{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BBB77I2HYL577IPYFG53Y4Y3HD","short_pith_number":"pith:BBB77I2H","schema_version":"1.0","canonical_sha256":"0843ffa347c2fbffa1f829bbbc731b38f9a4ab855b2497e86257869edae7d19c","source":{"kind":"arxiv","id":"2411.13050","version":1},"attestation_state":"computed","paper":{"title":"Topkima-Former: Low-energy, Low-Latency Inference for Transformers using top-k In-memory ADC","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AR","authors_text":"Arindam Basu, Hongjie Liu, Hongyang Shang, Junyi Yang, Shuai Dong, Xiaofeng Yang, Xiaoqi Peng, Ye Ke","submitted_at":"2024-11-20T05:51:35Z","abstract_excerpt":"Transformer model has gained prominence as a popular deep neural network architecture for neural language processing (NLP) and computer vision (CV) applications. However, the extensive use of nonlinear operations, like softmax, poses a performance bottleneck during transformer inference and comprises up to 40% of the total latency. Hence, we propose innovations at the circuit, architecture, and algorithm levels to accelerate the transformer. At the circuit level, we propose topkima-combining top-k activation selection with in-memory ADC (IMA) to implement a low-energy and low-latency softmax w"},"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":"2411.13050","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2024-11-20T05:51:35Z","cross_cats_sorted":[],"title_canon_sha256":"ccfd78df6f3aaa48ed40789b2b2efa393e6c86b9fe948d76c75f8a2841a262ee","abstract_canon_sha256":"26149de485450598ac119d7fcd8e8d76f7c0099395ee278d31c961eaa356ae89"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:38:04.274177Z","signature_b64":"3Xzz3u9d3UP6fbfyigLqIwK6JWJZZNpqswICysgCm4i3df8tsZeluj8ctVv/+FK7HE2fHdG6Dd43zArv6d1uCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0843ffa347c2fbffa1f829bbbc731b38f9a4ab855b2497e86257869edae7d19c","last_reissued_at":"2026-07-05T09:38:04.273704Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:38:04.273704Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Topkima-Former: Low-energy, Low-Latency Inference for Transformers using top-k In-memory ADC","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AR","authors_text":"Arindam Basu, Hongjie Liu, Hongyang Shang, Junyi Yang, Shuai Dong, Xiaofeng Yang, Xiaoqi Peng, Ye Ke","submitted_at":"2024-11-20T05:51:35Z","abstract_excerpt":"Transformer model has gained prominence as a popular deep neural network architecture for neural language processing (NLP) and computer vision (CV) applications. However, the extensive use of nonlinear operations, like softmax, poses a performance bottleneck during transformer inference and comprises up to 40% of the total latency. Hence, we propose innovations at the circuit, architecture, and algorithm levels to accelerate the transformer. At the circuit level, we propose topkima-combining top-k activation selection with in-memory ADC (IMA) to implement a low-energy and low-latency softmax w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.13050","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/2411.13050/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":"2411.13050","created_at":"2026-07-05T09:38:04.273766+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.13050v1","created_at":"2026-07-05T09:38:04.273766+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.13050","created_at":"2026-07-05T09:38:04.273766+00:00"},{"alias_kind":"pith_short_12","alias_value":"BBB77I2HYL57","created_at":"2026-07-05T09:38:04.273766+00:00"},{"alias_kind":"pith_short_16","alias_value":"BBB77I2HYL577IPY","created_at":"2026-07-05T09:38:04.273766+00:00"},{"alias_kind":"pith_short_8","alias_value":"BBB77I2H","created_at":"2026-07-05T09:38:04.273766+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/BBB77I2HYL577IPYFG53Y4Y3HD","json":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD.json","graph_json":"https://pith.science/api/pith-number/BBB77I2HYL577IPYFG53Y4Y3HD/graph.json","events_json":"https://pith.science/api/pith-number/BBB77I2HYL577IPYFG53Y4Y3HD/events.json","paper":"https://pith.science/paper/BBB77I2H"},"agent_actions":{"view_html":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD","download_json":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD.json","view_paper":"https://pith.science/paper/BBB77I2H","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.13050&json=true","fetch_graph":"https://pith.science/api/pith-number/BBB77I2HYL577IPYFG53Y4Y3HD/graph.json","fetch_events":"https://pith.science/api/pith-number/BBB77I2HYL577IPYFG53Y4Y3HD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD/action/storage_attestation","attest_author":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD/action/author_attestation","sign_citation":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD/action/citation_signature","submit_replication":"https://pith.science/pith/BBB77I2HYL577IPYFG53Y4Y3HD/action/replication_record"}},"created_at":"2026-07-05T09:38:04.273766+00:00","updated_at":"2026-07-05T09:38:04.273766+00:00"}