{"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/process-reward-models-that-think","title":"Process Reward Models That Think","arxiv_id":"2504.16828","date":"2025-04-23","proceeding":null,"authors":["Muhammad Khalifa","Rishabh Agarwal","Lajanugen Logeswaran","Jaekyeom Kim","Hao Peng","Moontae Lee","Honglak Lee","Lu Wang"],"abstract":"Step-by-step verifiers -- also known as process reward models (PRMs) -- are a key ingredient for test-time scaling. PRMs require step-level supervision, making them expensive to train. This work aims to build data-efficient PRMs as verbalized step-wise reward models that verify every step in the solution by generating a verification chain-of-thought (CoT). We propose ThinkPRM, a long CoT verifier fine-tuned on orders of magnitude fewer process labels than those required by discriminative PRMs. Our approach capitalizes on the inherent reasoning abilities of long CoT models, and outperforms LLM-as-a-Judge and discriminative verifiers -- using only 1% of the process labels in PRM800K -- across several challenging benchmarks. Specifically, ThinkPRM beats the baselines on ProcessBench, MATH-500, and AIME '24 under best-of-N selection and reward-guided search. In an out-of-domain evaluation on a subset of GPQA-Diamond and LiveCodeBench, our PRM surpasses discriminative verifiers trained on the full PRM800K by 8% and 4.5%, respectively. Lastly, under the same token budget, ThinkPRM scales up verification compute more effectively compared to LLM-as-a-Judge, outperforming it by 7.2% on a subset of ProcessBench. Our work highlights the value of generative, long CoT PRMs that can scale test-time compute for verification while requiring minimal supervision for training. Our code, data, and models will be released at https://github.com/mukhal/thinkprm.","url_abs":"https://arxiv.org/abs/2504.16828v2","url_pdf":"https://arxiv.org/pdf/2504.16828v2.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":"process-reward-models-that-think","repo_url":"https://github.com/mukhal/thinkprm","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"math","task_name":"Math"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2504.16828","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2504.16828"}},"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/mukhal/thinkprm","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":1,"unverified":9},"by_repo_kind":{"official":{"samples":10,"ran":1,"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":"1525d8bc0fc8d5a0","entry":"generate_completions","repo":"mukhal/thinkprm","repo_kind":"official","path":"search-and-learn/evaluation/model_utils.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/search-and-learn/evaluation/model_utils.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1525d8bc0fc8d5a0"}},{"code_sha256_prefix":"9bb1a529691054e6","entry":"clean_traj","repo":"mukhal/thinkprm","repo_kind":"official","path":"dataset/prm_dataset.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/dataset/prm_dataset.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"9bb1a529691054e6"}},{"code_sha256_prefix":"97eb4c263188e55d","entry":"format_train_verification_cot_for_thinkprm","repo":"mukhal/thinkprm","repo_kind":"official","path":"utils/helper.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/utils/helper.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"97eb4c263188e55d"}},{"code_sha256_prefix":"967e93846e542206","entry":"format_verification_cot_for_thinkprm","repo":"mukhal/thinkprm","repo_kind":"official","path":"utils/helper.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/utils/helper.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"967e93846e542206"}},{"code_sha256_prefix":"388299eedc9e4ec2","entry":"format_verification_cot_no_think","repo":"mukhal/thinkprm","repo_kind":"official","path":"utils/helper.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/utils/helper.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"388299eedc9e4ec2"}},{"code_sha256_prefix":"34f3f784f648c968","entry":"judge_answer","repo":"mukhal/thinkprm","repo_kind":"official","path":"utils/answer_utils.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/utils/answer_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"34f3f784f648c968"}},{"code_sha256_prefix":"ee71a94d9dd1b8e9","entry":"load_hf_lm_and_tokenizer","repo":"mukhal/thinkprm","repo_kind":"official","path":"search-and-learn/evaluation/model_utils.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/search-and-learn/evaluation/model_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"ee71a94d9dd1b8e9"}},{"code_sha256_prefix":"1a2bbb70c3c71e1e","entry":"preprocess_longcot","repo":"mukhal/thinkprm","repo_kind":"official","path":"prm/thinkprm_api.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/prm/thinkprm_api.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1a2bbb70c3c71e1e"}},{"code_sha256_prefix":"f319752266d199ba","entry":"retrieve_answer","repo":"mukhal/thinkprm","repo_kind":"official","path":"utils/answer_utils.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/utils/answer_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f319752266d199ba"}},{"code_sha256_prefix":"83252db0b404cdb0","entry":"retrieve_answer_from_dataset","repo":"mukhal/thinkprm","repo_kind":"official","path":"utils/answer_utils.py","file_url":"https://github.com/mukhal/thinkprm/blob/HEAD/utils/answer_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"83252db0b404cdb0"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}