{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4AZXZF4L4TDULFMIFWYGQZBY7M","short_pith_number":"pith:4AZXZF4L","schema_version":"1.0","canonical_sha256":"e0337c978be4c74595882db0686438fb2102c50965628f24a1dd32e0f2c57cf9","source":{"kind":"arxiv","id":"2506.13705","version":1},"attestation_state":"computed","paper":{"title":"TimeMaster: Training Time-Series Multimodal LLMs to Reason via Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Duanqing Xu, Junru Zhang, Lang Feng, Xu Guo, Yabo Dong, Yuhan Wu","submitted_at":"2025-06-16T17:12:26Z","abstract_excerpt":"Time-series reasoning remains a significant challenge in multimodal large language models (MLLMs) due to the dynamic temporal patterns, ambiguous semantics, and lack of temporal priors. In this work, we introduce TimeMaster, a reinforcement learning (RL)-based method that enables time-series MLLMs to perform structured, interpretable reasoning directly over visualized time-series inputs and task prompts. TimeMaster adopts a three-part structured output format, reasoning, classification, and domain-specific extension, and is optimized via a composite reward function that aligns format adherence"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2506.13705","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-16T17:12:26Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"a0d9e2520065de50f900df505456646e3115b54f3d4e7e41dafcae3d8fc3625c","abstract_canon_sha256":"fa6ce0770b4d3d0c053b815ac85155f4bd5c40df7bdceae18c20dcfe060ebee3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:22:25.269561Z","signature_b64":"JmWEoa0Q/jN9ulXywWHWZZbxn93Y/3T1uxvIdbEVq+qvosGE6SXtn0wyVTm5w0h55DofXPGM5cdMIVCZel0TBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e0337c978be4c74595882db0686438fb2102c50965628f24a1dd32e0f2c57cf9","last_reissued_at":"2026-07-05T11:22:25.269011Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:22:25.269011Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TimeMaster: Training Time-Series Multimodal LLMs to Reason via Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Duanqing Xu, Junru Zhang, Lang Feng, Xu Guo, Yabo Dong, Yuhan Wu","submitted_at":"2025-06-16T17:12:26Z","abstract_excerpt":"Time-series reasoning remains a significant challenge in multimodal large language models (MLLMs) due to the dynamic temporal patterns, ambiguous semantics, and lack of temporal priors. In this work, we introduce TimeMaster, a reinforcement learning (RL)-based method that enables time-series MLLMs to perform structured, interpretable reasoning directly over visualized time-series inputs and task prompts. TimeMaster adopts a three-part structured output format, reasoning, classification, and domain-specific extension, and is optimized via a composite reward function that aligns format adherence"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.13705","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2506.13705/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2506.13705","created_at":"2026-07-05T11:22:25.269073+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.13705v1","created_at":"2026-07-05T11:22:25.269073+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.13705","created_at":"2026-07-05T11:22:25.269073+00:00"},{"alias_kind":"pith_short_12","alias_value":"4AZXZF4L4TDU","created_at":"2026-07-05T11:22:25.269073+00:00"},{"alias_kind":"pith_short_16","alias_value":"4AZXZF4L4TDULFMI","created_at":"2026-07-05T11:22:25.269073+00:00"},{"alias_kind":"pith_short_8","alias_value":"4AZXZF4L","created_at":"2026-07-05T11:22:25.269073+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23214","citing_title":"Universal Interatomic Potentials as Configuration-Space Generators for One-Shot and Iterative Fine-Tuning of Ab Initio-Accurate Material-Specific Models","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2404.13004","citing_title":"A Unified Framework for Modeling Heterogeneous Financial Data via Dual-Granularity Prompting","ref_index":60,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M","json":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M.json","graph_json":"https://pith.science/api/pith-number/4AZXZF4L4TDULFMIFWYGQZBY7M/graph.json","events_json":"https://pith.science/api/pith-number/4AZXZF4L4TDULFMIFWYGQZBY7M/events.json","paper":"https://pith.science/paper/4AZXZF4L"},"agent_actions":{"view_html":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M","download_json":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M.json","view_paper":"https://pith.science/paper/4AZXZF4L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.13705&json=true","fetch_graph":"https://pith.science/api/pith-number/4AZXZF4L4TDULFMIFWYGQZBY7M/graph.json","fetch_events":"https://pith.science/api/pith-number/4AZXZF4L4TDULFMIFWYGQZBY7M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M/action/storage_attestation","attest_author":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M/action/author_attestation","sign_citation":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M/action/citation_signature","submit_replication":"https://pith.science/pith/4AZXZF4L4TDULFMIFWYGQZBY7M/action/replication_record"}},"created_at":"2026-07-05T11:22:25.269073+00:00","updated_at":"2026-07-05T11:22:25.269073+00:00"}