{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:W7PUI4TQIWD6DSEGQ2X3HOBFUY","short_pith_number":"pith:W7PUI4TQ","schema_version":"1.0","canonical_sha256":"b7df4472704587e1c88686afb3b825a60405dfd1a556700b4d76d53eecf7627b","source":{"kind":"arxiv","id":"2411.04965","version":1},"attestation_state":"computed","paper":{"title":"BitNet a4.8: 4-bit Activations for 1-bit LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Furu Wei, Hongyu Wang, Shuming Ma","submitted_at":"2024-11-07T18:41:50Z","abstract_excerpt":"Recent research on the 1-bit Large Language Models (LLMs), such as BitNet b1.58, presents a promising direction for reducing the inference cost of LLMs while maintaining their performance. In this work, we introduce BitNet a4.8, enabling 4-bit activations for 1-bit LLMs. BitNet a4.8 employs a hybrid quantization and sparsification strategy to mitigate the quantization errors introduced by the outlier channels. Specifically, we utilize 4-bit activations for inputs to the attention and feed-forward network layers, while sparsifying intermediate states followed with 8-bit quantization. Extensive "},"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":"2411.04965","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-11-07T18:41:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"3fa61d421148968ea338fc36ac78293e474188250459ede00886c8308e48d67c","abstract_canon_sha256":"4769d864016ef408e806b96723225cfb4df42282698dd1ee53bc28646de99e4b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:32:36.891353Z","signature_b64":"CqkkjQFfk3wERYT5VPGJjAhvjv8SJHhHQn1qftWGjvNwykLHpJcTeAmleylpjzAAkXDSI9O+VRXHXwLfu0OkBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b7df4472704587e1c88686afb3b825a60405dfd1a556700b4d76d53eecf7627b","last_reissued_at":"2026-07-05T09:32:36.890830Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:32:36.890830Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BitNet a4.8: 4-bit Activations for 1-bit LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Furu Wei, Hongyu Wang, Shuming Ma","submitted_at":"2024-11-07T18:41:50Z","abstract_excerpt":"Recent research on the 1-bit Large Language Models (LLMs), such as BitNet b1.58, presents a promising direction for reducing the inference cost of LLMs while maintaining their performance. In this work, we introduce BitNet a4.8, enabling 4-bit activations for 1-bit LLMs. BitNet a4.8 employs a hybrid quantization and sparsification strategy to mitigate the quantization errors introduced by the outlier channels. Specifically, we utilize 4-bit activations for inputs to the attention and feed-forward network layers, while sparsifying intermediate states followed with 8-bit quantization. Extensive "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.04965","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/2411.04965/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":"2411.04965","created_at":"2026-07-05T09:32:36.890896+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.04965v1","created_at":"2026-07-05T09:32:36.890896+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.04965","created_at":"2026-07-05T09:32:36.890896+00:00"},{"alias_kind":"pith_short_12","alias_value":"W7PUI4TQIWD6","created_at":"2026-07-05T09:32:36.890896+00:00"},{"alias_kind":"pith_short_16","alias_value":"W7PUI4TQIWD6DSEG","created_at":"2026-07-05T09:32:36.890896+00:00"},{"alias_kind":"pith_short_8","alias_value":"W7PUI4TQ","created_at":"2026-07-05T09:32:36.890896+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.05109","citing_title":"Tiny but Mighty: A Software-Hardware Co-Design Approach for Efficient Multimodal Inference on Battery-Powered Small Devices","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19167","citing_title":"LBLLM: Lightweight Binarization of Large Language Models via Three-Stage Distillation","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY","json":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY.json","graph_json":"https://pith.science/api/pith-number/W7PUI4TQIWD6DSEGQ2X3HOBFUY/graph.json","events_json":"https://pith.science/api/pith-number/W7PUI4TQIWD6DSEGQ2X3HOBFUY/events.json","paper":"https://pith.science/paper/W7PUI4TQ"},"agent_actions":{"view_html":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY","download_json":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY.json","view_paper":"https://pith.science/paper/W7PUI4TQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.04965&json=true","fetch_graph":"https://pith.science/api/pith-number/W7PUI4TQIWD6DSEGQ2X3HOBFUY/graph.json","fetch_events":"https://pith.science/api/pith-number/W7PUI4TQIWD6DSEGQ2X3HOBFUY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY/action/storage_attestation","attest_author":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY/action/author_attestation","sign_citation":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY/action/citation_signature","submit_replication":"https://pith.science/pith/W7PUI4TQIWD6DSEGQ2X3HOBFUY/action/replication_record"}},"created_at":"2026-07-05T09:32:36.890896+00:00","updated_at":"2026-07-05T09:32:36.890896+00:00"}