{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:PHMSFVWFRSYZ6RX27DL7S7JHXZ","short_pith_number":"pith:PHMSFVWF","schema_version":"1.0","canonical_sha256":"79d922d6c58cb19f46faf8d7f97d27be45bb173ba714c618637c9a65c94bb8f5","source":{"kind":"arxiv","id":"2605.27003","version":1},"attestation_state":"computed","paper":{"title":"Timestep-Aware SVDQuant-GPTQ for W4A4 Quantization of Wan2.2-I2V","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Dezhong Yao, Hai Jin, Junhao Wu","submitted_at":"2026-05-26T13:24:01Z","abstract_excerpt":"W4A4 quantization of large video diffusion Transformers offers substantial memory savings but is hindered by two main challenges: sparse large-magnitude activation outliers, and strongly timestep-dependent activation distributions across the multi-step denoising trajectory. These difficulties are compounded by Wan2.2-I2V's two-expert Mixture-of-Experts DiT design, whose high-noise and low-noise experts exhibit distinct quantization sensitivities that a single global calibration policy cannot capture. We propose a post-training quantization framework combining SVDQuant-based low-rank outlier co"},"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":"2605.27003","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-26T13:24:01Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e1e57dc60af29e40266e3f234b338bb305bdeab1135e5cf5d0106385b7f8e5ac","abstract_canon_sha256":"a18b357d6faec8e99341e5a1b8652074e4a9d38a99e4fe16facbf9b6c02d5819"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:06:23.621043Z","signature_b64":"QeQPULVo1XwPh9ZNLp36bM0tIe5/goYhuo0nbvJBjmrLlMYLcE7+loetolOZamHTIEGmQPkM9y1eripnmAsbBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"79d922d6c58cb19f46faf8d7f97d27be45bb173ba714c618637c9a65c94bb8f5","last_reissued_at":"2026-05-27T01:06:23.620438Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:06:23.620438Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Timestep-Aware SVDQuant-GPTQ for W4A4 Quantization of Wan2.2-I2V","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Dezhong Yao, Hai Jin, Junhao Wu","submitted_at":"2026-05-26T13:24:01Z","abstract_excerpt":"W4A4 quantization of large video diffusion Transformers offers substantial memory savings but is hindered by two main challenges: sparse large-magnitude activation outliers, and strongly timestep-dependent activation distributions across the multi-step denoising trajectory. These difficulties are compounded by Wan2.2-I2V's two-expert Mixture-of-Experts DiT design, whose high-noise and low-noise experts exhibit distinct quantization sensitivities that a single global calibration policy cannot capture. We propose a post-training quantization framework combining SVDQuant-based low-rank outlier co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.27003","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/2605.27003/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":"2605.27003","created_at":"2026-05-27T01:06:23.620520+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.27003v1","created_at":"2026-05-27T01:06:23.620520+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.27003","created_at":"2026-05-27T01:06:23.620520+00:00"},{"alias_kind":"pith_short_12","alias_value":"PHMSFVWFRSYZ","created_at":"2026-05-27T01:06:23.620520+00:00"},{"alias_kind":"pith_short_16","alias_value":"PHMSFVWFRSYZ6RX2","created_at":"2026-05-27T01:06:23.620520+00:00"},{"alias_kind":"pith_short_8","alias_value":"PHMSFVWF","created_at":"2026-05-27T01:06:23.620520+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ","json":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ.json","graph_json":"https://pith.science/api/pith-number/PHMSFVWFRSYZ6RX27DL7S7JHXZ/graph.json","events_json":"https://pith.science/api/pith-number/PHMSFVWFRSYZ6RX27DL7S7JHXZ/events.json","paper":"https://pith.science/paper/PHMSFVWF"},"agent_actions":{"view_html":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ","download_json":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ.json","view_paper":"https://pith.science/paper/PHMSFVWF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.27003&json=true","fetch_graph":"https://pith.science/api/pith-number/PHMSFVWFRSYZ6RX27DL7S7JHXZ/graph.json","fetch_events":"https://pith.science/api/pith-number/PHMSFVWFRSYZ6RX27DL7S7JHXZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ/action/storage_attestation","attest_author":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ/action/author_attestation","sign_citation":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ/action/citation_signature","submit_replication":"https://pith.science/pith/PHMSFVWFRSYZ6RX27DL7S7JHXZ/action/replication_record"}},"created_at":"2026-05-27T01:06:23.620520+00:00","updated_at":"2026-05-27T01:06:23.620520+00:00"}