{"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-2606-25556","title":"BiPACE: Bisimulation-Guided Policy Optimization with Action Counterfactual Estimation for LLM Agents","arxiv_id":"2606.25556","date":"2026-06-24","proceeding":null,"authors":["Hanyang Wang","Weijieying Ren","Yuxiang Zhang","Ding Cao","Zhizhao Zeng","Ke Zeng","Tianxiang Zhao"],"abstract":"Stepwise group-based RL is an attractive way to train long-horizon LLM agents without a learned critic: it reuses multiple sampled rollouts to estimate local advantages. Its weakness is less visible but more fundamental: every group-relative estimator assumes that the steps it compares are equivalent for credit assignment. We show that current agentic variants violate this assumption through a state-action credit mismatch. The observation-hash partition is overly fine on the state side, creating singleton groups with zero step-level signal, while a single within-group mean is too coarse on the action side, mixing state-value estimation with action-specific credit. We introduce BiPACE (Bisimulation-Guided Policy Optimization with Action Counterfactual Estimation), a drop-in advantage estimator that fixes both sides without adding a critic, auxiliary loss, or extra rollouts. BiGPO clusters steps by cosine distance in the actor's own hidden-state geometry, an empirical policy-induced proxy for bisimulation that substantially lowers the singleton rate left by observation hashing. PACE then recenters returns within each behavioral cluster using action-conditioned peer baselines; its Q-style instance estimates a local Q(s,a)-V(s) nonparametrically. On ALFWorld/Qwen2.5-7B, BiPACE_Q raises overall validation success from GiGPO's 90.8 to $97.1\\pm0.9$ over three seeds, and crosses the 95% threshold on every seed, which GiGPO never does within the same budget. On Qwen2.5-1.5B it reaches $93.5\\pm1.2$ versus GiGPO's 86.7, and on WebShop and TextCraft it improves over GRPO and GiGPO at both model scales. The measured BiPACE-specific overhead is 11.3% of a single training-step wall time. Yet it changes the estimator's comparison unit from surface identity to approximate behavioral equivalence plus action-side counterfactuals. The code is available at https://github.com/TianxiangZhao/BiPACE.","url_abs":"https://arxiv.org/abs/2606.25556","url_pdf":"https://arxiv.org/pdf/2606.25556","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=2606.25556","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2606.25556"}},"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/TianxiangZhao/BiPACE","reach":null}],"summary":{"ran":2,"ran_fixture":2,"ran_draft_wrong":1},"by_repo_kind":{"found_in_text":{"samples":5,"ran":5,"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":"56b20aecce68bcc2","entry":"HashNgramEmbedder","repo":"TianxiangZhao/BiPACE","repo_kind":"found_in_text","path":"gigpo/core_bigpo.py","file_url":"https://github.com/TianxiangZhao/BiPACE/blob/HEAD/gigpo/core_bigpo.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"well_formed","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"56b20aecce68bcc2"}},{"code_sha256_prefix":"f6c1765049ac517a","entry":"_get_logger","repo":"TianxiangZhao/BiPACE","repo_kind":"found_in_text","path":"gigpo/core_bigpo.py","file_url":"https://github.com/TianxiangZhao/BiPACE/blob/HEAD/gigpo/core_bigpo.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":"f6c1765049ac517a"}},{"code_sha256_prefix":"cc824d022da01a5b","entry":"_greedy_cosine_cluster","repo":"TianxiangZhao/BiPACE","repo_kind":"found_in_text","path":"gigpo/core_bigpo.py","file_url":"https://github.com/TianxiangZhao/BiPACE/blob/HEAD/gigpo/core_bigpo.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"cc824d022da01a5b"}},{"code_sha256_prefix":"4286994501fe1ec2","entry":"_to_text","repo":"TianxiangZhao/BiPACE","repo_kind":"found_in_text","path":"gigpo/core_bigpo.py","file_url":"https://github.com/TianxiangZhao/BiPACE/blob/HEAD/gigpo/core_bigpo.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"4286994501fe1ec2"}},{"code_sha256_prefix":"781da0b3c97a6650","entry":"build_step_group_bisim","repo":"TianxiangZhao/BiPACE","repo_kind":"found_in_text","path":"gigpo/core_bigpo.py","file_url":"https://github.com/TianxiangZhao/BiPACE/blob/HEAD/gigpo/core_bigpo.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"DEP_MISSING","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"781da0b3c97a6650"}}]},"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":null}