{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:ITDU443EZTKEEQXYXAB6P6SOSV","short_pith_number":"pith:ITDU443E","canonical_record":{"source":{"id":"2405.03133","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-06T03:06:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"429ebda2b916bb8347df3134470a6609d24d2b93d30537ee2bbfa4fb11377471","abstract_canon_sha256":"7249614c3e7efddb6eb32bb9367843141d8816248c7139e9a045fcb7768b9fb5"},"schema_version":"1.0"},"canonical_sha256":"44c74e7364ccd44242f8b803e7fa4e954462343e7f7daaac84c01b3f2bee2db0","source":{"kind":"arxiv","id":"2405.03133","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.03133","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"arxiv_version","alias_value":"2405.03133v2","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.03133","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"pith_short_12","alias_value":"ITDU443EZTKE","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"pith_short_16","alias_value":"ITDU443EZTKEEQXY","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"pith_short_8","alias_value":"ITDU443E","created_at":"2026-07-05T08:56:19Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:ITDU443EZTKEEQXYXAB6P6SOSV","target":"record","payload":{"canonical_record":{"source":{"id":"2405.03133","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-06T03:06:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"429ebda2b916bb8347df3134470a6609d24d2b93d30537ee2bbfa4fb11377471","abstract_canon_sha256":"7249614c3e7efddb6eb32bb9367843141d8816248c7139e9a045fcb7768b9fb5"},"schema_version":"1.0"},"canonical_sha256":"44c74e7364ccd44242f8b803e7fa4e954462343e7f7daaac84c01b3f2bee2db0","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:56:19.511284Z","signature_b64":"GjMHx6DdZXIU2BeaUy8ExctC3G3dKP39d+6QLDkv3r/3lUkQowP8CbrAv/N7kN7J/vHKDdNcQTr2SedN+8R+Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"44c74e7364ccd44242f8b803e7fa4e954462343e7f7daaac84c01b3f2bee2db0","last_reissued_at":"2026-07-05T08:56:19.510836Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:56:19.510836Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.03133","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":"2026-07-05T08:56:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FGDtPbEPVhkmPgKcMbaSEGRswgjHdj9lvY2EkbOAK25KBeX2RlqN0oRQo7g+nYUnO0V/+KSwXJzLRD2szu57DA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T09:41:53.030446Z"},"content_sha256":"6fc29cfbbc32205953ce8450bfd6c880f48ab8cd97a471a1e953602fc05034c0","schema_version":"1.0","event_id":"sha256:6fc29cfbbc32205953ce8450bfd6c880f48ab8cd97a471a1e953602fc05034c0"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:ITDU443EZTKEEQXYXAB6P6SOSV","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Lory: Fully Differentiable Mixture-of-Experts for Autoregressive Language Model Pre-training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Danqi Chen, Mengzhou Xia, Mike Lewis, Zexuan Zhong","submitted_at":"2024-05-06T03:06:33Z","abstract_excerpt":"Mixture-of-experts (MoE) models facilitate efficient scaling; however, training the router network introduces the challenge of optimizing a non-differentiable, discrete objective. Recently, a fully-differentiable MoE architecture, SMEAR, was proposed (Muqeeth et al., 2023), which softly merges experts in the parameter space; nevertheless, its effectiveness was only demonstrated in downstream fine-tuning on classification tasks. In this paper, we present Lory, the first approach that scales such architectures to autoregressive language model pre-training. Lory introduces two key techniques: (1)"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.03133","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/2405.03133/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":"2026-07-05T08:56:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"kq2LvwDivQbD54vu8j1hwRYsDDsyb3BAEpp251cSAU0wf+2+PNIxF1N5qfDo56gN3oX5VcolZRuWp1jvwFeRAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T09:41:53.030953Z"},"content_sha256":"3a53a230878476c922505cba1fa06ac33ab3651b2fe8326e36d72ec2198e8162","schema_version":"1.0","event_id":"sha256:3a53a230878476c922505cba1fa06ac33ab3651b2fe8326e36d72ec2198e8162"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ITDU443EZTKEEQXYXAB6P6SOSV/bundle.json","state_url":"https://pith