{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:WOXYTTAQMZUNXQW4AFQJ3TDAZL","short_pith_number":"pith:WOXYTTAQ","schema_version":"1.0","canonical_sha256":"b3af89cc106668dbc2dc01609dcc60cafc0b6cb464f1ed643c8adb885e550096","source":{"kind":"arxiv","id":"2405.20309","version":2},"attestation_state":"computed","paper":{"title":"Large Language Models Can Self-Improve At Web Agent Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ajay Patel, Chris Callison-Burch, Claudiu Leoveanu-Condrei, Marius-Constantin Dinu, Markus Hofmarcher, Sepp Hochreiter","submitted_at":"2024-05-30T17:52:36Z","abstract_excerpt":"Training models to act as agents that can effectively navigate and perform actions in a complex environment, such as a web browser, has typically been challenging due to lack of training data. Large language models (LLMs) have recently demonstrated some capability to navigate novel environments as agents in a zero-shot or few-shot fashion, purely guided by natural language instructions as prompts. Recent research has also demonstrated LLMs have the capability to exceed their base performance through self-improvement, i.e. fine-tuning on data generated by the model itself. In this work, we expl"},"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":"2405.20309","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-30T17:52:36Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"0979da4599177840ce2e5b049399fd17e310c67e3bd869fa729787caaa23b039","abstract_canon_sha256":"561eee3c43dce1bf8bde0d3374668683d2fdd0c27823e477630dfa313870c259"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:14:33.763029Z","signature_b64":"9wAo7stGTAZGNIpI9coOpr6OG2OBTbSDq0c4u3qqN2OWjbWBaPRUw+a9SpTq+Az1b1beiJc0ujIR24lVzvrjBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b3af89cc106668dbc2dc01609dcc60cafc0b6cb464f1ed643c8adb885e550096","last_reissued_at":"2026-07-05T09:14:33.762491Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:14:33.762491Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large Language Models Can Self-Improve At Web Agent Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ajay Patel, Chris Callison-Burch, Claudiu Leoveanu-Condrei, Marius-Constantin Dinu, Markus Hofmarcher, Sepp Hochreiter","submitted_at":"2024-05-30T17:52:36Z","abstract_excerpt":"Training models to act as agents that can effectively navigate and perform actions in a complex environment, such as a web browser, has typically been challenging due to lack of training data. Large language models (LLMs) have recently demonstrated some capability to navigate novel environments as agents in a zero-shot or few-shot fashion, purely guided by natural language instructions as prompts. Recent research has also demonstrated LLMs have the capability to exceed their base performance through self-improvement, i.e. fine-tuning on data generated by the model itself. In this work, we expl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.20309","kind":"arxiv","version":2},"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/2405.20309/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":"2405.20309","created_at":"2026-07-05T09:14:33.762552+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.20309v2","created_at":"2026-07-05T09:14:33.762552+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.20309","created_at":"2026-07-05T09:14:33.762552+00:00"},{"alias_kind":"pith_short_12","alias_value":"WOXYTTAQMZUN","created_at":"2026-07-05T09:14:33.762552+00:00"},{"alias_kind":"pith_short_16","alias_value":"WOXYTTAQMZUNXQW4","created_at":"2026-07-05T09:14:33.762552+00:00"},{"alias_kind":"pith_short_8","alias_value":"WOXYTTAQ","created_at":"2026-07-05T09:14:33.762552+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12191","citing_title":"Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application","ref_index":286,"is_internal_anchor":false},{"citing_arxiv_id":"2511.00739","citing_title":"Towards Understanding, Analyzing, and Optimizing Agentic AI Execution: A CPU-Centric Perspective","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2503.09572","citing_title":"Plan-and-Act: Improving Planning of Agents for Long-Horizon Tasks","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2603.26041","citing_title":"Rethinking Token Pruning for Historical Screenshots in GUI Visual Agents: Semantic, Spatial, and Temporal Perspectives","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL","json":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL.json","graph_json":"https://pith.science/api/pith-number/WOXYTTAQMZUNXQW4AFQJ3TDAZL/graph.json","events_json":"https://pith.science/api/pith-number/WOXYTTAQMZUNXQW4AFQJ3TDAZL/events.json","paper":"https://pith.science/paper/WOXYTTAQ"},"agent_actions":{"view_html":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL","download_json":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL.json","view_paper":"https://pith.science/paper/WOXYTTAQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.20309&json=true","fetch_graph":"https://pith.science/api/pith-number/WOXYTTAQMZUNXQW4AFQJ3TDAZL/graph.json","fetch_events":"https://pith.science/api/pith-number/WOXYTTAQMZUNXQW4AFQJ3TDAZL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL/action/storage_attestation","attest_author":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL/action/author_attestation","sign_citation":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL/action/citation_signature","submit_replication":"https://pith.science/pith/WOXYTTAQMZUNXQW4AFQJ3TDAZL/action/replication_record"}},"created_at":"2026-07-05T09:14:33.762552+00:00","updated_at":"2026-07-05T09:14:33.762552+00:00"}