{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZDDH5CEBC3WYDZV7J3DU2G7DFT","short_pith_number":"pith:ZDDH5CEB","schema_version":"1.0","canonical_sha256":"c8c67e888116ed81e6bf4ec74d1be32cda95bb08b7cf17e89cd12c848c625006","source":{"kind":"arxiv","id":"2606.31270","version":1},"attestation_state":"computed","paper":{"title":"Learning from Failure: Inference-Time Self-Improvement for Computer-Use Agents","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CY","cs.LG"],"primary_cat":"cs.CV","authors_text":"Ludwig Schmidt, Serena Yeung-Levy, Xiaohan Wang, Xueqiao Sun, Yuhui Zhang","submitted_at":"2026-06-30T07:44:37Z","abstract_excerpt":"Computer-use agents, which leverage multimodal large language models (MLLMs) to operate computers and complete tasks, have attracted significant attention for their utility and versatility. A major challenge in developing these agents is collecting large-scale, high-quality trajectories. The standard approach generates synthetic data through a self-improving loop: an agent is placed in a verifiable environment and iteratively fine-tuned on its successful trajectories. Despite its effectiveness, this paradigm exploits only successful trajectories and discards the failed ones, even though failur"},"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":"2606.31270","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2026-06-30T07:44:37Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CY","cs.LG"],"title_canon_sha256":"d67b2af9867465e9d3823b3782d1f702fc7382d80dae0f2ed50622867c953137","abstract_canon_sha256":"90673e347d1102a3c4adc0fbab7d0d1006d608216a4afab56467579e949cf077"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T01:17:57.680655Z","signature_b64":"xDZ8rX4tNycJ9dKKb7f/6QKwBUTwgQSjzbe0PND+lCTthijyrzH0o5KSaP6NR+rfeM9GRaRjF6QMjEkXqBE6CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c8c67e888116ed81e6bf4ec74d1be32cda95bb08b7cf17e89cd12c848c625006","last_reissued_at":"2026-07-01T01:17:57.680220Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T01:17:57.680220Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning from Failure: Inference-Time Self-Improvement for Computer-Use Agents","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CY","cs.LG"],"primary_cat":"cs.CV","authors_text":"Ludwig Schmidt, Serena Yeung-Levy, Xiaohan Wang, Xueqiao Sun, Yuhui Zhang","submitted_at":"2026-06-30T07:44:37Z","abstract_excerpt":"Computer-use agents, which leverage multimodal large language models (MLLMs) to operate computers and complete tasks, have attracted significant attention for their utility and versatility. A major challenge in developing these agents is collecting large-scale, high-quality trajectories. The standard approach generates synthetic data through a self-improving loop: an agent is placed in a verifiable environment and iteratively fine-tuned on its successful trajectories. Despite its effectiveness, this paradigm exploits only successful trajectories and discards the failed ones, even though failur"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.31270","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/2606.31270/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":"2606.31270","created_at":"2026-07-01T01:17:57.680286+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.31270v1","created_at":"2026-07-01T01:17:57.680286+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.31270","created_at":"2026-07-01T01:17:57.680286+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZDDH5CEBC3WY","created_at":"2026-07-01T01:17:57.680286+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZDDH5CEBC3WYDZV7","created_at":"2026-07-01T01:17:57.680286+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZDDH5CEB","created_at":"2026-07-01T01:17:57.680286+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT","json":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT.json","graph_json":"https://pith.science/api/pith-number/ZDDH5CEBC3WYDZV7J3DU2G7DFT/graph.json","events_json":"https://pith.science/api/pith-number/ZDDH5CEBC3WYDZV7J3DU2G7DFT/events.json","paper":"https://pith.science/paper/ZDDH5CEB"},"agent_actions":{"view_html":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT","download_json":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT.json","view_paper":"https://pith.science/paper/ZDDH5CEB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.31270&json=true","fetch_graph":"https://pith.science/api/pith-number/ZDDH5CEBC3WYDZV7J3DU2G7DFT/graph.json","fetch_events":"https://pith.science/api/pith-number/ZDDH5CEBC3WYDZV7J3DU2G7DFT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT/action/storage_attestation","attest_author":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT/action/author_attestation","sign_citation":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT/action/citation_signature","submit_replication":"https://pith.science/pith/ZDDH5CEBC3WYDZV7J3DU2G7DFT/action/replication_record"}},"created_at":"2026-07-01T01:17:57.680286+00:00","updated_at":"2026-07-01T01:17:57.680286+00:00"}