Papers › Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena

Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena

9 Jun 2023NeurIPS 2023 11arXiv:2306.05685archive 2025-07-28

Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric P. Xing, Hao Zhang, Joseph E. Gonzalez, Ion Stoica

Evaluating large language model (LLM) based chat assistants is challenging due to their broad capabilities and the inadequacy of existing benchmarks in measuring human preferences. To address this, we explore using strong LLMs as judges to evaluate these models on more open-ended questions. We examine the usage and limitations of LLM-as-a-judge, including position, verbosity, and self-enhancement biases, as well as limited reasoning ability, and propose solutions to mitigate some of them. We then verify the agreement between LLM judges and human preferences by introducing two benchmarks: MT-bench, a multi-turn question set; and Chatbot Arena, a crowdsourced battle platform. Our results reveal that strong LLM judges like GPT-4 can match both controlled and crowdsourced human preferences well, achieving over 80% agreement, the same level of agreement between humans. Hence, LLM-as-a-judge is a scalable and explainable way to approximate human preferences, which are otherwise very expensive to obtain. Additionally, we show our benchmark and traditional benchmarks complement each other by evaluating several variants of LLaMA and Vicuna. The MT-bench questions, 3K expert votes, and 30K conversations with human preferences are publicly available at https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge.

PaperPDFConference PDFCodeCode 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

No sample from the archive's official repository (lm-sys/fastchat) is linked to this paper; the 12 samples below come from 3 community repositories. Syntology harvested code from lm-sys/fastchat for 2 other papers.

lm-sys/fastchat: Syntology's graph links these samples to Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference (arXiv:2403.04132; that paper's own run record: 5 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 4 where Syntology's instrument failed) · 4 unverified) and DSPA: Dynamic SAE Steering for Data-Efficient Preference Alignment (arXiv:2603.21461; that paper's own run record: 5 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 4 where Syntology's instrument failed) · 2 unverified); nothing checks that they implement this paper's method.

Syntology 12 code samples harvested from 3 repositories linked to this paper: 9 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: community (archive-listed): 12 samples from 3 repositories, 9 ran (9 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 4 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-24 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

lm-sys/fastchat officialmentioned in paperpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference (arXiv:2403.04132; that paper's own run record: 5 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 4 where Syntology's instrument failed) · 4 unverified) and DSPA: Dynamic SAE Steering for Data-Efficient Preference Alignment (arXiv:2603.21461; that paper's own run record: 5 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 4 where Syntology's instrument failed) · 2 unverified); nothing checks that they implement this paper's method. report
PAIR-code/llm-comparator mentioned on GitHubtfApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to LLM Comparator: Visual Analytics for Side-by-Side Evaluation of Large Language Models (arXiv:2402.10524; that paper's own run record: 1 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
bjoernpl/fasteval mentioned on GitHubApache-2.0Syntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
dongping-chen/mllm-as-a-judge mentioned on GitHubpytorchSyntology: not harvested report
formulamonks/llm-benchmarker-suite mentioned on GitHubpytorchMITSyntology: 7 samples, 4 ran (0 instrument-failed) · 3 unverified report
ilyagusev/ping_pong_bench mentioned on GitHubApache-2.0Syntology: 4 samples, 4 ran (0 instrument-failed) · 0 unverified report
kuk/rulm-sbs2 mentioned on GitHubSyntology: not harvested report
lm-sys/routellm mentioned on GitHubpytorchApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to RouteLLM: Learning to Route LLMs with Preference Data (arXiv:2406.18665; that paper's own run record: 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 2 unverified); nothing checks that they implement this paper's method. report
ojiyumm/mt_bench_rwkv mentioned on GitHubpytorchSyntology: not harvested report
opengvlab/multi-modality-arena mentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to OmniMedVQA: A New Large-Scale Comprehensive Evaluation Benchmark for Medical LVLM (arXiv:2402.09181; that paper's own run record: 7 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 6 where Syntology's instrument failed) · 1 unverified); nothing checks that they implement this paper's method. report
theoremone/llm-benchmarker-suite mentioned on GitHubpytorchMITSyntology: not harvested 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

12 samples harvested: 9 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

9ran
3unverified

Licence: 0 of the 12 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.

