{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:DA56ANVEH4MBDZQM6SU5XHFDJA","short_pith_number":"pith:DA56ANVE","canonical_record":{"source":{"id":"2502.12089","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-02-17T18:06:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d2fdd972bd2de0e75c4f7c2e5d4a1f5a28d9b261eb11883e9eb239817aa6177a","abstract_canon_sha256":"cf80eb25237c626de625cbfa800f4b4105884f0aa305cb9a7626d99462643a54"},"schema_version":"1.0"},"canonical_sha256":"183be036a43f1811e60cf4a9db9ca34821d876408ea0c62aec24a1f1f9162d22","source":{"kind":"arxiv","id":"2502.12089","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.12089","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"arxiv_version","alias_value":"2502.12089v3","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.12089","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"pith_short_12","alias_value":"DA56ANVEH4MB","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"pith_short_16","alias_value":"DA56ANVEH4MBDZQM","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"pith_short_8","alias_value":"DA56ANVE","created_at":"2026-07-05T11:15:42Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:DA56ANVEH4MBDZQM6SU5XHFDJA","target":"record","payload":{"canonical_record":{"source":{"id":"2502.12089","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-02-17T18:06:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d2fdd972bd2de0e75c4f7c2e5d4a1f5a28d9b261eb11883e9eb239817aa6177a","abstract_canon_sha256":"cf80eb25237c626de625cbfa800f4b4105884f0aa305cb9a7626d99462643a54"},"schema_version":"1.0"},"canonical_sha256":"183be036a43f1811e60cf4a9db9ca34821d876408ea0c62aec24a1f1f9162d22","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:15:42.865366Z","signature_b64":"RM7YjNwTTjlhaVK20FmhXcXApJoaR23nxcb7t6XMEh23hX+gVB/veJpA1dC3lUP0O2yF2jvoRPwOxaxbfLR/Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"183be036a43f1811e60cf4a9db9ca34821d876408ea0c62aec24a1f1f9162d22","last_reissued_at":"2026-07-05T11:15:42.864925Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:15:42.864925Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.12089","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:15:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fGZkPumyrPv+WcE/u1KPTL3mH5yNtaEe9Vp70cKu6ay0Q3NEunpjnPOT5VUOZPXoew9MjsxX4p84chyJ8ntHAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T15:12:17.306475Z"},"content_sha256":"d139212a8ab952c821c0d07c519740df6ad02312c7aa3c70c18b6429f10c31f5","schema_version":"1.0","event_id":"sha256:d139212a8ab952c821c0d07c519740df6ad02312c7aa3c70c18b6429f10c31f5"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:DA56ANVEH4MBDZQM6SU5XHFDJA","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"How Compositional Generalization and Creativity Improve as Diffusion Models are Trained","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Alessandro Favero, Antonio Sclocchi, Francesco Cagnetta, Matthieu Wyart, Pascal Frossard","submitted_at":"2025-02-17T18:06:33Z","abstract_excerpt":"Natural data is often organized as a hierarchical composition of features. How many samples do generative models need in order to learn the composition rules, so as to produce a combinatorially large number of novel data? What signal in the data is exploited to learn those rules? We investigate these questions in the context of diffusion models both theoretically and empirically. Theoretically, we consider a simple probabilistic context-free grammar - a tree-like graphical model used to represent the hierarchical and compositional structure of data such as language and images. We demonstrate t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.12089","kind":"arxiv","version":3},"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/2502.12089/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":"2026-07-05T11:15:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DwcZQo8rrtUXKQYIPU/UgCbcEnR3lsrjmoS6n8lbGqpn7WY1pNi9fiZ9SzDxoF2k85M8mdp9yLbJWM6UuM/VDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T15:12:17.306854Z"},"content_sha256":"c7ee61ff0fa143bf7a83d1bbb9e56d0290c08c1dc5f872afcff71ef660c00876","schema_version":"1.0","event_id":"sha256:c7ee61ff0fa143bf7a83d1bbb9e56d0290c08c1dc5f872afcff71ef660c00876"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DA56ANVEH4MBDZQM6SU5XHFDJA/bundle.json","state_url":"https://pith.science/pith/DA56ANVEH4MBDZQM6SU5XHFDJA/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DA56ANVEH4MBDZQM6SU5XHFDJA/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":"2026-08-04T15:12:17Z","links":{"resolver":"https://pith.science/pith/DA56ANVEH4MBDZQM6SU5XHFDJA","bundle":"https://pith.science/pith/DA56ANVEH4MBDZQM6SU5XHFDJA/bundle.json","state":"https://pith.science/pith/DA56ANVEH4MBDZQM6SU5XHFDJA/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DA56ANVEH4MBDZQM6SU5XHFDJA/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:DA56ANVEH4MBDZQM6SU5XHFDJA","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":"cf80eb25237c626de625cbfa800f4b4105884f0aa305cb9a7626d99462643a54","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-02-17T18:06:33Z","title_canon_sha256":"d2fdd972bd2de0e75c4f7c2e5d4a1f5a28d9b261eb11883e9eb239817aa6177a"},"schema_version":"1.0","source":{"id":"2502.12089","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.12089","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"arxiv_version","alias_value":"2502.12089v3","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.12089","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"pith_short_12","alias_value":"DA56ANVEH4MB","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"pith_short_16","alias_value":"DA56ANVEH4MBDZQM","created_at":"2026-07-05T11:15:42Z"},{"alias_kind":"pith_short_8","alias_value":"DA56ANVE","created_at":"2026-07-05T11:15:42Z"}],"graph_snapshots":[{"event_id":"sha256:c7ee61ff0fa143bf7a83d1bbb9e56d0290c08c1dc5f872afcff71ef660c00876","target":"graph","created_at":"2026-07-05T11:15:42Z","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/2502.12089/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Natural data is often organized as a hierarchical composition of features. How many samples do generative models need in order to learn the composition rules, so as to produce a combinatorially large number of novel data? What signal in the data is exploited to learn those rules? We investigate these questions in the context of diffusion models both theoretically and empirically. Theoretically, we consider a simple probabilistic context-free grammar - a tree-like graphical model used to represent the hierarchical and compositional structure of data such as language and images. We demonstrate t","authors_text":"Alessandro Favero, Antonio Sclocchi, Francesco Cagnetta, Matthieu Wyart, Pascal Frossard","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-02-17T18:06:33Z","title":"How Compositional Generalization and Creativity Improve as Diffusion Models are Trained"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.12089","kind":"arxiv","version":3},"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:d139212a8ab952c821c0d07c519740df6ad02312c7aa3c70c18b6429f10c31f5","target":"record","created_at":"2026-07-05T11:15:42Z","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":"cf80eb25237c626de625cbfa800f4b4105884f0aa305cb9a7626d99462643a54","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-02-17T18:06:33Z","title_canon_sha256":"d2fdd972bd2de0e75c4f7c2e5d4a1f5a28d9b261eb11883e9eb239817aa6177a"},"schema_version":"1.0","source":{"id":"2502.12089","kind":"arxiv","version":3}},"canonical_sha256":"183be036a43f1811e60cf4a9db9ca34821d876408ea0c62aec24a1f1f9162d22","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"183be036a43f1811e60cf4a9db9ca34821d876408ea0c62aec24a1f1f9162d22","first_computed_at":"2026-07-05T11:15:42.864925Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:15:42.864925Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"RM7YjNwTTjlhaVK20FmhXcXApJoaR23nxcb7t6XMEh23hX+gVB/veJpA1dC3lUP0O2yF2jvoRPwOxaxbfLR/Bw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:15:42.865366Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.12089","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d139212a8ab952c821c0d07c519740df6ad02312c7aa3c70c18b6429f10c31f5","sha256:c7ee61ff0fa143bf7a83d1bbb9e56d0290c08c1dc5f872afcff71ef660c00876"],"state_sha256":"1cddd32e6d90eac1a4b7d5fc63ed1aeb43eea9548dd624046ab40603d6e65403"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"XTy08ZMgY+buMlLzVFPgEYINrtsIcBZXh06nCuZx4JDEc06yzS7iNjNmaOxOBiZ5ha3mQNso2obIsxaZIEqoAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T15:12:17.310829Z","bundle_sha256":"e2c3253f4d7e56656dd6577ba73467f03695579082e422b137e34c6a9e1e9c86"}}