{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:X5FH5XIF7ZITC7SYNOXKUQAVBX","short_pith_number":"pith:X5FH5XIF","schema_version":"1.0","canonical_sha256":"bf4a7edd05fe51317e586baeaa40150dc26c761294f8d2d927ab0235440f340a","source":{"kind":"arxiv","id":"2405.02749","version":1},"attestation_state":"computed","paper":{"title":"Sub-goal Distillation: A Method to Improve Small Language Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Elias Stengel-Eskin, Marc-Alexandre Cote, Maryam Hashemzadeh, Sarath Chandar","submitted_at":"2024-05-04T20:34:06Z","abstract_excerpt":"While Large Language Models (LLMs) have demonstrated significant promise as agents in interactive tasks, their substantial computational requirements and restricted number of calls constrain their practical utility, especially in long-horizon interactive tasks such as decision-making or in scenarios involving continuous ongoing tasks. To address these constraints, we propose a method for transferring the performance of an LLM with billions of parameters to a much smaller language model (770M parameters). Our approach involves constructing a hierarchical agent comprising a planning module, whic"},"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":"2405.02749","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-04T20:34:06Z","cross_cats_sorted":[],"title_canon_sha256":"c22cdf79c0fd100b335ab85afb61d3f77051711282edbc34897675618ee270c6","abstract_canon_sha256":"32c9e2ef50df0f9fc6a6d800be2eb33b2812d40042796abe7b1626ee7c4e16ca"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:15:53.244628Z","signature_b64":"aVHB4yV8LFEtV42sQmPInU0Z+T6WwyDvfRsDhUrJ9R8+t+PGEOLRd7L+xDTvl3mE1pHD1o1vcTs/P7UAiJF2DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bf4a7edd05fe51317e586baeaa40150dc26c761294f8d2d927ab0235440f340a","last_reissued_at":"2026-07-05T08:15:53.244131Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:15:53.244131Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sub-goal Distillation: A Method to Improve Small Language Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Elias Stengel-Eskin, Marc-Alexandre Cote, Maryam Hashemzadeh, Sarath Chandar","submitted_at":"2024-05-04T20:34:06Z","abstract_excerpt":"While Large Language Models (LLMs) have demonstrated significant promise as agents in interactive tasks, their substantial computational requirements and restricted number of calls constrain their practical utility, especially in long-horizon interactive tasks such as decision-making or in scenarios involving continuous ongoing tasks. To address these constraints, we propose a method for transferring the performance of an LLM with billions of parameters to a much smaller language model (770M parameters). Our approach involves constructing a hierarchical agent comprising a planning module, whic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.02749","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/2405.02749/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":"2405.02749","created_at":"2026-07-05T08:15:53.244190+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.02749v1","created_at":"2026-07-05T08:15:53.244190+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.02749","created_at":"2026-07-05T08:15:53.244190+00:00"},{"alias_kind":"pith_short_12","alias_value":"X5FH5XIF7ZIT","created_at":"2026-07-05T08:15:53.244190+00:00"},{"alias_kind":"pith_short_16","alias_value":"X5FH5XIF7ZITC7SY","created_at":"2026-07-05T08:15:53.244190+00:00"},{"alias_kind":"pith_short_8","alias_value":"X5FH5XIF","created_at":"2026-07-05T08:15:53.244190+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.17873","citing_title":"HINT-SD: Targeted Hindsight Self-Distillation for Long-Horizon Agents","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX","json":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX.json","graph_json":"https://pith.science/api/pith-number/X5FH5XIF7ZITC7SYNOXKUQAVBX/graph.json","events_json":"https://pith.science/api/pith-number/X5FH5XIF7ZITC7SYNOXKUQAVBX/events.json","paper":"https://pith.science/paper/X5FH5XIF"},"agent_actions":{"view_html":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX","download_json":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX.json","view_paper":"https://pith.science/paper/X5FH5XIF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.02749&json=true","fetch_graph":"https://pith.science/api/pith-number/X5FH5XIF7ZITC7SYNOXKUQAVBX/graph.json","fetch_events":"https://pith.science/api/pith-number/X5FH5XIF7ZITC7SYNOXKUQAVBX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX/action/storage_attestation","attest_author":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX/action/author_attestation","sign_citation":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX/action/citation_signature","submit_replication":"https://pith.science/pith/X5FH5XIF7ZITC7SYNOXKUQAVBX/action/replication_record"}},"created_at":"2026-07-05T08:15:53.244190+00:00","updated_at":"2026-07-05T08:15:53.244190+00:00"}