{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:Q6O45TFZHJOJMTTE2U3XZPOT2Y","short_pith_number":"pith:Q6O45TFZ","canonical_record":{"source":{"id":"2505.14136","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-20T09:39:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"30c5a2dff7ffd56a729da506bc36fc842a43f846f574a77bdf1f3325c1dfdb48","abstract_canon_sha256":"17c303555ccfa3f9923d62a1f7eb6cc6070f86b06986eb4fe25bcb6400de6e56"},"schema_version":"1.0"},"canonical_sha256":"879dceccb93a5c964e64d5377cbdd3d604a63b9492eca36b415f29ee7b66c594","source":{"kind":"arxiv","id":"2505.14136","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.14136","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"arxiv_version","alias_value":"2505.14136v2","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.14136","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"pith_short_12","alias_value":"Q6O45TFZHJOJ","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"pith_short_16","alias_value":"Q6O45TFZHJOJMTTE","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"pith_short_8","alias_value":"Q6O45TFZ","created_at":"2026-07-05T11:45:35Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:Q6O45TFZHJOJMTTE2U3XZPOT2Y","target":"record","payload":{"canonical_record":{"source":{"id":"2505.14136","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-20T09:39:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"30c5a2dff7ffd56a729da506bc36fc842a43f846f574a77bdf1f3325c1dfdb48","abstract_canon_sha256":"17c303555ccfa3f9923d62a1f7eb6cc6070f86b06986eb4fe25bcb6400de6e56"},"schema_version":"1.0"},"canonical_sha256":"879dceccb93a5c964e64d5377cbdd3d604a63b9492eca36b415f29ee7b66c594","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:45:35.798942Z","signature_b64":"LT7FCJQpB9u2EgkdhYdVThPXAWwDMdf0+QIxwTr6cG2RlDR70TiBQolaaJ9cpbF7dDbSjas5y5DNlgtvIak8CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"879dceccb93a5c964e64d5377cbdd3d604a63b9492eca36b415f29ee7b66c594","last_reissued_at":"2026-07-05T11:45:35.798423Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:45:35.798423Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2505.14136","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:45:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+FHRCS4kbxo7BxIBLm0fg9mV2KfoRX9Hm7o9NktbmHM2wJWO9KcbcjJnWYriYL/vZjwO+3uvj9XRtznXLIY0Ag==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T18:24:45.374671Z"},"content_sha256":"47343c98323b217fde9a21b49a00c1b5be11fc4859018373d1b141ad070d40f3","schema_version":"1.0","event_id":"sha256:47343c98323b217fde9a21b49a00c1b5be11fc4859018373d1b141ad070d40f3"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:Q6O45TFZHJOJMTTE2U3XZPOT2Y","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Local Mixtures of Experts: Essentially Free Test-Time Training via Model Merging","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andreas Krause, Ido Hakimi, Jonas H\\\"ubotter, Ryo Bertolissi","submitted_at":"2025-05-20T09:39:54Z","abstract_excerpt":"Mixture of expert (MoE) models are a promising approach to increasing model capacity without increasing inference cost, and are core components of many state-of-the-art language models. However, current MoE models typically use only few experts due to prohibitive training and inference cost. We propose Test-Time Model Merging (TTMM) which scales the MoE paradigm to an order of magnitude more experts and uses model merging to avoid almost any test-time overhead. We show that TTMM is an approximation of test-time training (TTT), which fine-tunes an expert model for each prediction task, i.e., pr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.14136","kind":"arxiv","version":2},"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/2505.14136/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":"2026-07-05T11:45:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6b5KM3tiNzGqLfONciV3+ghvP30tVSoMcOUVPg6/gGled35vBmprvhcAJdPFVlMudXUvXkZO//Mipqp7hPr7BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T18:24:45.375538Z"},"content_sha256":"e2b0c3d65d6d87698094b340eed3a5534fa04870dd2cf06f90b293d43d8e3d6c","schema_version":"1.0","event_id":"sha256:e2b0c3d65d6d87698094b340eed3a5534fa04870dd2cf06f90b293d43d8e3d6c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/Q6O45TFZHJOJMTTE2U3XZPOT2Y/bundle.json","state_url":"https://pith.science/pith/Q