{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:XK6LRY4L4N5CUUHK3PSLCXQMVG","short_pith_number":"pith:XK6LRY4L","canonical_record":{"source":{"id":"2606.07414","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T16:06:47Z","cross_cats_sorted":["cs.NE"],"title_canon_sha256":"edc54cab4866055b2d8d5ea96a89747731617e56eb0bd628fd1dc9a78947c90a","abstract_canon_sha256":"a2d0b294086d9912581af324a67acdfaafae8735149350cd675e43f057a5234e"},"schema_version":"1.0"},"canonical_sha256":"babcb8e38be37a2a50eadbe4b15e0ca9a4158fe4390c0c8ce7e0bd57900526e8","source":{"kind":"arxiv","id":"2606.07414","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.07414","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"arxiv_version","alias_value":"2606.07414v1","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07414","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"pith_short_12","alias_value":"XK6LRY4L4N5C","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"pith_short_16","alias_value":"XK6LRY4L4N5CUUHK","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"pith_short_8","alias_value":"XK6LRY4L","created_at":"2026-06-08T01:05:25Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:XK6LRY4L4N5CUUHK3PSLCXQMVG","target":"record","payload":{"canonical_record":{"source":{"id":"2606.07414","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T16:06:47Z","cross_cats_sorted":["cs.NE"],"title_canon_sha256":"edc54cab4866055b2d8d5ea96a89747731617e56eb0bd628fd1dc9a78947c90a","abstract_canon_sha256":"a2d0b294086d9912581af324a67acdfaafae8735149350cd675e43f057a5234e"},"schema_version":"1.0"},"canonical_sha256":"babcb8e38be37a2a50eadbe4b15e0ca9a4158fe4390c0c8ce7e0bd57900526e8","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-08T01:05:25.646641Z","signature_b64":"06mWkpecdNVmoFO0+o20YXrW8p4FYX4zjQD34kEdbt+pNzwcikpxp53usZ9hWKV1fTryWTiffWMPiGGC2PQ+Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"babcb8e38be37a2a50eadbe4b15e0ca9a4158fe4390c0c8ce7e0bd57900526e8","last_reissued_at":"2026-06-08T01:05:25.645846Z","signature_status":"signed_v1","first_computed_at":"2026-06-08T01:05:25.645846Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.07414","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":"2026-06-08T01:05:25Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ed88aR6L+YtF4HOql35B+LAS1epEmVD2RjxI9QV+UPbqFe0xZFM8bvw537FAKP0LOs6xz4JbOBHiIF6vCJ5RAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-29T14:57:54.255066Z"},"content_sha256":"d8122d777266c0839e08d46681ea047ebfd3ba6d12fb880e5cf9aad686d35e0c","schema_version":"1.0","event_id":"sha256:d8122d777266c0839e08d46681ea047ebfd3ba6d12fb880e5cf9aad686d35e0c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:XK6LRY4L4N5CUUHK3PSLCXQMVG","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Sparsely gated tiny linear experts","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.NE"],"primary_cat":"cs.LG","authors_text":"Simon Schug","submitted_at":"2026-06-05T16:06:47Z","abstract_excerpt":"Sparsity allows scaling model parameters without proportionally increasing computational cost. While mixture of experts (MoE) models are made increasingly sparse, individual experts typically remain large and dense. Here, we demonstrate that further increasing sparsity by shrinking each expert to consist of a single neuron and selecting a tiny fraction of many available neurons can improve compute efficiency and interpretability. Counterintuitively, the key to achieving both is removing the nonlinearity typically applied to the experts, resulting in a network of sparsely gated linear neurons ("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07414","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/2606.07414/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":"2026-06-08T01:05:25Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"H3WrnHQSeg6MN5vUOUCnrSRf4ZY4dcTbB4hFZI7VdKNb15PYK7Hdu0igknAKaYKzjWagqbQTpyn4UFMvjGTUCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-29T14:57:54.255449Z"},"content_sha256":"5d770f415ac6497fde83a83f9586ede203c771a421825c47daad253a75daaef6","schema_version":"1.0","event_id":"sha256:5d770f415ac6497fde83a83f9586ede203c771a421825c47daad253a75daaef6"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XK6LRY4L4N5CUUHK3PSLCXQMVG/bundle.json","state_url":"https://