Papers › From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline

From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline

17 Jun 2024arXiv:2406.11939archive 2025-07-28

Tianle Li, Wei-Lin Chiang, Evan Frick, Lisa Dunlap, Tianhao Wu, Banghua Zhu, Joseph E. Gonzalez, Ion Stoica

The rapid evolution of Large Language Models (LLMs) has outpaced the development of model evaluation, highlighting the need for continuous curation of new, challenging benchmarks. However, manual curation of high-quality, human-aligned benchmarks is expensive and time-consuming. To address this, we introduce BenchBuilder, an automated pipeline that leverages LLMs to curate high-quality, open-ended prompts from large, crowd-sourced datasets, enabling continuous benchmark updates without human in the loop. We apply BenchBuilder to datasets such as Chatbot Arena and WildChat-1M, extracting challenging prompts and utilizing LLM-as-a-Judge for automatic model evaluation. To validate benchmark quality, we propose new metrics to measure a benchmark's alignment with human preferences and ability to separate models. We release Arena-Hard-Auto, a benchmark consisting 500 challenging prompts curated by BenchBuilder. Arena-Hard-Auto provides 3x higher separation of model performances compared to MT-Bench and achieves 98.6% correlation with human preference rankings, all at a cost of $20. Our work sets a new framework for the scalable curation of automated benchmarks from extensive data.

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 21 code samples harvested from 2 repositories linked to this paper: 19 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 0 violated, 3 with no contract checked; 15 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 15 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 4 samples from 1 repository, 4 ran (0 with no instrument failure; 4 where Syntology's instrument failed); community (archive-listed): 17 samples from 1 repository, 15 ran (4 with no instrument failure; 11 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 2 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

lm-sys/arena-hard-auto officialmentioned in papermentioned on GitHubpytorchApache-2.0Syntology: 4 samples, 4 ran (4 instrument-failed) · 0 unverified report
lmarena/arena-hard-auto officialmentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Representation-Aware Advantage Estimation: Your Reward Model Provides More Than A Scalar Output (arXiv:2606.10528; that paper's own run record: 31 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 3 honoured, 4 violated, 2 with no contract checked; 22 where Syntology's instrument failed) · 9 unverified); nothing checks that they implement this paper's method. report
lm-sys/arena-hard mentioned on GitHubpytorchApache-2.0Syntology: 17 samples, 15 ran (11 instrument-failed) · 2 unverified report
lmarena/ppe mentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to How to Evaluate Reward Models for RLHF (arXiv:2410.14872; that paper's own run record: 7 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 1 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 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

21 samples harvested: 19 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 0 violated, 3 with no contract checked; 15 where Syntology's instrument failed) · 2 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.

1ran · honoured contract
15ran · our draft was wrong
3ran
2unverified

Licence: 1 of the 21 samples is pointer only, meaning Syntology does not serve that copy's text. For that sample, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 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.

lm-sys/arena-hard-auto official repository 4 samples, 4 ran (4 instrument-failed) · 0 unverified
get_endpoint BenchBuilder/label.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · c8136157d71c093b · report fixture: signature endpoint_list:list/1/dict · returned dict · level recorded 2026-09-24
load_json in 344 papers BenchBuilder/filter.py ran · our draft was wrong Apache-2.0 (permissive) · 6805dfa84dd06130 · report fixture: signature file_path:str/0/str · returned list · level recorded 2026-09-24
load_questions in 17 papers utils/completion.py ran · our draft was wrong Apache-2.0 (permissive) · 47892c5631a199d4 · report fixture: signature question_file:str/0/str · returned list · level recorded 2026-09-24
make_config in 11 papers BenchBuilder/label.py ran · our draft was wrong Apache-2.0 (permissive) here; the stored copy is pointer only · 07e6f41cf857a76f · report fixture: signature config_file:str/0/str · returned dict · level recorded 2026-09-22
lm-sys/arena-hard community (archive-listed) 17 samples, 15 ran (11 instrument-failed) · 2 unverified
calculate_score in 15 papers BenchBuilder/filter.py ran · honoured contract Apache-2.0 (permissive) · 06750b8d972efe27 · report fixture: signature conversation:dict/0/? · returned int · level recorded 2026-09-24
count_markdown_elements utils/add_markdown_info.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · df075cc813246dcd · report fixture: signature markdown_text:str/0/str, suffix:str/0/str · returned dict · level recorded 2026-09-24
create_llama3_body utils/bedrock_utils.py ran Apache-2.0 (permissive) · da15e93460148f70 · report fixture: signature messages:list/1/dict, max_gen_len:int/0/int, temperature:float/0/float, top_p:float/0/float, top_k:int/0/int · returned str · level recorded 2026-09-24
create_nova_messages utils/bedrock_utils.py ran · our draft was wrong Apache-2.0 (permissive) · 6743a5e05fc4ad6d · report fixture: signature messages:list/1/dict · returned list · level recorded 2026-09-24
extract_innermost_text utils/bedrock_utils.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 759c3030a0ca985f · report fixture: signature content:list/1/dict · returned str · level recorded 2026-09-24
format_confidence_interval show_result.py ran Apache-2.0 (permissive) · bf60419bbfec3311 · report fixture: signature mean_scores:DataFrame/0/?, lower_scores:DataFrame/0/?, upper_scores:DataFrame/0/?, baseline:str/0/str · returned DataFrame · level recorded 2026-09-24
get_categories in 9 papers qa_browser.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 31d3137625992d50 · report fixture: signature benchmark:str/0/str · returned list · level recorded 2026-09-24
get_element_counts utils/add_markdown_info.py ran · our draft was wrong Apache-2.0 (permissive) · 0a127509b23580f5 · report fixture: signature df:DataFrame/0/?, column:str/0/str · returned list · level recorded 2026-09-24
get_languages qa_browser.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 7ca039993ebdcc43 · report fixture: signature benchmark:str/0/str · returned list · level recorded 2026-09-24
get_model_style_metadata show_result.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · b1fbdabfc009df1c · report fixture: signature benchmark:str/0/str · returned dict · level recorded 2026-09-24
load_jsonl in 179 papers qa_browser.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · d049c92b5f5adfa3 · report fixture: signature file_path:str/0/str · returned DataFrame · level recorded 2026-09-24
load_jsonl in 179 papers BenchBuilder/filter.py ran Apache-2.0 (permissive) · e094bc160bb7d823 · report fixture: signature file_path:str/0/str · returned list · level recorded 2026-09-24
register_api in 5 papers utils/completion.py ran · our draft was wrong Apache-2.0 (permissive) · 7a5ee66f6fe09b60 · report fixture: signature api_type:str/0/str · returned function · level recorded 2026-09-24
register_engine utils/completion.py ran · our draft was wrong Apache-2.0 (permissive) · 0399a259308f77a0 · report fixture: signature engine_type:str/0/str · returned function · level recorded 2026-09-24
remove_pattern utils/add_markdown_info.py ran · our draft was wrong Apache-2.0 (permissive) · bae70f745e7b2d09 · report fixture: signature answer:str/0/str, pattern:Pattern/0/? · returned str · level recorded 2026-09-24
chat_completion_openai in 7 papers BenchBuilder/label.py unverified Apache-2.0 (permissive) · 0322093a9ad847a4 · report
load_judgments show_result.py unverified Apache-2.0 (permissive) · ff96063cc3cfa7d6 · report

Tasks

Chatbot

Datasets

Introduced by this paper, per the archive.

Arena-Hard-Auto

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

ALIGN

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