6O45TFZHJOJMTTE2U3XZPOT2Y/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/Q6O45TFZHJOJMTTE2U3XZPOT2Y/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":"2026-08-08T18:24:45Z","links":{"resolver":"https://pith.science/pith/Q6O45TFZHJOJMTTE2U3XZPOT2Y","bundle":"https://pith.science/pith/Q6O45TFZHJOJMTTE2U3XZPOT2Y/bundle.json","state":"https://pith.science/pith/Q6O45TFZHJOJMTTE2U3XZPOT2Y/state.json","well_known_bundle":"https://pith.science/.well-known/pith/Q6O45TFZHJOJMTTE2U3XZPOT2Y/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:Q6O45TFZHJOJMTTE2U3XZPOT2Y","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":"17c303555ccfa3f9923d62a1f7eb6cc6070f86b06986eb4fe25bcb6400de6e56","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-20T09:39:54Z","title_canon_sha256":"30c5a2dff7ffd56a729da506bc36fc842a43f846f574a77bdf1f3325c1dfdb48"},"schema_version":"1.0","source":{"id":"2505.14136","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.14136","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"arxiv_version","alias_value":"2505.14136v2","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.14136","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"pith_short_12","alias_value":"Q6O45TFZHJOJ","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"pith_short_16","alias_value":"Q6O45TFZHJOJMTTE","created_at":"2026-07-05T11:45:35Z"},{"alias_kind":"pith_short_8","alias_value":"Q6O45TFZ","created_at":"2026-07-05T11:45:35Z"}],"graph_snapshots":[{"event_id":"sha256:e2b0c3d65d6d87698094b340eed3a5534fa04870dd2cf06f90b293d43d8e3d6c","target":"graph","created_at":"2026-07-05T11:45:35Z","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/2505.14136/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Mixture of expert (MoE) models are a promising approach to increasing model capacity without increasing inference cost, and are core components of many state-of-the-art language models. However, current MoE models typically use only few experts due to prohibitive training and inference cost. We propose Test-Time Model Merging (TTMM) which scales the MoE paradigm to an order of magnitude more experts and uses model merging to avoid almost any test-time overhead. We show that TTMM is an approximation of test-time training (TTT), which fine-tunes an expert model for each prediction task, i.e., pr","authors_text":"Andreas Krause, Ido Hakimi, Jonas H\\\"ubotter, Ryo Bertolissi","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-20T09:39:54Z","title":"Local Mixtures of Experts: Essentially Free Test-Time Training via Model Merging"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.14136","kind":"arxiv","version":2},"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":"sha256:47343c98323b217fde9a21b49a00c1b5be11fc4859018373d1b141ad070d40f3","target":"record","created_at":"2026-07-05T11:45:35Z","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":"17c303555ccfa3f9923d62a1f7eb6cc6070f86b06986eb4fe25bcb6400de6e56","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-20T09:39:54Z","title_canon_sha256":"30c5a2dff7ffd56a729da506bc36fc842a43f846f574a77bdf1f3325c1dfdb48"},"schema_version":"1.0","source":{"id":"2505.14136","kind":"arxiv","version":2}},"canonical_sha256":"879dceccb93a5c964e64d5377cbdd3d604a63b9492eca36b415f29ee7b66c594","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"879dceccb93a5c964e64d5377cbdd3d604a63b9492eca36b415f29ee7b66c594","first_computed_at":"2026-07-05T11:45:35.798423Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:45:35.798423Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"LT7FCJQpB9u2EgkdhYdVThPXAWwDMdf0+QIxwTr6cG2RlDR70TiBQolaaJ9cpbF7dDbSjas5y5DNlgtvIak8CQ==","signature_status":"signed_v1","signed_at":"2026-07-05T11:45:35.798942Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.14136","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:47343c98323b217fde9a21b49a00c1b5be11fc4859018373d1b141ad070d40f3","sha256:e2b0c3d65d6d87698094b340eed3a5534fa04870dd2cf06f90b293d43d8e3d6c"],"state_sha256":"7b2c2e90bee371e004f800575d729f9993bbca64b87dfe4340f8fa47143e2728"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5mPq3XhNFsdyqc1rSDqbaPL4mWI8OCb7tRr6wjNyLGRknFU871Hvj2zRZYnDfuHD3fQbeum/oZ8t6s4QmK+tCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T18:24:45.381040Z","bundle_sha256":"55fa30de4f99898d9804a208d65d35b7263f9b81335e208dc74944a362ec65bb"}}