{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:KWQNFBPXPD2QQBBWWQTOT4AQS3","short_pith_number":"pith:KWQNFBPX","schema_version":"1.0","canonical_sha256":"55a0d285f778f5080436b426e9f01096c1a7e4ac835fd0e0350cdc0d409befc1","source":{"kind":"arxiv","id":"2303.17604","version":1},"attestation_state":"computed","paper":{"title":"Token Merging for Fast Stable Diffusion","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Daniel Bolya, Judy Hoffman","submitted_at":"2023-03-30T17:59:58Z","abstract_excerpt":"The landscape of image generation has been forever changed by open vocabulary diffusion models. However, at their core these models use transformers, which makes generation slow. Better implementations to increase the throughput of these transformers have emerged, but they still evaluate the entire model. In this paper, we instead speed up diffusion models by exploiting natural redundancy in generated images by merging redundant tokens. After making some diffusion-specific improvements to Token Merging (ToMe), our ToMe for Stable Diffusion can reduce the number of tokens in an existing Stable "},"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":"2303.17604","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-03-30T17:59:58Z","cross_cats_sorted":[],"title_canon_sha256":"327c14a5e273ce8bd6c6b60a39348ec609f18da0540209168987981cb3628633","abstract_canon_sha256":"38519870b8fd6a0920075c925cbef12b7ac67ef7906111fa5d601edf6f4a6531"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:56:36.067006Z","signature_b64":"KN4wcMpBI0Se9SY/xfT+NmKnlS7qbtt0d/qrso/jRs6k3M7leknYC9A2Of2YGv+A87X80GcPyFhhbk0VGBcsCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"55a0d285f778f5080436b426e9f01096c1a7e4ac835fd0e0350cdc0d409befc1","last_reissued_at":"2026-07-05T05:56:36.066552Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:56:36.066552Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Token Merging for Fast Stable Diffusion","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Daniel Bolya, Judy Hoffman","submitted_at":"2023-03-30T17:59:58Z","abstract_excerpt":"The landscape of image generation has been forever changed by open vocabulary diffusion models. However, at their core these models use transformers, which makes generation slow. Better implementations to increase the throughput of these transformers have emerged, but they still evaluate the entire model. In this paper, we instead speed up diffusion models by exploiting natural redundancy in generated images by merging redundant tokens. After making some diffusion-specific improvements to Token Merging (ToMe), our ToMe for Stable Diffusion can reduce the number of tokens in an existing Stable "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2303.17604","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/2303.17604/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":"2303.17604","created_at":"2026-07-05T05:56:36.066611+00:00"},{"alias_kind":"arxiv_version","alias_value":"2303.17604v1","created_at":"2026-07-05T05:56:36.066611+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.17604","created_at":"2026-07-05T05:56:36.066611+00:00"},{"alias_kind":"pith_short_12","alias_value":"KWQNFBPXPD2Q","created_at":"2026-07-05T05:56:36.066611+00:00"},{"alias_kind":"pith_short_16","alias_value":"KWQNFBPXPD2QQBBW","created_at":"2026-07-05T05:56:36.066611+00:00"},{"alias_kind":"pith_short_8","alias_value":"KWQNFBPX","created_at":"2026-07-05T05:56:36.066611+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.21042","citing_title":"Dynamic Video Generation: Shaping Video Generation Across Time and Space","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06270","citing_title":"Spark3R: Asymmetric Token Reduction Makes Fast Feed-Forward 3D Reconstruction","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2602.13357","citing_title":"AdaCorrection: Adaptive Offset Cache Correction for Accurate Diffusion Transformers","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06270","citing_title":"Spark3R: Asymmetric Token Reduction Makes Fast Feed-Forward 3D Reconstruction","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00707","citing_title":"PhysEdit: Physically-Consistent Region-Aware Image Editing via Adaptive Spatio-Temporal Reasoning","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13432","citing_title":"MaMe & MaRe: Matrix-Based Token Merging and Restoration for Efficient Visual Perception and Synthesis","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3","json":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3.json","graph_json":"https://pith.science/api/pith-number/KWQNFBPXPD2QQBBWWQTOT4AQS3/graph.json","events_json":"https://pith.science/api/pith-number/KWQNFBPXPD2QQBBWWQTOT4AQS3/events.json","paper":"https://pith.science/paper/KWQNFBPX"},"agent_actions":{"view_html":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3","download_json":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3.json","view_paper":"https://pith.science/paper/KWQNFBPX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2303.17604&json=true","fetch_graph":"https://pith.science/api/pith-number/KWQNFBPXPD2QQBBWWQTOT4AQS3/graph.json","fetch_events":"https://pith.science/api/pith-number/KWQNFBPXPD2QQBBWWQTOT4AQS3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3/action/storage_attestation","attest_author":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3/action/author_attestation","sign_citation":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3/action/citation_signature","submit_replication":"https://pith.science/pith/KWQNFBPXPD2QQBBWWQTOT4AQS3/action/replication_record"}},"created_at":"2026-07-05T05:56:36.066611+00:00","updated_at":"2026-07-05T05:56:36.066611+00:00"}