Datasets:

facebook
/

airs-bench

task stringlengths 22 52	category stringclasses 7 values	research_problem stringlengths 14 29	dataset stringlengths 12 40	metric stringclasses 9 values	metadata.yaml stringlengths 954 1.15k	project_description.md stringlengths 2.33k 26.1k	prepare.py stringlengths 3.09k 4.78k	evaluate_prepare.py stringlengths 2.18k 3.84k	evaluate.py stringlengths 1.67k 4.81k	custom_labels.py stringlengths 1.45k 5.36k ⌀	utils.py stringclasses 5 values
CodeGenerationAPPSPassAt5	Code	Code Generation	codeparrot/apps	Pass@5	metric_lower_is_better: false file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 - torchmetrics - pandas - numpy - torch - pyext prepare_code_python_requirements: - pyext logging_info: name: CodeGenerationAPP...	# Overview ## Task Description This is a Machine Learning project and your goal is to build a model that solves the project's TASK following the instructions provided below. TASK: Your task is to generate five independent Python programs for each competitive-programming problem. Each program must read from standar...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_from_disk # Configure logger with custom ...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. from json import load import os import sys import argparse import logging import shutil from dat...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json import pandas as pd import json from pathlib import Path from datasets import load_from_disk from u...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import json import pandas as pd from random import random from datasets import load_from_disk f...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import json import multiprocessing import numpy as np from tqdm import tqdm from testing_util import run_test from concur...
CodeRetrievalCodeXGlueMRR	Code	Code Retrieval	google/code_x_glue_tc_nl_code_search_adv	MRR	metric_lower_is_better: false file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 logging_info: name: CodeRetrievalCodeXGlueMRR dataset: google/code_x_glue_tc_nl_code_search_adv category: Code research_problem: Code...	# Overview ## Task Description This is a Machine Learning project and your goal is to complete the project's TASK following the instructions provided below. TASK: This is a NLP task to perform Code retrieval on google/code_x_glue_tc_nl_code_search_adv. ## Data ### Dataset Structure This is a retrieval task, so f...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging import re from datasets import load_dataset, load_from_disk # Confi...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import os import sys import argparse import logging import re import shutil from datasets import l...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json import numpy as np import pandas as pd import torch from datasets import loa...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. from datasets import load_from_disk import json import pandas as pd import os import argparse import re import copy impor...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. # Copyright (c) Microsoft Corporation. # Licensed under the MIT license. import logging import sys,json import numpy as ...
CoreferenceResolutionSuperGLUEWSCAccuracy	Text Extraction and Matching	Coreference Resolution	aps/super_glue	Accuracy	metric_lower_is_better: false file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 - torchmetrics - pandas - numpy - torch logging_info: name: CoreferenceResolutionSuperGLUEWSCAccuracy category: Text Extraction a...	# Overview ## Task Description This is a Machine Learning project and your goal is to build a model that solves the project's TASK following the instructions provided below. TASK: Your task is to resolve pronoun references in natural language sentences. You will be given a sentence containing an ambiguous pronoun and ...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_from_disk # Configure logger with custom ...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. from json import load import os import sys import argparse import logging import shutil from dat...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json import numpy as np import pandas as pd from datasets import load_from_disk ...	null	null
CoreferenceResolutionWinograndeAccuracy	Text Extraction and Matching	Coreference Resolution	allenai/winogrande	Accuracy	metric_lower_is_better: false file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 - torchmetrics - pandas - numpy - torch logging_info: name: CoreferenceResolutionWinograndeAccuracy category: Text Extraction and...	# Overview ## Task Description This is a Machine Learning project and your goal is to build a model that solves the project's TASK following the instructions provided below. TASK: Your task is to resolve ambiguous references in natural language sentences. You will be given a sentence containing a gap left for a possib...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_from_disk # Configure logger with custom ...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. from json import load import os import sys import argparse import logging import shutil from dat...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json import numpy as np import pandas as pd from datasets import load_from_disk ...	null	null
CvMolecularPropertyPredictionQm9MeanAbsoluteError	Molecules and Proteins ML	Molecular Property Prediction	nimashoghi/qm9	MeanAbsoluteError	metric_lower_is_better: true file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 - torchmetrics - pandas - numpy - torch logging_info: name: CvMolecularPropertyPredictionQm9MeanAbsoluteError dataset: nimashoghi/...	# Overview ## Task Description This is a Machine Learning project and your goal is to complete the project's TASK following the instructions provided below. TASK: Your task is to predict a molecular property of small molecules which is known as the heat capacity at constant volume (c_v). This is a fundamental...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_dataset, load_from_disk # Configure logge...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import os import sys import argparse import logging import shutil from datasets import load_from_...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json, numpy as np, pandas as pd # Conditional torch import - might be added by ge...	null	null