pith.science/pith/XK6LRY4L4N5CUUHK3PSLCXQMVG/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XK6LRY4L4N5CUUHK3PSLCXQMVG/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":"2026-06-29T14:57:54Z","links":{"resolver":"https://pith.science/pith/XK6LRY4L4N5CUUHK3PSLCXQMVG","bundle":"https://pith.science/pith/XK6LRY4L4N5CUUHK3PSLCXQMVG/bundle.json","state":"https://pith.science/pith/XK6LRY4L4N5CUUHK3PSLCXQMVG/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XK6LRY4L4N5CUUHK3PSLCXQMVG/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:XK6LRY4L4N5CUUHK3PSLCXQMVG","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":"a2d0b294086d9912581af324a67acdfaafae8735149350cd675e43f057a5234e","cross_cats_sorted":["cs.NE"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T16:06:47Z","title_canon_sha256":"edc54cab4866055b2d8d5ea96a89747731617e56eb0bd628fd1dc9a78947c90a"},"schema_version":"1.0","source":{"id":"2606.07414","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.07414","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"arxiv_version","alias_value":"2606.07414v1","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07414","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"pith_short_12","alias_value":"XK6LRY4L4N5C","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"pith_short_16","alias_value":"XK6LRY4L4N5CUUHK","created_at":"2026-06-08T01:05:25Z"},{"alias_kind":"pith_short_8","alias_value":"XK6LRY4L","created_at":"2026-06-08T01:05:25Z"}],"graph_snapshots":[{"event_id":"sha256:5d770f415ac6497fde83a83f9586ede203c771a421825c47daad253a75daaef6","target":"graph","created_at":"2026-06-08T01:05:25Z","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/2606.07414/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Sparsity allows scaling model parameters without proportionally increasing computational cost. While mixture of experts (MoE) models are made increasingly sparse, individual experts typically remain large and dense. Here, we demonstrate that further increasing sparsity by shrinking each expert to consist of a single neuron and selecting a tiny fraction of many available neurons can improve compute efficiency and interpretability. Counterintuitively, the key to achieving both is removing the nonlinearity typically applied to the experts, resulting in a network of sparsely gated linear neurons (","authors_text":"Simon Schug","cross_cats":["cs.NE"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T16:06:47Z","title":"Sparsely gated tiny linear experts"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07414","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:d8122d777266c0839e08d46681ea047ebfd3ba6d12fb880e5cf9aad686d35e0c","target":"record","created_at":"2026-06-08T01:05:25Z","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":"a2d0b294086d9912581af324a67acdfaafae8735149350cd675e43f057a5234e","cross_cats_sorted":["cs.NE"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T16:06:47Z","title_canon_sha256":"edc54cab4866055b2d8d5ea96a89747731617e56eb0bd628fd1dc9a78947c90a"},"schema_version":"1.0","source":{"id":"2606.07414","kind":"arxiv","version":1}},"canonical_sha256":"babcb8e38be37a2a50eadbe4b15e0ca9a4158fe4390c0c8ce7e0bd57900526e8","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"babcb8e38be37a2a50eadbe4b15e0ca9a4158fe4390c0c8ce7e0bd57900526e8","first_computed_at":"2026-06-08T01:05:25.645846Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-08T01:05:25.645846Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"06mWkpecdNVmoFO0+o20YXrW8p4FYX4zjQD34kEdbt+pNzwcikpxp53usZ9hWKV1fTryWTiffWMPiGGC2PQ+Dw==","signature_status":"signed_v1","signed_at":"2026-06-08T01:05:25.646641Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.07414","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d8122d777266c0839e08d46681ea047ebfd3ba6d12fb880e5cf9aad686d35e0c","sha256:5d770f415ac6497fde83a83f9586ede203c771a421825c47daad253a75daaef6"],"state_sha256":"04a8790ea97e95c1b2050e4ea6e56048053a4ac8f505ff5af9a03eef9f18b9e1"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RuPNiaZ0kHaQQ4sT/JWOzfxM670zwZgyKkjuruFPsgdVg8kjY4BA8nQDhQ5UFwqRPqGV75u2YYfdNZnLUswMCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-29T14:57:54.257373Z","bundle_sha256":"cb460ad4f6bae6210934dae6933b5bcfadb3b2572d4c6872fce7ce01b7c883f1"}}