{"_canonicalization":{"envelope_id":"axm_ + sha256(envelope minus {signature, axiom_id, anchors})","envelope_signature":"ed25519(envelope minus {signature, axiom_id})","json":"sort_keys=True, separators=(',',':'), ensure_ascii=False, allow_nan=False, utf-8","leaf_hash":"sha256(0x00 || canonical_json(envelope_full))","seal_signature":"ed25519(seal minus {signature, sig_algorithm})"},"axiom_id":"axm_daa1bfe001ca0514bfb462cca7d4060ae7046c2fff9aaaa91dbedb8a536077fb","bitcoin_anchor":{"bitcoin_attestations":[],"calendar_attestations":[],"ots_url":"","stamped_at":"","status":"pending_next_stamp"},"envelope":{"anchors":[{"chain":"crovia.axiom_graph","height":0,"merkle_proof":"spider_vendor_press_v1","root_at_anchor":"spider_vendor_press_v1"}],"axiom_id":"axm_daa1bfe001ca0514bfb462cca7d4060ae7046c2fff9aaaa91dbedb8a536077fb","axiom_type":"AX.OBS","body":{"axiom_subtype":"news.vendor_press.v1","category":"news","fingerprint":"e6c13073513dd86e924287f0a79d48a7245be8ee685609f6417318d8bd3c34c5","published":"Fri, 29 May 2026 00:00:00 -0400","receipt_hash":"e6c13073513dd86e924287f0a79d48a7245be8ee685609f6417318d8bd3c34c5","schema":"spider.news.vendor_press.v1","spider":"vendor_press","spider_record":{"axiom_subtype":"news.vendor_press.v1","category":"news","decision_hint":"POSITIVE","envelope_target":"AX.OBS","fingerprint":"e6c13073513dd86e924287f0a79d48a7245be8ee685609f6417318d8bd3c34c5","observed_at":"2026-05-29T04:43:58.478092Z","parent_run_hash":"0fcd87efcfe67ccb9952f747541debc16793919a4d20fd71ca0ad5516a0a13ee","published":"Fri, 29 May 2026 00:00:00 -0400","runtime_version":"0.1.0","schema":"spider.news.vendor_press.v1","source_status":200,"source_url":"https://export.arxiv.org/rss/cs.AI","spider":"vendor_press","summary_excerpt":"arXiv:2605.30348v1 Announce Type: cross \nAbstract: The pretraining data mixture of Large Language Models (LLMs) constitutes their \"digital DNA\", shaping model behaviors, capabilities, and failure modes. Yet this composition is rarely disclosed, making post-hoc auditing of data combination or provenance difficult. In this work, we formalize $\\textbf{{Data Mixture Surgery (DMS)}}$: given only generated text from a target LLM, estimate the domain-level distribution of its pretraining corpus under a predefined taxonomy. We propose $\\textbf{{LLMSurgeon}}$, a strong framework that casts DMS as an inverse problem under the label-shift assumption. Rather than directly aggregating classifier outputs, LLMSurgeon estimates a calibrated $\\textit{soft}$ confusion matrix and solves a constrained inverse problem to correct systematic domain confusion and recover the latent mixture prior. To evaluate, we introduce $\\textbf{{LLMScan}}$, a recipe-verifiable evaluation suite built from open-source LLMs w","title":"LLMSurgeon: Diagnosing Data Mixture of Large Language Models","url":"https://arxiv.org/abs/2605.30348","vendor":"arxiv_cs_ai"},"summary":"arXiv:2605.30348v1 Announce Type: cross \nAbstract: The pretraining data mixture of Large Language Models (LLMs) constitutes their \"digital DNA\", shaping model behaviors, capabilities, and failure modes. Yet this composition is rarely disclosed, making post-hoc auditing of data combination or provenance difficult. In this work, we formalize $\\textbf{{Data Mixture Surgery (DMS)}}$: given only generated text from a target LLM, estimate the domain-level distribution of its pretraining corpus under a predefined taxonomy. We propose $\\textbf{{LLMSurgeon}}$, a strong framework that casts DMS as an inverse problem under the label-shift assumption. Rather than directly aggregating classifier outputs, LLMSurgeon estimates a calibrated $\\textit{soft}$ confusion matrix and solves a constrained inverse problem to correct systematic domain confusion and recover the latent mixture prior. To evaluate, we introduce $\\textbf{{LLMScan}}$, a recipe-verifiable evaluation suite built from open-source LLMs w","title":"LLMSurgeon: