{"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-2607-04364","title":"RL Forgets! Towards Continual Policy Optimization","arxiv_id":"2607.04364","date":"2026-07-05","proceeding":null,"authors":["Mao-Lin Luo","Zhe-Xu Wang","Zi-Hao Zhou","Bo Ye","Jian Zhao","Min-Ling Zhang","Tong Wei"],"abstract":"Continual post-training is becoming a central paradigm for adapting vision-language models to evolving tasks. Recent work has increasingly favored reinforcement learning over supervised fine-tuning, driven by the belief that reinforcement learning is inherently less prone to forgetting. However, the belief remains insufficiently validated, as existing evidence is largely drawn from outdated or homogeneous benchmarks. We revisit this assumption under recent and diverse multimodal reasoning tasks. To this end, we introduce MRCL, a Multimodal Reasoning Continual Learning benchmark. Experiments on MRCL show that standard reinforcement learning still suffers from severe catastrophic forgetting during continual post-training. We trace this failure to an objective mismatch: the KL regularization used in common policy optimization methods is evaluated on current-task data, whereas forgetting is caused by behavioral drift on prior-task distributions. To address this problem, we propose Continual Policy Optimization (CPO), a replay-free framework grounded in a prior-task behavioral KL objective. CPO relaxes the intractable historical KL constraint into sparse parameter-movement regularization, limiting policy drift without storing old data. Extensive experiments across multiple model scales show that CPO consistently reduces forgetting while preserving, and in some cases improving, pretrained model capabilities. On Qwen3-VL-8B, CPO reduces forgetting by 13.7% and improves pretrained capability by 7.0%. The implementation code is available at https://github.com/MaolinLuo/CPO.","url_abs":"https://arxiv.org/abs/2607.04364","url_pdf":"https://arxiv.org/pdf/2607.04364","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=2607.04364","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2607.04364"}},"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/MaolinLuo/CPO","reach":null}],"summary":{"ran":12,"unverified":4},"by_repo_kind":{"found_in_text":{"samples":16,"ran":12,"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":"f7bc07d97a8d5304","entry":"build_content_list_with_interleaved_images","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/dataset/grpo_dataset.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/dataset/grpo_dataset.py","link_basis":"first_harvest_node","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":"f7bc07d97a8d5304"}},{"code_sha256_prefix":"0ad8fc4d50d8c375","entry":"build_content_list_with_interleaved_images","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/dataset/sft_dataset.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/dataset/sft_dataset.py","link_basis":"first_harvest_node","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":"0ad8fc4d50d8c375"}},{"code_sha256_prefix":"fab9bb76d601b06a","entry":"clean_expression","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/train/reward_funcs.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/train/reward_funcs.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"fab9bb76d601b06a"}},{"code_sha256_prefix":"7b5840b320919712","entry":"extract_last_boxed","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/eval/inference.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/eval/inference.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7b5840b320919712"}},{"code_sha256_prefix":"fa6d2859d3461c9e","entry":"get_model_name_from_path","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/utils.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/utils.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"fa6d2859d3461c9e"}},{"code_sha256_prefix":"d8fd759aa8357c1e","entry":"is_lora_model","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/utils.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/utils.py","link_basis":"first_harvest_node","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":"d8fd759aa8357c1e"}},{"code_sha256_prefix":"3be2d5c5ed531180","entry":"parse_image_placeholders","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/dataset/grpo_dataset.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/dataset/grpo_dataset.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"3be2d5c5ed531180"}},{"code_sha256_prefix":"c71bdd2f1919c76b","entry":"parse_image_placeholders","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/dataset/sft_dataset.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/dataset/sft_dataset.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"c71bdd2f1919c76b"}},{"code_sha256_prefix":"22c6606760ef21db","entry":"parse_options","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/eval/inference.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/eval/inference.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"22c6606760ef21db"}},{"code_sha256_prefix":"e93a985e3fa84beb","entry":"process_image","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/eval/inference.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/eval/inference.py","link_basis":"first_harvest_node","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":"e93a985e3fa84beb"}},{"code_sha256_prefix":"7188da24beec6bd9","entry":"remove_vision_tokens","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/dataset/grpo_dataset.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/dataset/grpo_dataset.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7188da24beec6bd9"}},{"code_sha256_prefix":"533c2d12f5c89416","entry":"truncate_sequence","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/dataset/data_utils.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/dataset/data_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":"533c2d12f5c89416"}},{"code_sha256_prefix":"80cc18ed2ef66de3","entry":"extract_weights","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/train/extract_weights.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/train/extract_weights.py","link_basis":"first_harvest_node","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":"80cc18ed2ef66de3"}},{"code_sha256_prefix":"b233a30c3d5f6f11","entry":"load_full_model","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/compute_importance_mask.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/compute_importance_mask.py","link_basis":"first_harvest_node","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":"b233a30c3d5f6f11"}},{"code_sha256_prefix":"3a9a6c8b3e97ccf8","entry":"load_state_dict_cpu","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/compute_importance_mask.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/compute_importance_mask.py","link_basis":"first_harvest_node","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":"3a9a6c8b3e97ccf8"}},{"code_sha256_prefix":"f4c34181924a9fa4","entry":"merge_gradient_chunks","repo":"MaolinLuo/CPO","repo_kind":"found_in_text","path":"src/train/extract_gradients.py","file_url":"https://github.com/MaolinLuo/CPO/blob/HEAD/src/train/extract_gradients.py","link_basis":"first_harvest_node","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":"f4c34181924a9fa4"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":{"kind":"leaderboard_placements","source":"Syntology's leaderboard-shaped extractor over the paper's own arXiv-HTML tables: a model pointed at a cell, the number was read from that cell and checked against the board's metric, dataset, split and scale, and an independent check accepted the entry; not reviewed by the paper's authors or the archive's editors","extractor_model":"global.anthropic.claude-sonnet-4-5-20250929-v1:0","verifier_model":null,"prompt_sha":"fa63d4bb9d755694","coverage":{"sentence":"Syntology has checked 6,264 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","papers_newer_than_archive":9581,"papers_checked":6264},"entries":[],"not_placed":{"boards":1,"rejected_by_independent_check":0,"refused_by_a_rule":1,"check_did_not_answer":0,"proposed_without_a_cell":0,"declined_by_site":0}}}