{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:M3ESM2RLMWSWR45HCQJAYRWYFX","short_pith_number":"pith:M3ESM2RL","canonical_record":{"source":{"id":"2508.04096","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.SD","submitted_at":"2025-08-06T05:28:18Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"173fe38c0966606d83a3fb7b589caece25b4e4eb991acb90cd9f31293eaf5595","abstract_canon_sha256":"d9ea12ba6af3e621a1d1ccf5076e4d45bc3ffe69c0713efdce29cc77c7cc1ddc"},"schema_version":"1.0"},"canonical_sha256":"66c9266a2b65a568f3a714120c46d82dd0fa92003220de657bea1097988b5d59","source":{"kind":"arxiv","id":"2508.04096","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.04096","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"arxiv_version","alias_value":"2508.04096v1","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.04096","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"pith_short_12","alias_value":"M3ESM2RLMWSW","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"pith_short_16","alias_value":"M3ESM2RLMWSWR45H","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"pith_short_8","alias_value":"M3ESM2RL","created_at":"2026-07-05T11:49:20Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:M3ESM2RLMWSWR45HCQJAYRWYFX","target":"record","payload":{"canonical_record":{"source":{"id":"2508.04096","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.SD","submitted_at":"2025-08-06T05:28:18Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"173fe38c0966606d83a3fb7b589caece25b4e4eb991acb90cd9f31293eaf5595","abstract_canon_sha256":"d9ea12ba6af3e621a1d1ccf5076e4d45bc3ffe69c0713efdce29cc77c7cc1ddc"},"schema_version":"1.0"},"canonical_sha256":"66c9266a2b65a568f3a714120c46d82dd0fa92003220de657bea1097988b5d59","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:49:20.532049Z","signature_b64":"c0Ydc88zM8t/iVaPzvTufl0xwJnAHuoM4GquYK3TiV8wiJHuwIOtZln1+jiUq6dFpij5LyAXZjQUc1zpqHeLBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"66c9266a2b65a568f3a714120c46d82dd0fa92003220de657bea1097988b5d59","last_reissued_at":"2026-07-05T11:49:20.531534Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:49:20.531534Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2508.04096","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":"2026-07-05T11:49:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LAbio+vNxmma7DTSWwxMiYFtFVXUqcrKzFywYmdLGaK/Rl7QtmN0BIw3G5TbftO4jsOy1u6lZpcHPf9d9eaUDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T07:54:56.058751Z"},"content_sha256":"164e0d1a38c43826242811419bc9beae3b061628af310ca711f9bfb3b1a3d7bb","schema_version":"1.0","event_id":"sha256:164e0d1a38c43826242811419bc9beae3b061628af310ca711f9bfb3b1a3d7bb"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:M3ESM2RLMWSWR45HCQJAYRWYFX","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Efficient Scaling for LLM-based ASR","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.SD","authors_text":"Bingshen Mu, Dong Yu, Kun Wei, Lei Xie, Yiwen Shao","submitted_at":"2025-08-06T05:28:18Z","abstract_excerpt":"Large language model (LLM)-based automatic speech recognition (ASR) achieves strong performance but often incurs high computational costs. This work investigates how to obtain the best LLM-ASR performance efficiently. Through comprehensive and controlled experiments, we find that pretraining the speech encoder before integrating it with the LLM leads to significantly better scaling efficiency than the standard practice of joint post-training of LLM-ASR. Based on this insight, we propose a new multi-stage LLM-ASR training strategy, EFIN: Encoder First Integration. Among all training strategies "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.04096","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/2508.04096/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":"2026-07-05T11:49:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"XhwGVvCpCch93aox2Qo/EKE45cMJAqAtomAMlz8hjxnB7aTpuk2BhiO1zwhe1TRwNP7mJVLO7c/fOdHihNyJBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T07:54:56.059236Z"},"content_sha256":"9a53a912ed675f6a4af9e01f7868ab8606b6ca8e0eed0038407956b8c5332435","schema_version":"1.0","event_id":"sha256:9a53a912ed675f6a4af9e01f7868ab8606b6ca8e0eed0038407956b8c5332435"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/M3ESM2RLMWSWR45HCQJAYRWYFX/bundle.json","state_url":"https://pith.science/pith/M3ESM2RLMWSWR45HCQJAYRWYFX/