{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LJWX5MBAUIHHOGJVPOIS2EMTZ3","short_pith_number":"pith:LJWX5MBA","schema_version":"1.0","canonical_sha256":"5a6d7eb020a20e7719357b912d1193cee561536e573d8db87a1f9f0d97955dc9","source":{"kind":"arxiv","id":"2504.01931","version":4},"attestation_state":"computed","paper":{"title":"On the Role of Feedback in Test-Time Scaling of Agentic AI Workflows","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ahmad Beirami, Amrit Singh Bedi, Furong Huang, Hamid Palangi, Jindong Gu, Mohammadreza Pourreza, Nino Scherrer, Ruoxi Sun, Souradip Chakraborty, Tomas Pfister, Yiwen Song","submitted_at":"2025-04-02T17:40:47Z","abstract_excerpt":"Agentic AI workflows (systems that autonomously plan and act) are becoming widespread, yet their task success rate on complex tasks remains low. A promising solution is inference-time alignment, which uses extra compute at test time to improve performance. Inference-time alignment relies on three components: sampling, evaluation, and feedback. While most prior work studies sampling and automatic evaluation, feedback remains underexplored. To study the role of feedback, we introduce Iterative Agent Decoding (IAD), a procedure that repeatedly inserts feedback extracted from different forms of cr"},"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":"2504.01931","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-02T17:40:47Z","cross_cats_sorted":[],"title_canon_sha256":"83dbf0e6dd538e3385e3610c19fc2e5175f10d4173af4306db5dd58d88582549","abstract_canon_sha256":"11c7c920ea5fc185237a663e2f2f3b51d08fc0701ddd8ccea16769d3ea2cc0e9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:33:22.494253Z","signature_b64":"069KIKUC0UHwBKq1XwdP2uK4YYQKm5YAreRIAEWImbHFvB/KBtscVtCO33a9O0tkr8kDQjDeu3PUY0ArjaeTAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5a6d7eb020a20e7719357b912d1193cee561536e573d8db87a1f9f0d97955dc9","last_reissued_at":"2026-07-05T11:33:22.493740Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:33:22.493740Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Role of Feedback in Test-Time Scaling of Agentic AI Workflows","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ahmad Beirami, Amrit Singh Bedi, Furong Huang, Hamid Palangi, Jindong Gu, Mohammadreza Pourreza, Nino Scherrer, Ruoxi Sun, Souradip Chakraborty, Tomas Pfister, Yiwen Song","submitted_at":"2025-04-02T17:40:47Z","abstract_excerpt":"Agentic AI workflows (systems that autonomously plan and act) are becoming widespread, yet their task success rate on complex tasks remains low. A promising solution is inference-time alignment, which uses extra compute at test time to improve performance. Inference-time alignment relies on three components: sampling, evaluation, and feedback. While most prior work studies sampling and automatic evaluation, feedback remains underexplored. To study the role of feedback, we introduce Iterative Agent Decoding (IAD), a procedure that repeatedly inserts feedback extracted from different forms of cr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.01931","kind":"arxiv","version":4},"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/2504.01931/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":"2504.01931","created_at":"2026-07-05T11:33:22.493805+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.01931v4","created_at":"2026-07-05T11:33:22.493805+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.01931","created_at":"2026-07-05T11:33:22.493805+00:00"},{"alias_kind":"pith_short_12","alias_value":"LJWX5MBAUIHH","created_at":"2026-07-05T11:33:22.493805+00:00"},{"alias_kind":"pith_short_16","alias_value":"LJWX5MBAUIHHOGJV","created_at":"2026-07-05T11:33:22.493805+00:00"},{"alias_kind":"pith_short_8","alias_value":"LJWX5MBA","created_at":"2026-07-05T11:33:22.493805+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.09002","citing_title":"Security Considerations for Multi-agent Systems","ref_index":212,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19341","citing_title":"Evaluation-driven Scaling for Scientific Discovery","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3","json":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3.json","graph_json":"https://pith.science/api/pith-number/LJWX5MBAUIHHOGJVPOIS2EMTZ3/graph.json","events_json":"https://pith.science/api/pith-number/LJWX5MBAUIHHOGJVPOIS2EMTZ3/events.json","paper":"https://pith.science/paper/LJWX5MBA"},"agent_actions":{"view_html":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3","download_json":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3.json","view_paper":"https://pith.science/paper/LJWX5MBA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.01931&json=true","fetch_graph":"https://pith.science/api/pith-number/LJWX5MBAUIHHOGJVPOIS2EMTZ3/graph.json","fetch_events":"https://pith.science/api/pith-number/LJWX5MBAUIHHOGJVPOIS2EMTZ3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3/action/storage_attestation","attest_author":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3/action/author_attestation","sign_citation":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3/action/citation_signature","submit_replication":"https://pith.science/pith/LJWX5MBAUIHHOGJVPOIS2EMTZ3/action/replication_record"}},"created_at":"2026-07-05T11:33:22.493805+00:00","updated_at":"2026-07-05T11:33:22.493805+00:00"}