{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:PZC67D334BO67U6BLGUEPEY2AC","short_pith_number":"pith:PZC67D33","schema_version":"1.0","canonical_sha256":"7e45ef8f7be05defd3c159a847931a008776f57e9bc7c65fdaa29cd985a58489","source":{"kind":"arxiv","id":"2507.00417","version":1},"attestation_state":"computed","paper":{"title":"ASTRO: Teaching Language Models to Reason by Reflecting and Backtracking In-Context","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Anirudh Goyal, Hannaneh Hajishirzi, Joongwon Kim, Liang Tan, Srinivasan Iyer, Tianlu Wang","submitted_at":"2025-07-01T04:10:15Z","abstract_excerpt":"We introduce ASTRO, the \"Autoregressive Search-Taught Reasoner\", a framework for training language models to reason like search algorithms, explicitly leveraging self-reflection, backtracking, and exploration in their outputs. Recently, training large language models (LLMs) via reinforcement learning (RL) has led to the advent of reasoning models with greatly enhanced reasoning capabilities. Open-source replications of reasoning models, while successful, build upon models that already exhibit strong reasoning capabilities along with search behavior observed even before RL. As a result, it is y"},"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":"2507.00417","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-07-01T04:10:15Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"018e02435f72bee751b8400abff6601be874dcf73ff3a3714077a45d3cf220b0","abstract_canon_sha256":"63b36e0b7525603d517a787d7e607ba0ed4363248b33a9fc1bd0d5b57f762c86"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:30:08.558145Z","signature_b64":"TY7HwmDY5BYy0jxOZU5L91urOSsJpeIceJ/koyaAAEBRDoQeUSQdiA2uMUFscZbm8rhB+rFwiwpVCipfLrfnBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7e45ef8f7be05defd3c159a847931a008776f57e9bc7c65fdaa29cd985a58489","last_reissued_at":"2026-07-05T11:30:08.557610Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:30:08.557610Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ASTRO: Teaching Language Models to Reason by Reflecting and Backtracking In-Context","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Anirudh Goyal, Hannaneh Hajishirzi, Joongwon Kim, Liang Tan, Srinivasan Iyer, Tianlu Wang","submitted_at":"2025-07-01T04:10:15Z","abstract_excerpt":"We introduce ASTRO, the \"Autoregressive Search-Taught Reasoner\", a framework for training language models to reason like search algorithms, explicitly leveraging self-reflection, backtracking, and exploration in their outputs. Recently, training large language models (LLMs) via reinforcement learning (RL) has led to the advent of reasoning models with greatly enhanced reasoning capabilities. Open-source replications of reasoning models, while successful, build upon models that already exhibit strong reasoning capabilities along with search behavior observed even before RL. As a result, it is y"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.00417","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/2507.00417/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":"2507.00417","created_at":"2026-07-05T11:30:08.557695+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.00417v1","created_at":"2026-07-05T11:30:08.557695+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.00417","created_at":"2026-07-05T11:30:08.557695+00:00"},{"alias_kind":"pith_short_12","alias_value":"PZC67D334BO6","created_at":"2026-07-05T11:30:08.557695+00:00"},{"alias_kind":"pith_short_16","alias_value":"PZC67D334BO67U6B","created_at":"2026-07-05T11:30:08.557695+00:00"},{"alias_kind":"pith_short_8","alias_value":"PZC67D33","created_at":"2026-07-05T11:30:08.557695+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07492","citing_title":"Search, Fail, Recover: A Training Framework for Correction-Aware Reasoning","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2605.25507","citing_title":"Credit Assignment with Resets in Language Model Reasoning","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06556","citing_title":"Robots Need More than VLA and World Models","ref_index":141,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC","json":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC.json","graph_json":"https://pith.science/api/pith-number/PZC67D334BO67U6BLGUEPEY2AC/graph.json","events_json":"https://pith.science/api/pith-number/PZC67D334BO67U6BLGUEPEY2AC/events.json","paper":"https://pith.science/paper/PZC67D33"},"agent_actions":{"view_html":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC","download_json":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC.json","view_paper":"https://pith.science/paper/PZC67D33","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.00417&json=true","fetch_graph":"https://pith.science/api/pith-number/PZC67D334BO67U6BLGUEPEY2AC/graph.json","fetch_events":"https://pith.science/api/pith-number/PZC67D334BO67U6BLGUEPEY2AC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC/action/storage_attestation","attest_author":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC/action/author_attestation","sign_citation":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC/action/citation_signature","submit_replication":"https://pith.science/pith/PZC67D334BO67U6BLGUEPEY2AC/action/replication_record"}},"created_at":"2026-07-05T11:30:08.557695+00:00","updated_at":"2026-07-05T11:30:08.557695+00:00"}