GMolecularPropertyPredictionQm9MeanAbsoluteError	Molecules and Proteins ML	Molecular Property Prediction	nimashoghi/qm9	MeanAbsoluteError	metric_lower_is_better: true file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 - torchmetrics - pandas - numpy - torch logging_info: name: GMolecularPropertyPredictionQm9MeanAbsoluteError dataset: nimashoghi/q...	# Overview ## Task Description This is a Machine Learning project and your goal is to complete the project's TASK following the instructions provided below. TASK: Your task is to predict a molecular property of small molecules which is known as the Gibbs free energy (G). This is a fundamental thermodynamic propert...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_dataset, load_from_disk # Configure logge...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import os import sys import argparse import logging import shutil from datasets import load_from_...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json, numpy as np, pandas as pd # Conditional torch import - might be added by ge...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import os import argparse import pandas as pd from datasets import load_from_disk def main( global_shared_data_dir,...	null
GraphRegressionZincMae	Molecules and Proteins ML	Graph Regression	graphs-datasets/ZINC	MAE	metric_lower_is_better: true file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 - torchmetrics - pandas - numpy - torch logging_info: name: GraphRegressionZincMae dataset: graphs-datasets/ZINC category: Molec...	# Overview ## Task Description This is a Machine Learning project and your goal is to complete the project's TASK following the instructions provided below. TASK: This is a Molecules, proteins etc task to perform Graph Regression on graphs-datasets/ZINC. Your predictions will be scored against the y column of the...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_dataset, load_from_disk # Configure logge...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import os import sys import argparse import logging import shutil from datasets import load_from_...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json import ast import numpy as np import pandas as pd import torch from datasets...	null	null
MathQuestionAnsweringSVAMPAccuracy	Math	Math Question Answering	ChilleD/SVAMP	Accuracy	metric_lower_is_better: false file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 logging_info: name: MathQuestionAnsweringSVAMPAccuracy category: Math research_problem: Math Question Answering output_type: text-gen...	# Overview ## Task Description This is a Machine Learning project and your goal is to build a model that solves the project's TASK following the instructions provided below. TASK: Your task is solve math world prolems. Each example presents a short story followed by a specific question. Your task is to read the text a...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_from_disk # Configure logger with custom ...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. from json import load import os import sys import argparse import logging import shutil from dat...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json import numpy as np import pandas as pd from datasets import load_from_disk ...	null	null
QuestionAnsweringDuoRCAccuracy	Question Answering	Question Answering	ibm-research/duorc	Accuracy	metric_lower_is_better: false file_export_globs: - submission.csv container_python_requirements: - datasets==4.0.0 evaluate_container_python_requirements: - datasets==4.0.0 - torchmetrics - pandas - numpy - torch logging_info: name: QuestionAnsweringDuoRCAccuracy category: Question Answering researc...	# Overview ## Task Description This is a Machine Learning project and your goal is to build a model that solves the project's TASK following the instructions provided below. TASK: Your task is to answer questions given a large context. You will be provided a title of a story and context surrounding its plot, then will...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import sys import logging from datasets import load_from_disk # Configure logger with custom ...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. from json import load import os import sys import argparse import logging import shutil from dat...	#!/usr/bin/env python3 # Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse, json import numpy as np import pandas as pd from datasets import load_from_disk ...	# Copyright (c) Meta Platforms, Inc. and affiliates. # All rights reserved. # # This source code is licensed under the license found in the # LICENSE file in the root directory of this source tree. import argparse import os import numpy as np from pathlib import Path from datasets import load_from_disk def parse_ar...	null
QuestionAnsweringEli5Rouge1	Question Answering	Question Answering	Pavithree/eli5	Rouge1	"metric_lower_is_better: false\nfile_export_globs:\n - submission.csv\ncontainer_python_requirement(...TRUNCATED)	"\n# Overview\n## Task Description\nThis is a Machine Learning project and your goal is to complete (...TRUNCATED)	"# Copyright (c) Meta Platforms, Inc. and affiliates.\n# All rights reserved.\n#\n# This source code(...TRUNCATED)	"#!/usr/bin/env python3\n# Copyright (c) Meta Platforms, Inc. and affiliates.\n# All rights reserved(...TRUNCATED)	"#!/usr/bin/env python3\n# Copyright (c) Meta Platforms, Inc. and affiliates.\n# All rights reserved(...TRUNCATED)	"# Copyright (c) Meta Platforms, Inc. and affiliates.\n# All rights reserved.\n#\n# This source code(...TRUNCATED)	null