bjoernpl/fasteval community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
should_filter_process_output evaluation/models/fastchat.py ran fingerprinted Apache-2.0 (permissive) · 056987a0c7465ba0 · report fixture: signature process_name:str/0/str, line:str/0/str · returned bool · level recorded 2026-09-24
formulamonks/llm-benchmarker-suite community (archive-listed) 7 samples, 4 ran (0 instrument-failed) · 3 unverified
calculate_accuracy in 65 papers metrics/accuracy.py ran fingerprinted MIT (permissive) · 8cd285fe9e20db6b · report fixture: signature true_positives:int/0/int, true_negatives:int/0/int, total_samples:int/0/int · returned float · level recorded 2026-09-25
calculate_f1_score in 14 papers metrics/f1_score.py ran fingerprinted MIT (permissive) · 289ff24d57f60bca · report fixture: signature true_positives:int/0/int, false_positives:int/0/int, false_negatives:int/0/int · returned float · level recorded 2026-09-25
count_true_positives_negatives metrics/utils.py ran MIT (permissive) · 68aa24da5846ab68 · report fixture: signature predictions:list/1/int, targets:list/1/int, positive_label:int/0/int · returned tuple · level recorded 2026-09-24
generate_completion in 5 papers eval_levels/large_language_model.py ran MIT (permissive) · 93e8875d6f69828c · report fixture: signature language_model:callable/0/?, prompt:str/0/str, max_length:int/0/int · returned str · level recorded 2026-09-24
calculate_bleu_score metrics/bleu_score.py unverified MIT (permissive) · 5c196faf66a8f85f · report
get_completion_score eval_levels/gpt_completions_api.py unverified MIT (permissive) · 31d81ed97dc19096 · report
load_language_model eval_levels/large_language_model.py unverified MIT (permissive) · 8b546d8a3cbca5d0 · report
ilyagusev/ping_pong_bench community (archive-listed) 4 samples, 4 ran (0 instrument-failed) · 0 unverified
bootstrap_mean in 5 papers src/build_table_v2.py ran Apache-2.0 (permissive) · 97e1d403e329ddba · report fixture: signature data:list/1/float, n_bootstrap:int/0/int · returned tuple · level recorded 2026-09-24
display_str src/build_table_v2.py ran fingerprinted Apache-2.0 (permissive) · 33a7dc7b274b93b6 · report fixture: signature text:str/0/str · returned str · level recorded 2026-09-24
to_markdown in 6 papers src/browser.py ran Apache-2.0 (permissive) · 5f0936b6dd8764d2 · report fixture: signature record:dict/0/? · returned str · level recorded 2026-09-25
to_meta src/browser.py ran Apache-2.0 (permissive) · 876eef1b6c8e5394 · report fixture: signature record:dict/0/? · returned str · level recorded 2026-09-24

Tasks

ChatbotLanguage ModellingLarge Language ModelLong-Context Understanding

Datasets

Introduced by this paper, per the archive.

MT-Bench

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-13b-v1.5-16k 12k 1.4 #6 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-13b-v1.5-16k 16k 0.9 #6 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-13b-v1.5-16k 1k 53.4 #6 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-13b-v1.5-16k 2k 29.2 #6 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-13b-v1.5-16k 4k 13.1 #6 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-13b-v1.5-16k 6k 4.3 #6 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-13b-v1.5-16k 8k 2.2 #6 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-7b-v1.5-16k 12k 1.9 #8 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-7b-v1.5-16k 16k 1.0 #8 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-7b-v1.5-16k 1k 37.0 #8 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-7b-v1.5-16k 2k 11.1 #8 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-7b-v1.5-16k 4k 5.8 #8 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-7b-v1.5-16k 6k 3.2 #8 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) Vicuna-7b-v1.5-16k 8k 1.8 #8 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) LongChat-7b-v1.5-32k 12k 1.6 #9 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) LongChat-7b-v1.5-32k 16k 0.8 #9 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) LongChat-7b-v1.5-32k 1k 32.4 #9 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) LongChat-7b-v1.5-32k 2k 10.7 #9 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) LongChat-7b-v1.5-32k 4k 5.7 #9 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) LongChat-7b-v1.5-32k 6k 3.1 #9 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (BestAnswer) LongChat-7b-v1.5-32k 8k 1.9 #9 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-13b-v1.5-16k 16k 3.1 #3 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-13b-v1.5-16k 2k 5.4 #3 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-13b-v1.5-16k 4k 5.0 #3 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-13b-v1.5-16k 8k 2.4 #3 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) LongChat-7b-v1.5-32k 16k 2.5 #4 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) LongChat-7b-v1.5-32k 2k 5.3 #4 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) LongChat-7b-v1.5-32k 4k 5.0 #4 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) LongChat-7b-v1.5-32k 8k 3.1 #4 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-7b-v1.5-16k 16k 1.7 #5 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-7b-v1.5-16k 2k 5.3 #5 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-7b-v1.5-16k 4k 2.2 #5 of 10 Archive leaderboard report
Long-Context Understanding Ada-LEval (TSort) Vicuna-7b-v1.5-16k 8k 2.3 #5 of 10 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutGPT-4Label SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

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