{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:T5OAOHY5D7LMISS2FKQUUVM6P3","short_pith_number":"pith:T5OAOHY5","schema_version":"1.0","canonical_sha256":"9f5c071f1d1fd6c44a5a2aa14a559e7eda223639ad0b95ae6d90111181129a9f","source":{"kind":"arxiv","id":"2412.01819","version":4},"attestation_state":"computed","paper":{"title":"Switti: Designing Scale-Wise Transformers for Text-to-Image Synthesis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Anton Voronov, Denis Kuznedelev, Dmitry Baranchuk, Mikhail Khoroshikh, Valentin Khrulkov","submitted_at":"2024-12-02T18:57:41Z","abstract_excerpt":"This work presents Switti, a scale-wise transformer for text-to-image generation. We start by adapting an existing next-scale prediction autoregressive (AR) architecture to T2I generation, investigating and mitigating training stability issues in the process. Next, we argue that scale-wise transformers do not require causality and propose a non-causal counterpart facilitating ~21% faster sampling and lower memory usage while also achieving slightly better generation quality. Furthermore, we reveal that classifier-free guidance at high-resolution scales is often unnecessary and can even degrade"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2412.01819","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-12-02T18:57:41Z","cross_cats_sorted":[],"title_canon_sha256":"89734e165626217a39b129564da612b6ade3ee832dd618708589f68971f8e402","abstract_canon_sha256":"8eeb2cb70aa8d3da6c1c47bddcf24b01e20dd25a94725c81cb46f9c377d13fed"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:35:42.619049Z","signature_b64":"HbCHAWb3oNvxlWPmHUhNHTIRed5KuIEU5jn0c4OlbYtJ/fak0XvxZ9YIiLnSLF0Cs+mLQEYYhdtRiyAI4dC9AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9f5c071f1d1fd6c44a5a2aa14a559e7eda223639ad0b95ae6d90111181129a9f","last_reissued_at":"2026-07-05T10:35:42.618517Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:35:42.618517Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Switti: Designing Scale-Wise Transformers for Text-to-Image Synthesis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Anton Voronov, Denis Kuznedelev, Dmitry Baranchuk, Mikhail Khoroshikh, Valentin Khrulkov","submitted_at":"2024-12-02T18:57:41Z","abstract_excerpt":"This work presents Switti, a scale-wise transformer for text-to-image generation. We start by adapting an existing next-scale prediction autoregressive (AR) architecture to T2I generation, investigating and mitigating training stability issues in the process. Next, we argue that scale-wise transformers do not require causality and propose a non-causal counterpart facilitating ~21% faster sampling and lower memory usage while also achieving slightly better generation quality. Furthermore, we reveal that classifier-free guidance at high-resolution scales is often unnecessary and can even degrade"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.01819","kind":"arxiv","version":4},"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/2412.01819/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2412.01819","created_at":"2026-07-05T10:35:42.618577+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.01819v4","created_at":"2026-07-05T10:35:42.618577+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.01819","created_at":"2026-07-05T10:35:42.618577+00:00"},{"alias_kind":"pith_short_12","alias_value":"T5OAOHY5D7LM","created_at":"2026-07-05T10:35:42.618577+00:00"},{"alias_kind":"pith_short_16","alias_value":"T5OAOHY5D7LMISS2","created_at":"2026-07-05T10:35:42.618577+00:00"},{"alias_kind":"pith_short_8","alias_value":"T5OAOHY5","created_at":"2026-07-05T10:35:42.618577+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06078","citing_title":"Knowledge Distillation for Visual Autoregressive Models","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14591","citing_title":"Prompt-Guided Image Editing with Masked Logit Nudging in Visual Autoregressive Models","ref_index":45,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3","json":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3.json","graph_json":"https://pith.science/api/pith-number/T5OAOHY5D7LMISS2FKQUUVM6P3/graph.json","events_json":"https://pith.science/api/pith-number/T5OAOHY5D7LMISS2FKQUUVM6P3/events.json","paper":"https://pith.science/paper/T5OAOHY5"},"agent_actions":{"view_html":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3","download_json":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3.json","view_paper":"https://pith.science/paper/T5OAOHY5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.01819&json=true","fetch_graph":"https://pith.science/api/pith-number/T5OAOHY5D7LMISS2FKQUUVM6P3/graph.json","fetch_events":"https://pith.science/api/pith-number/T5OAOHY5D7LMISS2FKQUUVM6P3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3/action/storage_attestation","attest_author":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3/action/author_attestation","sign_citation":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3/action/citation_signature","submit_replication":"https://pith.science/pith/T5OAOHY5D7LMISS2FKQUUVM6P3/action/replication_record"}},"created_at":"2026-07-05T10:35:42.618577+00:00","updated_at":"2026-07-05T10:35:42.618577+00:00"}