{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:QEL4HKLMTI4QJMQL32TN7UQGNO","short_pith_number":"pith:QEL4HKLM","canonical_record":{"source":{"id":"2306.07629","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-06-13T08:57:54Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"acec35eb64829b1a36c30afb3e7f970cec7326eefafea01465e7ae29805a157f","abstract_canon_sha256":"ff25315d6c87b11a6d20b92d2ebc27bb82fe942a3763acf57869dd844c81cdfa"},"schema_version":"1.0"},"canonical_sha256":"8117c3a96c9a3904b20bdea6dfd2066b9d7f70c7509f056aae77454796cbd03c","source":{"kind":"arxiv","id":"2306.07629","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2306.07629","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"arxiv_version","alias_value":"2306.07629v4","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.07629","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"pith_short_12","alias_value":"QEL4HKLMTI4Q","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"pith_short_16","alias_value":"QEL4HKLMTI4QJMQL","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"pith_short_8","alias_value":"QEL4HKLM","created_at":"2026-07-05T08:27:22Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:QEL4HKLMTI4QJMQL32TN7UQGNO","target":"record","payload":{"canonical_record":{"source":{"id":"2306.07629","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-06-13T08:57:54Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"acec35eb64829b1a36c30afb3e7f970cec7326eefafea01465e7ae29805a157f","abstract_canon_sha256":"ff25315d6c87b11a6d20b92d2ebc27bb82fe942a3763acf57869dd844c81cdfa"},"schema_version":"1.0"},"canonical_sha256":"8117c3a96c9a3904b20bdea6dfd2066b9d7f70c7509f056aae77454796cbd03c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:27:22.509187Z","signature_b64":"1DCIEXPOHds+nXnGyk6qpv+RdYjrPDvZndgStUZlrTVaIZgc+3oYJzZ28PGFexEzRd8S1YTMKMhIHFaWRc2ACQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8117c3a96c9a3904b20bdea6dfd2066b9d7f70c7509f056aae77454796cbd03c","last_reissued_at":"2026-07-05T08:27:22.508774Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:27:22.508774Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2306.07629","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:27:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+k0V0QsLj3nfKYR+eXTKktE+bqs5qHyeFh4RBAbxfDLoOzvjaEs5vSh6W0Wkb4X5LA66UxyhfrykxMMgE2E6CA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T10:05:09.755476Z"},"content_sha256":"f1e954faed493687abb885a19c6b5cdcd86563d705778de20b7c33b6532aa589","schema_version":"1.0","event_id":"sha256:f1e954faed493687abb885a19c6b5cdcd86563d705778de20b7c33b6532aa589"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:QEL4HKLMTI4QJMQL32TN7UQGNO","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"SqueezeLLM: Dense-and-Sparse Quantization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Amir Gholami, Coleman Hooper, Kurt Keutzer, Michael W. Mahoney, Sehoon Kim, Sheng Shen, Xiuyu Li, Zhen Dong","submitted_at":"2023-06-13T08:57:54Z","abstract_excerpt":"Generative Large Language Models (LLMs) have demonstrated remarkable results for a wide range of tasks. However, deploying these models for inference has been a significant challenge due to their unprecedented resource requirements. This has forced existing deployment frameworks to use multi-GPU inference pipelines, which are often complex and costly, or to use smaller and less performant models. In this work, we demonstrate that the main bottleneck for generative inference with LLMs is memory bandwidth, rather than compute, specifically for single batch inference. While quantization has emerg"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.07629","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/2306.07629/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":"2026-07-05T08:27:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Pa1bFbJ/l5INV2DwwuRfj6wQf62QjGAooNhwFlnTD+XEgQu5IOJu0Vky8221EWh60hlrND24KHZIcRr6u7nQBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T10:05:09.755993Z"},"content_sha256":"48715dbca0a53921ad66e46949f72b17ca084f6abe9a48f8e84e8256589df1d8","schema_version":"1.0","event_id":"sha256:48715dbca0a53921ad66e46949f72b17ca084f6abe9a48f8e84e8256589df1d8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/QEL4HKLMTI4QJMQL32TN7UQGNO/bundle.json","state_url":"https://pith.science/pith/QEL4HKLMTI4QJMQL32TN7UQGNO/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/QEL4HKLMTI4QJMQL32TN7UQGNO/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":"2026-08-06T10:05:09Z","links":{"resolver":"https://pith.science/pith/QEL4HKLMTI4QJMQL32TN7UQGNO","bundle":"https://pith.science/pith/QEL4HKLMTI4QJMQL32TN7UQGNO/bundle.json","state":"https://pith.science/pith/QEL4HKLMTI4QJMQL32TN7UQGNO/state.json","well_known_bundle":"https://pith.science/.well-known/pith/QEL4HKLMTI4QJMQL32TN7UQGNO/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:QEL4HKLMTI4QJMQL32TN7UQGNO","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":"ff25315d6c87b11a6d20b92d2ebc27bb82fe942a3763acf57869dd844c81cdfa","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-06-13T08:57:54Z","title_canon_sha256":"acec35eb64829b1a36c30afb3e7f970cec7326eefafea01465e7ae29805a157f"},"schema_version":"1.0","source":{"id":"2306.07629","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2306.07629","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"arxiv_version","alias_value":"2306.07629v4","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.07629","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"pith_short_12","alias_value":"QEL4HKLMTI4Q","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"pith_short_16","alias_value":"QEL4HKLMTI4QJMQL","created_at":"2026-07-05T08:27:22Z"},{"alias_kind":"pith_short_8","alias_value":"QEL4HKLM","created_at":"2026-07-05T08:27:22Z"}],"graph_snapshots":[{"event_id":"sha256:48715dbca0a53921ad66e46949f72b17ca084f6abe9a48f8e84e8256589df1d8","target":"graph","created_at":"2026-07-05T08:27:22Z","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/2306.07629/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Generative Large Language Models (LLMs) have demonstrated remarkable results for a wide range of tasks. However, deploying these models for inference has been a significant challenge due to their unprecedented resource requirements. This has forced existing deployment frameworks to use multi-GPU inference pipelines, which are often complex and costly, or to use smaller and less performant models. In this work, we demonstrate that the main bottleneck for generative inference with LLMs is memory bandwidth, rather than compute, specifically for single batch inference. While quantization has emerg","authors_text":"Amir Gholami, Coleman Hooper, Kurt Keutzer, Michael W. Mahoney, Sehoon Kim, Sheng Shen, Xiuyu Li, Zhen Dong","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-06-13T08:57:54Z","title":"SqueezeLLM: Dense-and-Sparse Quantization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.07629","kind":"arxiv","version":4},"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:f1e954faed493687abb885a19c6b5cdcd86563d705778de20b7c33b6532aa589","target":"record","created_at":"2026-07-05T08:27:22Z","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":"ff25315d6c87b11a6d20b92d2ebc27bb82fe942a3763acf57869dd844c81cdfa","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-06-13T08:57:54Z","title_canon_sha256":"acec35eb64829b1a36c30afb3e7f970cec7326eefafea01465e7ae29805a157f"},"schema_version":"1.0","source":{"id":"2306.07629","kind":"arxiv","version":4}},"canonical_sha256":"8117c3a96c9a3904b20bdea6dfd2066b9d7f70c7509f056aae77454796cbd03c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8117c3a96c9a3904b20bdea6dfd2066b9d7f70c7509f056aae77454796cbd03c","first_computed_at":"2026-07-05T08:27:22.508774Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:27:22.508774Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"1DCIEXPOHds+nXnGyk6qpv+RdYjrPDvZndgStUZlrTVaIZgc+3oYJzZ28PGFexEzRd8S1YTMKMhIHFaWRc2ACQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:27:22.509187Z","signed_message":"canonical_sha256_bytes"},"source_id":"2306.07629","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f1e954faed493687abb885a19c6b5cdcd86563d705778de20b7c33b6532aa589","sha256:48715dbca0a53921ad66e46949f72b17ca084f6abe9a48f8e84e8256589df1d8"],"state_sha256":"7a3630a80713b3f922db56c71bbddc00c602fc3137b758ea03a0ece77ce146cd"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"3vPwku3XuxNfWnb+gjxkYEPOXWN0nz9FHMZTe8IDYacnayzcNYnnhoAtO0/g+aUItUIlOo6WpvCskQz0Mv84AA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T10:05:09.761057Z","bundle_sha256":"6739e2d91ad6440161428a6c54ef7d04864d4ef9220b16a29c66ec97c97f2c9c"}}