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/M3ESM2RLMWSWR45HCQJAYRWYFX/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":"2026-08-04T07:54:56Z","links":{"resolver":"https://pith.science/pith/M3ESM2RLMWSWR45HCQJAYRWYFX","bundle":"https://pith.science/pith/M3ESM2RLMWSWR45HCQJAYRWYFX/bundle.json","state":"https://pith.science/pith/M3ESM2RLMWSWR45HCQJAYRWYFX/state.json","well_known_bundle":"https://pith.science/.well-known/pith/M3ESM2RLMWSWR45HCQJAYRWYFX/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:M3ESM2RLMWSWR45HCQJAYRWYFX","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":"d9ea12ba6af3e621a1d1ccf5076e4d45bc3ffe69c0713efdce29cc77c7cc1ddc","cross_cats_sorted":["eess.AS"],"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.SD","submitted_at":"2025-08-06T05:28:18Z","title_canon_sha256":"173fe38c0966606d83a3fb7b589caece25b4e4eb991acb90cd9f31293eaf5595"},"schema_version":"1.0","source":{"id":"2508.04096","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.04096","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"arxiv_version","alias_value":"2508.04096v1","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.04096","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"pith_short_12","alias_value":"M3ESM2RLMWSW","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"pith_short_16","alias_value":"M3ESM2RLMWSWR45H","created_at":"2026-07-05T11:49:20Z"},{"alias_kind":"pith_short_8","alias_value":"M3ESM2RL","created_at":"2026-07-05T11:49:20Z"}],"graph_snapshots":[{"event_id":"sha256:9a53a912ed675f6a4af9e01f7868ab8606b6ca8e0eed0038407956b8c5332435","target":"graph","created_at":"2026-07-05T11:49:20Z","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/2508.04096/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language model (LLM)-based automatic speech recognition (ASR) achieves strong performance but often incurs high computational costs. This work investigates how to obtain the best LLM-ASR performance efficiently. Through comprehensive and controlled experiments, we find that pretraining the speech encoder before integrating it with the LLM leads to significantly better scaling efficiency than the standard practice of joint post-training of LLM-ASR. Based on this insight, we propose a new multi-stage LLM-ASR training strategy, EFIN: Encoder First Integration. Among all training strategies ","authors_text":"Bingshen Mu, Dong Yu, Kun Wei, Lei Xie, Yiwen Shao","cross_cats":["eess.AS"],"headline":"","license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.SD","submitted_at":"2025-08-06T05:28:18Z","title":"Efficient Scaling for LLM-based ASR"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.04096","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":"sha256:164e0d1a38c43826242811419bc9beae3b061628af310ca711f9bfb3b1a3d7bb","target":"record","created_at":"2026-07-05T11:49:20Z","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":"d9ea12ba6af3e621a1d1ccf5076e4d45bc3ffe69c0713efdce29cc77c7cc1ddc","cross_cats_sorted":["eess.AS"],"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.SD","submitted_at":"2025-08-06T05:28:18Z","title_canon_sha256":"173fe38c0966606d83a3fb7b589caece25b4e4eb991acb90cd9f31293eaf5595"},"schema_version":"1.0","source":{"id":"2508.04096","kind":"arxiv","version":1}},"canonical_sha256":"66c9266a2b65a568f3a714120c46d82dd0fa92003220de657bea1097988b5d59","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"66c9266a2b65a568f3a714120c46d82dd0fa92003220de657bea1097988b5d59","first_computed_at":"2026-07-05T11:49:20.531534Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:49:20.531534Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"c0Ydc88zM8t/iVaPzvTufl0xwJnAHuoM4GquYK3TiV8wiJHuwIOtZln1+jiUq6dFpij5LyAXZjQUc1zpqHeLBg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:49:20.532049Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.04096","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:164e0d1a38c43826242811419bc9beae3b061628af310ca711f9bfb3b1a3d7bb","sha256:9a53a912ed675f6a4af9e01f7868ab8606b6ca8e0eed0038407956b8c5332435"],"state_sha256":"5735ce05711e5ec67451304b2614c262bb4f3edee2720426d8ee822d716b8b51"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"KqB9/N+RLzYua1ufTSNSwjNxcEBg0Fr1D1U6BqhghcgvCCQ6BePupK3Rm36yrj/Xnsbtu+LppKKn3haxXIOUDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T07:54:56.063232Z","bundle_sha256":"d8d68cb111e3982d42240ef2dde460f15477fd465eb3ca10c87c7c29a4e848b2"}}