Papers › tinyBenchmarks: evaluating LLMs with fewer examples

tinyBenchmarks: evaluating LLMs with fewer examples

22 Feb 2024arXiv:2402.14992archive 2025-07-28

Felipe Maia Polo, Lucas Weber, Leshem Choshen, Yuekai Sun, Gongjun Xu, Mikhail Yurochkin

The versatility of large language models (LLMs) led to the creation of diverse benchmarks that thoroughly test a variety of language models' abilities. These benchmarks consist of tens of thousands of examples making evaluation of LLMs very expensive. In this paper, we investigate strategies to reduce the number of evaluations needed to assess the performance of an LLM on several key benchmarks. For example, we show that to accurately estimate the performance of an LLM on MMLU, a popular multiple-choice QA benchmark consisting of 14K examples, it is sufficient to evaluate this LLM on 100 curated examples. We release evaluation tools and tiny versions of popular benchmarks: Open LLM Leaderboard, MMLU, HELM, and AlpacaEval 2.0. Our empirical analysis demonstrates that these tools and tiny benchmarks are sufficient to reliably and efficiently reproduce the original evaluation results.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 39 code samples harvested from 3 repositories linked to this paper: 25 ran (of which 0 constructed an object rather than computing a result; 22 with no instrument failure: 4 honoured, 1 violated, 17 with no contract checked; 3 where Syntology's instrument failed) · 14 unverified. Where Syntology's instrument failed: 3 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 22 samples from 2 repositories, 10 ran (7 with no instrument failure; 3 where Syntology's instrument failed); community (archive-listed): 17 samples from 1 repository, 15 ran (15 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 1 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-22 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

felipemaiapolo/efficbench officialmentioned in papermentioned on GitHubSyntology: 17 samples, 7 ran (2 instrument-failed) · 10 unverified report
felipemaiapolo/tinybenchmarks officialmentioned in papermentioned on GitHubMITSyntology: 5 samples, 3 ran (1 instrument-failed) · 2 unverified report
mariushobbhahn/swebench-verified-mini mentioned on GitHubMITSyntology: 17 samples, 15 ran (0 instrument-failed) · 2 unverified report
sentient-agi/oml-1.0-fingerprinting mentioned on GitHubpytorchApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Training AI to be Loyal (arXiv:2502.15720; that paper's own run record: 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified); nothing checks that they implement this paper's method. report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

39 samples harvested: 25 ran (of which 0 constructed an object rather than computing a result; 22 with no instrument failure: 4 honoured, 1 violated, 17 with no contract checked; 3 where Syntology's instrument failed) · 14 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

4ran · honoured contract
1ran · violated contract
3ran · our draft was wrong
17ran
14unverified

Licence: 17 of the 39 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 3 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