End of preview. Expand in Data Studio

AIRS-Bench: a Suite of Tasks for Frontier AI Research Science Agents

The AI Research Science Benchmark (AIRS-Bench) quantifies the autonomous research abilities of LLM agents in the area of machine learning. AIRS-Bench comprises 20 tasks from state-of-the-art machine learning papers spanning diverse domains: NLP, Code, Math, biochemical modelling, and time series forecasting.

Each task is specified by a ⟨problem, dataset, metric⟩ triplet and a SOTA value. The agent receives the full task specification and is expected to develop a solution that generates predictions on a test set, which are then evaluated and compared against the state-of-the-art (SOTA) score from a published paper.

For full details see the paper and the GitHub repository.

Dataset Description

This dataset contains the task specification files for the 20 AIRS-Bench tasks, formatted for use with the aira-dojo agentic harness.

Category	# Tasks
Text Classification	2
Question Answering	4
Text Extraction and Matching	3
Molecules and Proteins ML	5
Time Series	3
Code	2
Math	1

Data Fields

Column	Type	Description
`task`	`string`	Task identifier (directory name, e.g. `SentimentAnalysisYelpReviewFullAccuracy`)
`category`	`string`	High-level domain category (e.g. `Text Classification`, `Code`)
`research_problem`	`string`	The specific research problem the task addresses
`dataset`	`string`	HuggingFace dataset identifier used for the task
`metric`	`string`	Evaluation metric (e.g. `Accuracy`, `MeanAbsoluteError`, `Rouge1`)
`metadata.yaml`	`string`	Full content of the task metadata file (dataset config, SOTA info, requirements)
`project_description.md`	`string`	The task prompt provided to the agent
`prepare.py`	`string`	Dataset preparation script (creates train/test splits, hides test labels)
`evaluate_prepare.py`	`string`	Evaluation data preparation script (creates test labels for scoring)
`evaluate.py`	`string`	Evaluation script used to score the agent's submission
`custom_labels.py`	`string`	Optional custom label handler for non-standard label formats (empty if unused)
`utils.py`	`string`	Optional shared utilities across task scripts (empty if unused)

Citation

@article{lupidi2026airsbenchsuitetasksfrontier,
      title={AIRS-Bench: a Suite of Tasks for Frontier AI Research Science Agents},
      author={Alisia Lupidi and Bhavul Gauri and Thomas Simon Foster and Bassel Al Omari and Despoina Magka and Alberto Pepe and Alexis Audran-Reiss and Muna Aghamelu and Nicolas Baldwin and Lucia Cipolina-Kun and Jean-Christophe Gagnon-Audet and Chee Hau Leow and Sandra Lefdal and Hossam Mossalam and Abhinav Moudgil and Saba Nazir and Emanuel Tewolde and Isabel Urrego and Jordi Armengol Estape and Amar Budhiraja and Gaurav Chaurasia and Abhishek Charnalia and Derek Dunfield and Karen Hambardzumyan and Daniel Izcovich and Martin Josifoski and Ishita Mediratta and Kelvin Niu and Parth Pathak and Michael Shvartsman and Edan Toledo and Anton Protopopov and Roberta Raileanu and Alexander Miller and Tatiana Shavrina and Jakob Foerster and Yoram Bachrach},
      year={2026},
      eprint={2602.06855},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2602.06855},
}