{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:U5SKOZYE3IYX3CMXILH7IMIEE3","short_pith_number":"pith:U5SKOZYE","canonical_record":{"source":{"id":"2507.11371","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-15T14:44:29Z","cross_cats_sorted":["cs.MA"],"title_canon_sha256":"9339e4fbc71a6c409d4a67c3f8eddecd191cc33e3e9ecb879c6e4db0c74e3738","abstract_canon_sha256":"42daa41b113aa63da792fbd193597ba534a53a007c4184114d64e9afa09c3fdb"},"schema_version":"1.0"},"canonical_sha256":"a764a76704da317d899742cff4310426fa4acf93edb2efc65f12698670b03ea5","source":{"kind":"arxiv","id":"2507.11371","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.11371","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"arxiv_version","alias_value":"2507.11371v1","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.11371","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"pith_short_12","alias_value":"U5SKOZYE3IYX","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"pith_short_16","alias_value":"U5SKOZYE3IYX3CMX","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"pith_short_8","alias_value":"U5SKOZYE","created_at":"2026-07-05T11:37:38Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:U5SKOZYE3IYX3CMXILH7IMIEE3","target":"record","payload":{"canonical_record":{"source":{"id":"2507.11371","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-15T14:44:29Z","cross_cats_sorted":["cs.MA"],"title_canon_sha256":"9339e4fbc71a6c409d4a67c3f8eddecd191cc33e3e9ecb879c6e4db0c74e3738","abstract_canon_sha256":"42daa41b113aa63da792fbd193597ba534a53a007c4184114d64e9afa09c3fdb"},"schema_version":"1.0"},"canonical_sha256":"a764a76704da317d899742cff4310426fa4acf93edb2efc65f12698670b03ea5","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:37:38.018017Z","signature_b64":"PbJKBeLsPHiakaOBtK11hNkvbV158OnobvDRSkRM/xNWh7fRjQDuDyKr5N295y7zGvpyS9azYWGudNsvvCY1DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a764a76704da317d899742cff4310426fa4acf93edb2efc65f12698670b03ea5","last_reissued_at":"2026-07-05T11:37:38.017485Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:37:38.017485Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2507.11371","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:37:38Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"39lOGkpaLbeLhzEJsaB9F44i4b/pmbU97ZifF0Xt2ny1oNmn1r0XlnvK3vupiZ7N8DigeGix46lLn8qCXCYMBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T13:05:09.283147Z"},"content_sha256":"c78da24790969509e93c82281a9ed962104e83b6191b7adf93249a3fac3d615e","schema_version":"1.0","event_id":"sha256:c78da24790969509e93c82281a9ed962104e83b6191b7adf93249a3fac3d615e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:U5SKOZYE3IYX3CMXILH7IMIEE3","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Step-wise Policy for Rare-tool Knowledge (SPaRK): Offline RL that Drives Diverse Tool Use in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.MA"],"primary_cat":"cs.LG","authors_text":"Gabriel Bo, Justin Gu, Koa Chang","submitted_at":"2025-07-15T14:44:29Z","abstract_excerpt":"We present Step-wise Policy for Rare-tool Knowledge (SPaRK), a novel reinforcement learning framework that teaches large language models to explore diverse tool usage patterns beyond conventional high-temperature sampling. Building on recent advances in step-wise reinforcement learning, we introduce a dual-objective reward system that simultaneously optimizes for answer quality and tool diversity, training a Llama-3.1 8B model through offline PPO on synthetically generated trajectories from the MMLU-Pro dataset. Our approach uniquely employs a rarity-first exploitation strategy where a GPT-4o "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.11371","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/2507.11371/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:37:38Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QTmQFtsjAJsPh0hyp2B7gu/ZVNX3G3VEqV4sPqleVMIsibyL9iw4Cf9wp0P3mwnibJOZUSLkdBYAUswactMbDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T13:05:09.284117Z"},"content_sha256":"e58444f6d719692ae43d2d1b230b6a3849d7dd58d21c61296090652a6ad68bd9","schema_version":"1.0","event_id":"sha256:e58444f6d719692ae43d2d1b230b6a3849d7dd58d21c61296090652a6ad68bd9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/U5SKOZYE3IYX3CMXILH7IMIEE3/bundle.json","state_url":"https://pith.science/pith/U5SKOZYE3IYX3CMXILH7IMIEE3/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/U5SKOZYE3IYX3CMXILH7IMIEE3/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-10T13:05:09Z","links":{"resolver":"https://pith.science/pith/U5SKOZYE3IYX3CMXILH7IMIEE3","bundle":"https://pith.science/pith/U5SKOZYE3IYX3CMXILH7IMIEE3/bundle.json","state":"https://pith.science/pith/U5SKOZYE3IYX3CMXILH7IMIEE3/state.json","well_known_bundle":"https://pith.science/.well-known/pith/U5SKOZYE3IYX3CMXILH7IMIEE3/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:U5SKOZYE3IYX3CMXILH7IMIEE3","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"42daa41b113aa63da792fbd193597ba534a53a007c4184114d64e9afa09c3fdb","cross_cats_sorted":["cs.MA"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-15T14:44:29Z","title_canon_sha256":"9339e4fbc71a6c409d4a67c3f8eddecd191cc33e3e9ecb879c6e4db0c74e3738"},"schema_version":"1.0","source":{"id":"2507.11371","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.11371","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"arxiv_version","alias_value":"2507.11371v1","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.11371","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"pith_short_12","alias_value":"U5SKOZYE3IYX","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"pith_short_16","alias_value":"U5SKOZYE3IYX3CMX","created_at":"2026-07-05T11:37:38Z"},{"alias_kind":"pith_short_8","alias_value":"U5SKOZYE","created_at":"2026-07-05T11:37:38Z"}],"graph_snapshots":[{"event_id":"sha256:e58444f6d719692ae43d2d1b230b6a3849d7dd58d21c61296090652a6ad68bd9","target":"graph","created_at":"2026-07-05T11:37:38Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2507.11371/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We present Step-wise Policy for Rare-tool Knowledge (SPaRK), a novel reinforcement learning framework that teaches large language models to explore diverse tool usage patterns beyond conventional high-temperature sampling. Building on recent advances in step-wise reinforcement learning, we introduce a dual-objective reward system that simultaneously optimizes for answer quality and tool diversity, training a Llama-3.1 8B model through offline PPO on synthetically generated trajectories from the MMLU-Pro dataset. Our approach uniquely employs a rarity-first exploitation strategy where a GPT-4o ","authors_text":"Gabriel Bo, Justin Gu, Koa Chang","cross_cats":["cs.MA"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-15T14:44:29Z","title":"Step-wise Policy for Rare-tool Knowledge (SPaRK): Offline RL that Drives Diverse Tool Use in LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.11371","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c78da24790969509e93c82281a9ed962104e83b6191b7adf93249a3fac3d615e","target":"record","created_at":"2026-07-05T11:37:38Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"42daa41b113aa63da792fbd193597ba534a53a007c4184114d64e9afa09c3fdb","cross_cats_sorted":["cs.MA"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-15T14:44:29Z","title_canon_sha256":"9339e4fbc71a6c409d4a67c3f8eddecd191cc33e3e9ecb879c6e4db0c74e3738"},"schema_version":"1.0","source":{"id":"2507.11371","kind":"arxiv","version":1}},"canonical_sha256":"a764a76704da317d899742cff4310426fa4acf93edb2efc65f12698670b03ea5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a764a76704da317d899742cff4310426fa4acf93edb2efc65f12698670b03ea5","first_computed_at":"2026-07-05T11:37:38.017485Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:37:38.017485Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"PbJKBeLsPHiakaOBtK11hNkvbV158OnobvDRSkRM/xNWh7fRjQDuDyKr5N295y7zGvpyS9azYWGudNsvvCY1DA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:37:38.018017Z","signed_message":"canonical_sha256_bytes"},"source_id":"2507.11371","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c78da24790969509e93c82281a9ed962104e83b6191b7adf93249a3fac3d615e","sha256:e58444f6d719692ae43d2d1b230b6a3849d7dd58d21c61296090652a6ad68bd9"],"state_sha256":"a20798ac2772627ef05b32abfba5bceb2de11f8f251137c9c2a45a3c0afcf61c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lQplk1IILckdt2PRqqCK1/EFNIPzvzLfNKOgCr2akl7g7tLHuET82MVIlmjcD08kuMVMTQN7HZUdKCO231Z4Dg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-10T13:05:09.291244Z","bundle_sha256":"40dfcbbc316c75de191051fa8ee555ad407b85e575dab31ec0b52ace7289b3f4"}}