.science/pith/ITDU443EZTKEEQXYXAB6P6SOSV/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ITDU443EZTKEEQXYXAB6P6SOSV/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":"2026-08-08T09:41:53Z","links":{"resolver":"https://pith.science/pith/ITDU443EZTKEEQXYXAB6P6SOSV","bundle":"https://pith.science/pith/ITDU443EZTKEEQXYXAB6P6SOSV/bundle.json","state":"https://pith.science/pith/ITDU443EZTKEEQXYXAB6P6SOSV/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ITDU443EZTKEEQXYXAB6P6SOSV/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:ITDU443EZTKEEQXYXAB6P6SOSV","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":"7249614c3e7efddb6eb32bb9367843141d8816248c7139e9a045fcb7768b9fb5","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-06T03:06:33Z","title_canon_sha256":"429ebda2b916bb8347df3134470a6609d24d2b93d30537ee2bbfa4fb11377471"},"schema_version":"1.0","source":{"id":"2405.03133","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.03133","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"arxiv_version","alias_value":"2405.03133v2","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.03133","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"pith_short_12","alias_value":"ITDU443EZTKE","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"pith_short_16","alias_value":"ITDU443EZTKEEQXY","created_at":"2026-07-05T08:56:19Z"},{"alias_kind":"pith_short_8","alias_value":"ITDU443E","created_at":"2026-07-05T08:56:19Z"}],"graph_snapshots":[{"event_id":"sha256:3a53a230878476c922505cba1fa06ac33ab3651b2fe8326e36d72ec2198e8162","target":"graph","created_at":"2026-07-05T08:56:19Z","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/2405.03133/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Mixture-of-experts (MoE) models facilitate efficient scaling; however, training the router network introduces the challenge of optimizing a non-differentiable, discrete objective. Recently, a fully-differentiable MoE architecture, SMEAR, was proposed (Muqeeth et al., 2023), which softly merges experts in the parameter space; nevertheless, its effectiveness was only demonstrated in downstream fine-tuning on classification tasks. In this paper, we present Lory, the first approach that scales such architectures to autoregressive language model pre-training. Lory introduces two key techniques: (1)","authors_text":"Danqi Chen, Mengzhou Xia, Mike Lewis, Zexuan Zhong","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-06T03:06:33Z","title":"Lory: Fully Differentiable Mixture-of-Experts for Autoregressive Language Model Pre-training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.03133","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":"sha256:6fc29cfbbc32205953ce8450bfd6c880f48ab8cd97a471a1e953602fc05034c0","target":"record","created_at":"2026-07-05T08:56:19Z","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":"7249614c3e7efddb6eb32bb9367843141d8816248c7139e9a045fcb7768b9fb5","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-06T03:06:33Z","title_canon_sha256":"429ebda2b916bb8347df3134470a6609d24d2b93d30537ee2bbfa4fb11377471"},"schema_version":"1.0","source":{"id":"2405.03133","kind":"arxiv","version":2}},"canonical_sha256":"44c74e7364ccd44242f8b803e7fa4e954462343e7f7daaac84c01b3f2bee2db0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"44c74e7364ccd44242f8b803e7fa4e954462343e7f7daaac84c01b3f2bee2db0","first_computed_at":"2026-07-05T08:56:19.510836Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:56:19.510836Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"GjMHx6DdZXIU2BeaUy8ExctC3G3dKP39d+6QLDkv3r/3lUkQowP8CbrAv/N7kN7J/vHKDdNcQTr2SedN+8R+Aw==","signature_status":"signed_v1","signed_at":"2026-07-05T08:56:19.511284Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.03133","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6fc29cfbbc32205953ce8450bfd6c880f48ab8cd97a471a1e953602fc05034c0","sha256:3a53a230878476c922505cba1fa06ac33ab3651b2fe8326e36d72ec2198e8162"],"state_sha256":"dbb41770e66298dadcef9e09b589a214918ebc4a3ffbf83df92882256d62f6f2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RP/AK2zA7voPl5DlIk6RyBrPkoZqQyO3CtxOzwMRVfTXQ05eS1WpD8CRVU3Y7VB+gvIwOsBuJVJDVBpdvA5sCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T09:41:53.034696Z","bundle_sha256":"9a2efc585d3b27321fcd703e155d626fcc15fa7efb6fcbaf95cab72e00617fb3"}}