{"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/arxiv-2603-09403","title":"LLM as a Meta-Judge: Synthetic Data for NLP Evaluation Metric Validation","arxiv_id":"2603.09403","date":"2026-03-10","proceeding":null,"authors":["Lukáš Eigler","Jindřich Libovický","David Hurych"],"abstract":"Validating evaluation metrics for NLG typically relies on expensive and time-consuming human annotations, which predominantly exist only for English datasets. We propose LLM as a Meta-Judge, a scalable framework that utilizes LLMs to generate synthetic evaluation datasets via controlled semantic degradation of real data, replacing human judgment. We validate our approach using meta-correlation, measuring the alignment between metric rankings derived from synthetic data and those from standard human benchmarks. Experiments across Machine Translation, Question Answering, and Summarization demonstrate that synthetic validation serves as a reliable proxy for human judgment, achieving meta-correlations exceeding 0.9 in multilingual QA and proves to be a viable alternative where human judgments are unavailable or too expensive to obtain. Our code and data are publicly available at https://github.com/eiglerl/meta-judge.","url_abs":"https://arxiv.org/abs/2603.09403","url_pdf":"https://arxiv.org/pdf/2603.09403","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2603.09403","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2603.09403"}},"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":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/eiglerl/meta-judge","reach":null}],"summary":{"unverified":11},"by_repo_kind":{"found_in_text":{"samples":11,"ran":0,"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":11,"samples":[{"code_sha256_prefix":"e132bad3afe9ef9d","entry":"compute_pair_meta_correlation","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/metrics/meta_correlation.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/metrics/meta_correlation.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"e132bad3afe9ef9d"}},{"code_sha256_prefix":"f6ea80c8d379467f","entry":"extract_curves","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/finetuning/plot_training_curves.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/finetuning/plot_training_curves.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"f6ea80c8d379467f"}},{"code_sha256_prefix":"5f9303d1a6f27bd6","entry":"extract_response","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/finetuning/train_grpo.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/finetuning/train_grpo.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"5f9303d1a6f27bd6"}},{"code_sha256_prefix":"3cd33267f77e0022","entry":"find_last_trainer_state","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/finetuning/plot_training_curves.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/finetuning/plot_training_curves.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3cd33267f77e0022"}},{"code_sha256_prefix":"b09dbc63544053c4","entry":"load_json","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/metrics/meta_correlation.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/metrics/meta_correlation.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b09dbc63544053c4"}},{"code_sha256_prefix":"195f354f266a47ed","entry":"parse_id","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/generation/analyze_metric_failures.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/generation/analyze_metric_failures.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"195f354f266a47ed"}},{"code_sha256_prefix":"6e321f0d298cf07d","entry":"process_parts","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/generation/analyze_reasoning_quality.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/generation/analyze_reasoning_quality.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"6e321f0d298cf07d"}},{"code_sha256_prefix":"2352b202bfd29f4b","entry":"resolve_data_sources","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/finetuning/train_sft.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/finetuning/train_sft.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"2352b202bfd29f4b"}},{"code_sha256_prefix":"04b0f324f7c8bb9c","entry":"safe_corr","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/metrics/meta_correlation.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/metrics/meta_correlation.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"04b0f324f7c8bb9c"}},{"code_sha256_prefix":"85eb13a5e936ad11","entry":"smooth","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/finetuning/plot_training_curves.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/finetuning/plot_training_curves.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"85eb13a5e936ad11"}},{"code_sha256_prefix":"7dc44e93ccf63aff","entry":"strip_thinking","repo":"eiglerl/meta-judge","repo_kind":"found_in_text","path":"src/finetuning/train_grpo.py","file_url":"https://github.com/eiglerl/meta-judge/blob/HEAD/src/finetuning/train_grpo.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7dc44e93ccf63aff"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.CL","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}