felipemaiapolo/efficbench official repository 17 samples, 7 ran (2 instrument-failed) · 10 unverified
get_data in 223 papers run_experiment.py ran no licence file found · pointer only · f79b7ee882bf1224 · report fixture: returned tuple · level recorded 2026-09-24
get_datas generating_data/download-openllmleaderboard/download_model_names/openllm.py ran · honoured contract no licence file found · pointer only · 3f37f4d4e8b9e152 · report fixture: signature data:dict/0/? · returned list · level recorded 2026-09-24
get_lambda in 8 papers utils.py ran no licence file found · pointer only · c030bb619c641f8c · report fixture: signature b:Tensor/0/float, v:Tensor/0/float · returned Tensor · level recorded 2026-09-24
item_curve utils.py ran · honoured contract fingerprinted no licence file found · pointer only · d173066d5ff1fb67 · report fixture: signature theta:ndarray/2/float, a:ndarray/2/float, b:ndarray/2/float · returned ndarray · level recorded 2026-09-24
load_irt_parameters irt.py ran · our draft was wrong no licence file found · pointer only · a744fa187a860f72 · report fixture: signature model_name:str/0/str · returned tuple · level recorded 2026-09-24
sigmoid in 216 papers
census: 36 implementations from 56 papers, 7 distinct outputs across 3 buckets, one shared input per bucket
utils.py ran · honoured contract fingerprinted no licence file found · pointer only · 613d5fa522ce2fe7 · report fixture: signature z:ndarray/2/float · returned ndarray · level recorded 2026-09-24
winrate plots.py ran · our draft was wrong fingerprinted no licence file found · pointer only · f9a2099f222c55c0 · report fixture: signature x:ndarray/2/float, axis:int/0/int · returned ndarray · level recorded 2026-09-24
calculate_accuracies acc.py unverified no licence file found · pointer only · 665d15ac51c28fd5 · report
compute_acc_pirt acc.py unverified no licence file found · pointer only · db63652efa566108 · report
estimate_ability_parameters irt.py unverified no licence file found · pointer only · a0e44adf60096817 · report
evaluate_scenarios experiments.py unverified no licence file found · pointer only · 1153f14380c21738 · report
get_commit_dates_models generating_data/download_alpaca/preprocess_alpaca_eval.py unverified no licence file found · pointer only · 9562e2171d59db3a · report
get_last_commit_date generating_data/download_alpaca/preprocess_alpaca_eval.py unverified no licence file found · pointer only · 5c4864e2b480cdcb · report
get_random selection.py unverified no licence file found · pointer only · 68f19e5ceaec8296 · report
load_scores in 10 papers plots.py unverified no licence file found · pointer only · 9d2ac225248cfcc2 · report
run_adaptive_selection selection.py unverified no licence file found · pointer only · 6620cc858a942cb4 · report
select_initial_adaptive_items selection.py unverified no licence file found · pointer only · 5c7d27e94d6596c8 · report
felipemaiapolo/tinybenchmarks official repository 5 samples, 3 ran (1 instrument-failed) · 2 unverified
item_curve demo/utils.py ran · honoured contract fingerprinted MIT (permissive) · 921caf5fc3c2b540 · report fixture: signature theta:ndarray/2/float, a:ndarray/1/float, b:float/0/float · returned ndarray · level recorded 2026-09-24
prepare_data in 87 papers tutorials/utils.py ran · our draft was wrong MIT (permissive) · ef3515d9063526b8 · report fixture: signature scenarios:dict/0/?, data:dict/0/? · returned tuple · level recorded 2026-09-24
sigmoid in 216 papers
census: 36 implementations from 56 papers, 7 distinct outputs across 3 buckets, one shared input per bucket this sample: bucket 1 of 3, arg 1: rank 1, kind float, dtype float64
demo/utils.py ran · violated contract fingerprinted MIT (permissive) · a911f9cacbc9faf4 · report fixture: signature z:ndarray/1/float · returned ndarray · level recorded 2026-09-22
estimate_ability_parameters tutorials/irt.py unverified MIT (permissive) · 5cfb956fb08b8f0e · report
fit_theta demo/utils.py unverified MIT (permissive) · bfcec22b9dae71a3 · report
mariushobbhahn/swebench-verified-mini community (archive-listed) 17 samples, 15 ran (0 instrument-failed) · 2 unverified
check_file_exists in 5 papers run_all.py ran MIT (permissive) · 3e37b7d40582ad9d · report fixture: signature file_path:PosixPath/0/? · returned bool · level recorded 2026-09-24
compute_kmeans generate_subsets.py ran MIT (permissive) · 44ffb79d93d3714c · report fixture: signature data:DataFrame/0/?, n_clusters:int/0/int, random_state:int/0/int · returned tuple · level recorded 2026-09-24
convert_to_gb get_docker_image_sizes.py ran MIT (permissive) · 607e7cb82032195e · report fixture: signature size_str:str/0/str · returned float · level recorded 2026-09-24
count_test_results extract_data_from_logs.py ran MIT (permissive) · 8ae80ca052410f1d · report fixture: signature test_output:str/0/str · returned tuple · level recorded 2026-09-24
draw_representative_sample generate_subsets.py ran MIT (permissive) · ac93c19afba2b273 · report fixture: signature data:DataFrame/0/?, labels:ndarray/1/int, n:int/0/int, random_state:int/0/int · returned DataFrame · level recorded 2026-09-24
filter_dataset_by_ids make_new_huggingface_dataset.py ran MIT (permissive) · bac16ddf5e843b55 · report fixture: signature dataset:dict/0/?, ids:list/1/str · returned DatasetDict · level recorded 2026-09-25
get_score_and_test_counts_from_sample extract_data_from_logs.py ran MIT (permissive) · 479a232346a8cdc4 · report fixture: signature sample:MockSample/0/? · returned tuple · level recorded 2026-09-24
load_sample_ids make_new_huggingface_dataset.py ran MIT (permissive) · 6a02dcfbdb7bea7d · report fixture: signature path:PosixPath/0/? · returned list · level recorded 2026-09-24
map_environments get_docker_image_sizes.py ran MIT (permissive) · f81ca1d465c962aa · report fixture: signature df:DataFrame/0/?, mapping_path:PosixPath/0/? · returned DataFrame · level recorded 2026-09-24
merge_data_with_metadata add_metadata_to_data.py ran MIT (permissive) · d7d8fe87aefb67aa · report fixture: signature data:DataFrame/0/?, annotation:DataFrame/0/? · returned DataFrame · level recorded 2026-09-24
merge_model_scores extract_data_from_logs.py ran MIT (permissive) · a2fa41c1789608ea · report fixture: signature df_list:list/1/DataFrame · returned DataFrame · level recorded 2026-09-25
merge_with_environment_data add_metadata_to_data.py ran MIT (permissive) · ef12fdf35ff27c06 · report fixture: signature df:DataFrame/0/?, size_env_data:DataFrame/0/? · returned DataFrame · level recorded 2026-09-24
prepare_data in 87 papers generate_subsets.py ran MIT (permissive) · cc7ed1b21213f9b3 · report fixture: returned DataFrame · level recorded 2026-09-24
prepare_numeric_data compare_subsets.py ran MIT (permissive) · f20343e9cf5ded69 · report fixture: signature data:DataFrame/0/?, replace_difficulty_with_number:bool/0/bool · returned DataFrame · level recorded 2026-09-24
process_docker_output get_docker_image_sizes.py ran MIT (permissive) · fb3c33baacb5cdd5 · report fixture: signature df:DataFrame/0/? · returned DataFrame · level recorded 2026-09-25
compare_difficulty_distributions compare_subsets.py unverified MIT (permissive) · bf1d584f11529d41 · report
compare_means compare_subsets.py unverified MIT (permissive) · fd966a7ef33f7a71 · report

Tasks

MMLUMultiple-choice

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections