks canonical (classifier label)
classifier basis: census
7 ran · 3 no recorded run
FunctionStatusWhere it lives
algorithm_status Ran
ran; our contract draft was wrong, not the code
harsha-simhadri/big-ann-benchmarks/benchmark/algorithms/definitions.py
code served (permissive licence) · get_code("b78ed1fd13d41cb3")
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 InstantiationStatus. 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
from enum import Enum

class InstantiationStatus(Enum):
    AVAILABLE = 0
    NO_CONSTRUCTOR = 1
    NO_MODULE = 2

# Create a stand-in definition class with module and constructor attributes
class Definition:
    def __init__(self, module, constructor):
        self.module = module
        self.constructor = constructor

# Test case 1: Valid module and constructor
definition = Definition(module="os.path", constructor="join")

ARGS = {
    "definition": definition
}
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 InstantiationStatus — a InstantiationStatus output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a InstantiationStatus 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
buddy_up Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
harsha-simhadri/big-ann-benchmarks/benchmark/algorithms/buddy_t1.py
code served (permissive licence) · get_code("59da4457b6940dfd")
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 (2, 128) 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 numpy as np

torch.manual_seed(0)

# Create a small numpy array with shape (2, 128) to match the expected dimensions
# The function indexes into this array using column indices from the buddy lists
points = np.random.randn(2, 128).astype(np.float32)

# Test with 'bigann' dataset which has a non-empty buddy list
dataset_prefix = 'bigann'

ARGS = {
    'points': points,
    'dataset_prefix': dataset_prefix,
    'copy': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — 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 (2, 128) — shape (2, 128) 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
build Ran
ran; our contract draft was wrong, not the code
harsha-simhadri/big-ann-benchmarks/install.py
code served (permissive licence) · get_code("4a00404229a68dcc")
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 subprocess

def build(tag, args, dockerfile):
    print('Building %s...' % tag)
    if args is not None and len(args) != 0:
        q = ' '.join(['--build-arg ' + x.replace(' ', '\\ ') for x in args])
    else:
        q = ''
    try:
        command = 'docker build %s --rm -t %s -f' % (q, tag)
        command += ' %s .' % dockerfile
        print(command)
        subprocess.check_call(command, shell=True)
        return {tag: 'success'}
    except subprocess.CalledProcessError:
        return {tag: 'fail'}


ARGS = {
    'tag': 'my-docker-image:latest',
    'args': ['VERSION=1.0', 'ENV=production'],
    'dockerfile': 'Dockerfile'
}
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 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
build_multiprocess Ran
ran; our contract draft was wrong, not the code
harsha-simhadri/big-ann-benchmarks/install.py
code served (permissive licence) · get_code("46577c49b50c09a9")
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 subprocess

def build(tag, args, dockerfile):
    print('Building %s...' % tag)
    if args is not None and len(args) != 0:
        q = ' '.join(['--build-arg ' + x.replace(' ', '\\ ') for x in args])
    else:
        q = ''
    try:
        command = 'docker build %s --rm -t %s -f' % (q, tag)
        command += ' %s .' % dockerfile
        print(command)
        subprocess.check_call(command, shell=True)
        return {tag: 'success'}
    except subprocess.CalledProcessError:
        return {tag: 'fail'}

def build_multiprocess(args):
    return build(*args)

ARGS = {
    'args': (
        'my-image:latest',
        ['KEY1=value1', 'KEY2=value2'],
        'Dockerfile'
    )
}
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 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
get_unique_algorithms Ran
ran; our contract draft was wrong, not the code
harsha-simhadri/big-ann-benchmarks/benchmark/algorithms/definitions.py
code served (permissive licence) · get_code("15a8353f5fe218be")
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 tempfile
import os
import yaml

# Create a temporary YAML file with test data
temp_dir = tempfile.mkdtemp()
definition_file = os.path.join(temp_dir, "definitions.yaml")

# Create test definition data matching the nested structure expected by get_unique_algorithms
test_definitions = {
    "point1": {
        "metric1": {
            "algorithm_a": {},
            "algorithm_b": {}
        },
        "metric2": {
            "algorithm_c": {}
        }
    },
    "point2": {
        "metric3": {
            "algorithm_a": {},
            "algorithm_d": {}
        }
    }
}

# Write the YAML file
with open(definition_file, 'w') as f:
    yaml.dump(test_definitions, f)

ARGS = {
    "definition_file": definition_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
yaml — 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
instantiate_algorithm Ran
ran; our contract draft was wrong, not the code
harsha-simhadri/big-ann-benchmarks/benchmark/algorithms/definitions.py
code served (permissive licence) · get_code("a8a893f658ddb978")
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 torch
import importlib
from unittest.mock import Mock

torch.manual_seed(0)

