{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/from-crowdsourced-data-to-high-quality","title":"From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline","arxiv_id":"2406.11939","date":"2024-06-17","proceeding":null,"authors":["Tianle Li","Wei-Lin Chiang","Evan Frick","Lisa Dunlap","Tianhao Wu","Banghua Zhu","Joseph E. Gonzalez","Ion Stoica"],"abstract":"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.","url_abs":"https://arxiv.org/abs/2406.11939v2","url_pdf":"https://arxiv.org/pdf/2406.11939v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"from-crowdsourced-data-to-high-quality","repo_url":"https://github.com/lm-sys/arena-hard-auto","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"from-crowdsourced-data-to-high-quality","repo_url":"https://github.com/lmarena/arena-hard-auto","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"from-crowdsourced-data-to-high-quality","repo_url":"https://github.com/lm-sys/arena-hard","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"from-crowdsourced-data-to-high-quality","repo_url":"https://github.com/lmarena/ppe","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"chatbot","task_name":"Chatbot"}],"methods":[{"method_slug":"align","method_name":"ALIGN"}],"datasets_introduced":[{"slug":"arena-hard-auto","name":"Arena-Hard-Auto","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2406.11939","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2406.11939"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lm-sys/arena-hard","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lmarena/arena-hard-auto","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lmarena/ppe","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lm-sys/arena-hard-auto","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"ran":17,"ran_draft_wrong":1,"unverified":3},"by_repo_kind":{"official":{"samples":4,"ran":4,"repositories":1},"listed":{"samples":17,"ran":14,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"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"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"06750b8d972efe27","entry":"calculate_score","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"BenchBuilder/filter.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/BenchBuilder/filter.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"06750b8d972efe27"}},{"code_sha256_prefix":"df075cc813246dcd","entry":"count_markdown_elements","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/add_markdown_info.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/add_markdown_info.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"df075cc813246dcd"}},{"code_sha256_prefix":"da15e93460148f70","entry":"create_llama3_body","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/bedrock_utils.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/bedrock_utils.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"da15e93460148f70"}},{"code_sha256_prefix":"6743a5e05fc4ad6d","entry":"create_nova_messages","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/bedrock_utils.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/bedrock_utils.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"6743a5e05fc4ad6d"}},{"code_sha256_prefix":"759c3030a0ca985f","entry":"extract_innermost_text","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/bedrock_utils.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/bedrock_utils.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"759c3030a0ca985f"}},{"code_sha256_prefix":"bf60419bbfec3311","entry":"format_confidence_interval","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"show_result.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/show_result.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"bf60419bbfec3311"}},{"code_sha256_prefix":"31d3137625992d50","entry":"get_categories","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"qa_browser.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/qa_browser.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"31d3137625992d50"}},{"code_sha256_prefix":"0a127509b23580f5","entry":"get_element_counts","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/add_markdown_info.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/add_markdown_info.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0a127509b23580f5"}},{"code_sha256_prefix":"c8136157d71c093b","entry":"get_endpoint","repo":"lm-sys/arena-hard-auto","repo_kind":"official","path":"BenchBuilder/label.py","file_url":"https://github.com/lm-sys/arena-hard-auto/blob/HEAD/BenchBuilder/label.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c8136157d71c093b"}},{"code_sha256_prefix":"7ca039993ebdcc43","entry":"get_languages","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"qa_browser.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/qa_browser.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7ca039993ebdcc43"}},{"code_sha256_prefix":"b1fbdabfc009df1c","entry":"get_model_style_metadata","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"show_result.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/show_result.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"b1fbdabfc009df1c"}},{"code_sha256_prefix":"6805dfa84dd06130","entry":"load_json","repo":"lm-sys/arena-hard-auto","repo_kind":"official","path":"BenchBuilder/filter.py","file_url":"https://github.com/lm-sys/arena-hard-auto/blob/HEAD/BenchBuilder/filter.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"6805dfa84dd06130"}},{"code_sha256_prefix":"d049c92b5f5adfa3","entry":"load_jsonl","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"qa_browser.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/qa_browser.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"d049c92b5f5adfa3"}},{"code_sha256_prefix":"47892c5631a199d4","entry":"load_questions","repo":"lm-sys/arena-hard-auto","repo_kind":"official","path":"utils/completion.py","file_url":"https://github.com/lm-sys/arena-hard-auto/blob/HEAD/utils/completion.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"47892c5631a199d4"}},{"code_sha256_prefix":"07e6f41cf857a76f","entry":"make_config","repo":"lm-sys/arena-hard-auto","repo_kind":"official","path":"BenchBuilder/label.py","file_url":"https://github.com/lm-sys/arena-hard-auto/blob/HEAD/BenchBuilder/label.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"07e6f41cf857a76f"}},{"code_sha256_prefix":"7a5ee66f6fe09b60","entry":"register_api","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/completion.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/completion.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7a5ee66f6fe09b60"}},{"code_sha256_prefix":"0399a259308f77a0","entry":"register_engine","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/completion.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/completion.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0399a259308f77a0"}},{"code_sha256_prefix":"bae70f745e7b2d09","entry":"remove_pattern","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"utils/add_markdown_info.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/utils/add_markdown_info.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"bae70f745e7b2d09"}},{"code_sha256_prefix":"0322093a9ad847a4","entry":"chat_completion_openai","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"BenchBuilder/label.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/BenchBuilder/label.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0322093a9ad847a4"}},{"code_sha256_prefix":"e094bc160bb7d823","entry":"load_jsonl","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"BenchBuilder/filter.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/BenchBuilder/filter.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e094bc160bb7d823"}},{"code_sha256_prefix":"ff96063cc3cfa7d6","entry":"load_judgments","repo":"lm-sys/arena-hard","repo_kind":"listed","path":"show_result.py","file_url":"https://github.com/lm-sys/arena-hard/blob/HEAD/show_result.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ff96063cc3cfa7d6"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}