{"_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_b832edb70220ad9313d7a8a79bc7de06e60daa8272a4ddab2ad1ff76b802e0dd","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_b832edb70220ad9313d7a8a79bc7de06e60daa8272a4ddab2ad1ff76b802e0dd","axiom_type":"AX.OBS","body":{"axiom_subtype":"news.vendor_press.v1","category":"news","fingerprint":"5cdf72df933ae5f3a70249621fb719db6b3b609e5bc42ac01726fc5f58098777","published":"Tue, 09 Jun 2026 00:00:00 -0400","receipt_hash":"5cdf72df933ae5f3a70249621fb719db6b3b609e5bc42ac01726fc5f58098777","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":"5cdf72df933ae5f3a70249621fb719db6b3b609e5bc42ac01726fc5f58098777","observed_at":"2026-06-09T04:43:45.619596Z","parent_run_hash":"f2344865fd128464efd1bacba326b5a7ccea707694b8c5650dd51ae8c46ac8a1","published":"Tue, 09 Jun 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:2511.11041v2 Announce Type: replace-cross \nAbstract: We find that current sentence-embedding models produce outputs with a consistent bias: every embedding $e$ decomposes as $\\tilde e + \\mu$, where the mean $\\mu$ is near-identical across all sentences. We study two training-free corrections -- subtracting $\\mu$ directly (R1), or projecting each embedding off the mean direction (R2) -- and show, via a first-order error-propagation argument, that R2 cancels the parallel component of mean-estimation error that R1 retains. Across 38 models on the Massive Multilingual Text Embedding Benchmark (MMTEB)~\\citep{MMTEB}, R2 yields consistent classification gains (paired $\\bar t = 3.31$, 29 of 38 models with $t>2$, zero losses), and the per-model mean norm $\\Vert\\mu\\Vert$ correlates with which models benefit most. A nine-method dose-response ablation on five models further reveals that mild single-direction removal helps, but full principal component analysis (PCA) whitening hurts every mode","title":"Correcting Mean Bias in Text Embeddings: A Refined Renormalization with Training-Free Improvements on MMTEB","url":"https://arxiv.org/abs/2511.11041","vendor":"arxiv_cs_ai"},"summary":"arXiv:2511.11041v2 Announce Type: replace-cross \nAbstract: We find that current sentence-embedding models produce outputs with a consistent bias: every embedding $e$ decomposes as $\\tilde e + \\mu$, where the mean $\\mu$ is near-identical across all sentences. We study two training-free corrections -- subtracting $\\mu$ directly (R1), or projecting each embedding off the mean direction (R2) -- and show, via a first-order error-propagation argument, that R2 cancels the parallel component of mean-estimation error that R1 retains. Across 38 models on the Massive Multilingual Text Embedding Benchmark (MMTEB)~\\citep{MMTEB}, R2 yields consistent classification gains (paired $\\bar t = 3.31$, 29 of 38 models with $t>2$, zero losses), and the per-model mean norm $\\Vert\\mu\\Vert$ correlates with which models benefit most. A nine-method dose-response ablation on five models further reveals that mild single-direction removal helps, but full principal component analysis (PCA) whitening hurts every mode","title":"Correcting Mean Bias in Text Embeddings: A Refined Renormalization with Training-Free Improvements on MMTEB","vendor":"arxiv_cs_ai"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-06-09T04:43:45Z","notes":"Spider vendor_press (news) news.vendor_press.v1","object":{"captured_by":"crovia.spider.vendor_press","primary_source_url":"https://arxiv.org/abs/2511.11041"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:7e9c127468f145b3aed3a5b91c0c1974c90c89dde456147cb6be505a0d6d391062310a5d54dcf49001848a8d3b2b2c3d56efd6372f39bb27618150f0b3107505","signer":"crovia.substrate","subject":{"observed_at":"2026-06-09T04:43:45Z","source_collector":"spider:vendor_press","target_id":"https://arxiv.org/abs/2511.11041"},"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":"b9032b401fdabefc4a1ee45d8472eb966f4731b3b61ce724dbe7a302f2188e28","leaf_index":224556,"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":"dd350389b94c0cbffa9ecf86f76b8c9a99cbec0e66fe272079a5d12c1c214160","side":"right"},{"sibling":"dcc6758a7585e0cecbe1c9d6a4b125a7930c51d53acf8b1701e0ec8a3cc3aadc","side":"right"},{"sibling":"46b5d8d36c704a4820292361790913495752f8792fe48223f65e78da977b6122","side":"left"},{"sibling":"de3817339030c0cadcef5ac48ee6eab102c18d750771c830e6e77fe52ed99b2f","side":"left"},{"sibling":"86115d2e02fb669f24d2259c12c5627939626a794cdf949f70cf6ff8d4461804","side":"right"},{"sibling":"8cb8d803d2ce30194ad0ac084483b71b61aef7a447a22d4937d90f5d6854b6a7","side":"left"},{"sibling":"7e5dbd67aa1e8898289b0ff2e3ec46622c5fc90def2622ec8f9fb42d76c33dae","side":"right"},{"sibling":"ce39e453038de7aa49f7571634d02890601b8ad6700079b4c6e4dc3e1d2e8b38","side":"right"},{"sibling":"f54580a307d4bb82e453931aa73e9a0c590486cb8df06af79ad4674f1c4f6963","side":"left"},{"sibling":"b2df6a4bb3e928f0b447931cc688ae01d2415773a2b07cfed0b1cba689078aed","side":"right"},{"sibling":"b1c9ec856caa0fd46bb47b46f18c59ebcd295d774ca17adb3b46f05d394a6a5d","side":"left"},{"sibling":"24fdc29d461691aedb6fa920758206b5bb43851f477ef7a04c34aaed84b8971b","side":"left"},{"sibling":"036922da4e1e2c46d948f070454bfad299b7406fb00735ea9d8bd1e687f5f445","side":"right"},{"sibling":"533d82482604463aa4a281b9d8b85917b383b7c5f924b2b494039524c55e8797","side":"left"},{"sibling":"b2590791b920ca2a4ed39de126d2c0b1a10d9e7e62f572f12425f214e767b6e1","side":"left"},{"sibling":"87c6b850dfec08ac35a693d9db3a3315250a68adb1cfab9b1015f212b63b15bd","side":"right"},{"sibling":"c300cf0154c136afc09b1702a0be98f4ba5b6dc5cf57e8cc714ec1eaf4196eff","side":"left"},{"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":224761,"merkle_root":"e9f7b49b652e869ab97ffba9c5a31356b2d0e3dc5d00bb28944adf737c46b1e7","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260609T103805Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-06-09T14:15:34Z","sig_algorithm":"ed25519","signature":"8ad8076fb12c8e486ae1d1559a9a7ba8e2ee996a9ad3d8ba7bcdbdbd88ab3a15bcb429707aca6d3e9d8b97e2ba755b3dcc77b1abb6601ccb829842719a6fb30d","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_b832edb70220ad9313d7a8a79bc7de06e60daa8272a4ddab2ad1ff76b802e0dd"}}