# Create a minimal definition object that has the required attributes:
# - module: string name of a module
# - constructor: string name of a callable in that module
# - arguments: list/tuple of positional arguments to pass to the constructor

definition = Mock()
definition.module = 'os.path'
definition.constructor = 'join'
definition.arguments = ['/tmp', 'file.txt']

ARGS = {
    'definition': definition
}
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 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
knn_search_batched Ran
ran; our contract draft was wrong, not the code
harsha-simhadri/big-ann-benchmarks/benchmark/algorithms/buddy_t1.py
code served (permissive licence) · get_code("7902eacd100022b7")
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 3, 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 numpy as np

class MockFaissIndex:
    """Mock FAISS index that implements the search method"""
    def __init__(self, dim, num_vectors):
        self.dim = dim
        self.num_vectors = num_vectors
        # Create dummy data for the index
        self.data = np.random.randn(num_vectors, dim).astype(np.float32)
    
    def search(self, xq, k):
        """
        Mock search that returns k nearest neighbors
        Returns (distances, indices) both as numpy arrays
        """
        # xq shape: (batch_size, dim)
        # Return distances and indices of shape (batch_size, k)
        batch_size = xq.shape[0]
        # Simple mock: return dummy distances and indices
        distances = np.random.randn(batch_size, k).astype(np.float32)
        indices = np.random.randint(0, self.num_vectors, size=(batch_size, k))
        return distances, indices

# Create mock index
index = MockFaissIndex(dim=4, num_vectors=10)

# Create query data
xq = np.random.randn(6, 4).astype(np.float32)

# Set parameters
k = 3
bs = 2

ARGS = {
    "index": index,
    "xq": xq,
    "k": k,
    "bs": bs
}
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 tuple, ndim 3, 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
buddy_random Not yet run
unverified (harvested, no recorded run)
harsha-simhadri/big-ann-benchmarks/benchmark/algorithms/buddy_t1.py
code served (permissive licence) · get_code("094a6e57546103ae")
How it ran

no invocation recorded.

cleaned_run_metric Not yet run
unverified (harvested, no recorded run)
harsha-simhadri/big-ann-benchmarks/data_export.py
code served (permissive licence) · get_code("8441efbcc93c8387")
How it ran

no invocation recorded.

memmap_bin_file Not yet run
unverified (harvested, no recorded run)
harsha-simhadri/big-ann-benchmarks/benchmark/algorithms/cuanns_ivfpq.py
code served (permissive licence) · get_code("b023210a176f40ab")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 1 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.

FunctionAlso present inEarliest-datedOthers named
build_multiprocess 2 other papers
1 dated earlier, 1 later
2308.07061 (2023-08)
Machine Unlearning: Solutions and Challenges
2410.18926 (2024-10)
LoRANN: Low-Rank Matrix Factorization for Approximate Nearest Neighbor Search
algorithm_status 1 other paper
1 dated earlier
1807.05614 (2018-07)
ANN-Benchmarks: A Benchmarking Tool for Approximate Nearest Neighbor Algorithms
get_unique_algorithms 1 other paper
1 dated earlier
1807.05614 (2018-07)
ANN-Benchmarks: A Benchmarking Tool for Approximate Nearest Neighbor Algorithms
instantiate_algorithm 1 other paper
1 dated earlier
1807.05614 (2018-07)
ANN-Benchmarks: A Benchmarking Tool for Approximate Nearest Neighbor Algorithms

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

The 2023 Big ANN Challenge, held at NeurIPS 2023, focused on advancing the state-of-the-art in indexing data structures and search algorithms for practical variants of Approximate Nearest Neighbor (ANN) search that reflect the growing complexity and diversity of workloads. Unlike prior challenges that emphasized scaling up classical ANN search [21], this competition addressed filtered search, out-of-distribution data, sparse and streaming variants of ANNS. Participants developed and submitted innovative solutions that were evaluated on new standard datasets with constrained computational resources. The results showcased significant improvements in search accuracy and efficiency over industry-standard baselines, with notable contributions from both academic and industrial teams. This paper summarizes the competition tracks, datasets, evaluation metrics, and the innovative approaches of the top-performing submissions, providing insights into the current advancements and future directions in the field of approximate nearest neighbor search.

For agents

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

get_harvested_code_for_paper("2409.17424")
get_code_for_paper("2409.17424")
have("2409.17424")

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

curl https://syntology.ai/api/ran/2409.17424.json

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

[![Syntology run record](https://syntology.ai/api/ran/2409.17424.svg)](https://syntology.ai/paper/2409.17424)

Connect an agent — have() is free.