Diagnosing Data Mixture of Large Language Models","vendor":"arxiv_cs_ai"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-05-29T04:43:58Z","notes":"Spider vendor_press (news) news.vendor_press.v1","object":{"captured_by":"crovia.spider.vendor_press","primary_source_url":"https://arxiv.org/abs/2605.30348"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:fe6e8d8c2f55618857c2b105606a266515134b34370288bdb736f89253b23c3945154d14625dccd733d8b83c2316d5bfa1d86b9ad581005315d77103d69c8c05","signer":"crovia.substrate","subject":{"observed_at":"2026-05-29T04:43:58Z","source_collector":"spider:vendor_press","target_id":"https://arxiv.org/abs/2605.30348"},"tsa":{"authority":"crovia.substrate.bootstrap","rfc3161_token":"{\"kind\":\"crovia.bootstrap.tsa\",\"source_jsonl\":\"/opt/crovia/spider/data/news/vendor_press_v1.jsonl\",\"source_seal_merkle_root\":\"spider_vendor_press_v1\",\"upgrade_path\":\"Sessione H \\u2014 OpenTimestamps weekly anchor\"}"},"zk_mode":"clear","zk_proof":null},"ledger":{"leaf_hash":"77061b9e4f8af2aec3b186b64d9f08238cf6f53abe1e9a3a78193c528f5b8750","leaf_index":157996,"ledger_path":"/opt/crovia/substrate/axiom_ledger.jsonl"},"merkle_proof":{"hash_alg":"sha256","leaf_prefix":"0x00","node_prefix":"0x01","odd_leaf_rule":"duplicate_last","path":[{"sibling":"8d6461e4bbcb4b4df899febe35f5e041ab9c109224448379548624a9e50ffa0d","side":"right"},{"sibling":"7ea5327681ff46a6f063a2caeb6b0ea53ef5e971dae77c5cceb065a032e7bdd4","side":"right"},{"sibling":"496aab625051367b724cd10bacc81981bc346ca05e71475fa0db9c9f1c47e866","side":"left"},{"sibling":"04ced925b4d67fc085d4da650388a0e59f23acb16dac07b0a829124a311bde87","side":"left"},{"sibling":"7f6a339ed3296eb8c1b0e9df69bd9b6f37bd934b2752df85196f47a848973e5a","side":"right"},{"sibling":"fea7174f0a136a206dde2903ff85ee04d9ef78a16cc9ccf7baf52adfad38ec9a","side":"left"},{"sibling":"6ec9548fe4b05de3b418873e8f539457b0d358597426bbc33615ea29f980e57e","side":"right"},{"sibling":"7d396b40cab1f5921441ef7dd8b2f11230a90a7ab021a73b9ae8fa5ac45029fa","side":"right"},{"sibling":"002471ff5893cd2a2890119cbfe4a9456bfb46cf31fa6aee377650afe6ae0c92","side":"left"},{"sibling":"1dcc44e23fbb0218b13591e4b584eca3600dcf365769cb741e0ecd33b25b8c56","side":"right"},{"sibling":"fcf16a6f44025801f5b83e928acce764352ddbc06ae3043d8cde9a409933e6d8","side":"right"},{"sibling":"78982294dee68f9db9288c64d7e507c7865fda96e1b6f7ccff5c8bb152e93c49","side":"left"},{"sibling":"995b421824624a8282c7f44e64c64ee35344800f477ae1845b41be14d3fab94c","side":"right"},{"sibling":"66331bac84ca0f8983eb09fac7eaf95af234f1b82680b793eabff4ee25caac40","side":"left"},{"sibling":"35ca36cee447f0ef7064a25d55f59357c66901e31427729c4c1d8b14aa8adb6c","side":"left"},{"sibling":"eef0e8906a749d3470f89beeedc723f37a5737010bbb0dcc7cf91515338e5a3e","side":"right"},{"sibling":"1a07e481a9407d71aad078ce854cdeee362163c887fe10f889b0ecf0b5e749ad","side":"right"},{"sibling":"d841ad93efda0869e5eb97678f348f03f5caab4353e05ff4bf18f47fb945b822","side":"left"}]},"schema":"crovia.axiom_proof.v1","seal":{"first_collector_run_id":"","first_receipt_hash":"","jsonl_path":"/opt/crovia/substrate/axiom_ledger.jsonl","key_id":"430895f101d38164","last_collector_run_id":"","last_receipt_hash":"","leaf_count":158251,"merkle_root":"485e6b31fe60c8beba5b394808c7e4c32448b2ff65c2482c480ca0e2a2eda718","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260529T053701Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-05-29T05:37:37Z","sig_algorithm":"ed25519","signature":"bbf9f005201182fce4f9d94c7a9d01a508b56daf7d9611bd73514f5f616bc059d0e5e1f2edfc95716e6fe08ef5fae38b558cbf7f2fd8f9d5dfe4c34a54c83005","signer_version":"1.1.0"},"trust_root":{"key_id":"430895f101d38164","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","signature_algorithm":"ed25519","url":"/registry/canon/TRUST_ROOT.md"},"verifier":{"spec":"/registry/canon/AXIOM_RECEIPT_v1.md","url":"/v/axm_daa1bfe001ca0514bfb462cca7d4060ae7046c2fff9aaaa91dbedb8a536077fb"}}