{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:KW54KIDFZCTH4FO4O4RS6VPIV7","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":"93d3c5dddc0333e446f23e0119f911c68110fb647f6e9d9a5b4dc35c69d290b1","cross_cats_sorted":["cs.AI","cs.CL","cs.LG","cs.MM"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-28T17:59:12Z","title_canon_sha256":"831ffdfc694e4500642984da60782b2e5f1362bd1a2f56a85198952d10fff66a"},"schema_version":"1.0","source":{"id":"2303.16199","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2303.16199","created_at":"2026-05-17T23:39:19Z"},{"alias_kind":"arxiv_version","alias_value":"2303.16199v3","created_at":"2026-05-17T23:39:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.16199","created_at":"2026-05-17T23:39:19Z"},{"alias_kind":"pith_short_12","alias_value":"KW54KIDFZCTH","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"KW54KIDFZCTH4FO4","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"KW54KIDF","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:61116247afe57e9d38ab998a0c8d53264d2254ec25159a9e4948740efbd6214e","target":"graph","created_at":"2026-05-17T23:39:19Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"With our efficient training, LLaMA-Adapter can generate high-quality responses, comparable to Alpaca with fully fine-tuned 7B parameters."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The zero-initialized attention mechanism with zero gating adaptively injects the new instructional cues into LLaMA while effectively preserving its pre-trained knowledge."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"LLaMA-Adapter turns frozen LLaMA 7B into a capable instruction follower using only 1.2M new parameters and zero-init attention, matching Alpaca while extending to image-conditioned reasoning on ScienceQA and COCO."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"LLaMA-Adapter adapts frozen LLaMA to follow instructions using only 1.2 million added parameters."}],"snapshot_sha256":"50a8db6213e450be59e73c21dc566ff19f998250c31b89e53264e3b522e0b4ee"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"We present LLaMA-Adapter, a lightweight adaption method to efficiently fine-tune LLaMA into an instruction-following model. Using 52K self-instruct demonstrations, LLaMA-Adapter only introduces 1.2M learnable parameters upon the frozen LLaMA 7B model, and costs less than one hour for fine-tuning on 8 A100 GPUs. Specifically, we adopt a set of learnable adaption prompts, and prepend them to the word tokens at higher transformer layers. Then, a zero-initialized attention mechanism with zero gating is proposed, which adaptively injects the new instructional cues into LLaMA, while effectively pres","authors_text":"Aojun Zhou, Chris Liu, Hongsheng Li, Jiaming Han, Pan Lu, Peng Gao, Renrui Zhang, Shilin Yan, Xiangfei Hu, Yu Qiao","cross_cats":["cs.AI","cs.CL","cs.LG","cs.MM"],"headline":"LLaMA-Adapter adapts frozen LLaMA to follow instructions using only 1.2 million added parameters.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-28T17:59:12Z","title":"LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attention"},"references":{"count":278,"internal_anchors":45,"resolved_work":278,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Alpaca-lora. https://github.com/tloen/alpaca-lora, 2023","work_id":"9a06075c-ae6a-4dd1-a289-ebde72983a0f","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Flamingo: a visual language model for few-shot learning","work_id":"01d90a48-13d8-4bc8-a06a-97adb5201146","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Open llm leaderboard","work_id":"5d0d6bb6-6bf7-45f4-a39e-40222560248f","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Language models are few-shot learners","work_id":"b5af3a68-2622-4421-b39b-b1d2fbde2d8d","year":1901},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":6,"title":"Introduction to the conll-2004 shared task: Semantic role labeling","work_id":"c0bd2045-d243-4c6d-a996-2634541c2b6b","year":2004}],"snapshot_sha256":"14c936933f9c610e90f0dbbeae5ab392bfc5944dc15cd86e833a0b401fe6c09b"},"source":{"id":"2303.16199","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-14T23:01:25.060113Z","id":"033679df-4a70-4f12-9af8-1266f199bfda","model_set":{"reader":"grok-4.3"},"one_line_summary":"LLaMA-Adapter turns frozen LLaMA 7B into a capable instruction follower using only 1.2M new parameters and zero-init attention, matching Alpaca while extending to image-conditioned reasoning on ScienceQA and COCO.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"LLaMA-Adapter adapts frozen LLaMA to follow instructions using only 1.2 million added parameters.","strongest_claim":"With our efficient training, LLaMA-Adapter can generate high-quality responses, comparable to Alpaca with fully fine-tuned 7B parameters.","weakest_assumption":"The zero-initialized attention mechanism with zero gating adaptively injects the new instructional cues into LLaMA while effectively preserving its pre-trained knowledge."}},"verdict_id":"033679df-4a70-4f12-9af8-1266f199bfda"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c1cf0770929562ff4e1cac2066ab440d8a2ad8d2a26c69d8092657c379443fc7","target":"record","created_at":"2026-05-17T23:39:19Z","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":"93d3c5dddc0333e446f23e0119f911c68110fb647f6e9d9a5b4dc35c69d290b1","cross_cats_sorted":["cs.AI","cs.CL","cs.LG","cs.MM"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-28T17:59:12Z","title_canon_sha256":"831ffdfc694e4500642984da60782b2e5f1362bd1a2f56a85198952d10fff66a"},"schema_version":"1.0","source":{"id":"2303.16199","kind":"arxiv","version":3}},"canonical_sha256":"55bbc52065c8a67e15dc77232f55e8afc43b5a16ab565fd34a0b47466d3502a0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"55bbc52065c8a67e15dc77232f55e8afc43b5a16ab565fd34a0b47466d3502a0","first_computed_at":"2026-05-17T23:39:19.917730Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:19.917730Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"+IWjWxXZ/fywm56IzaBdXj8oSF1oOvrIxY0G+5gAL6FyBmmw4l+yeKQCqbLC1pNtuoGfw5kwXnTUfhAtbSbJAA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:19.918526Z","signed_message":"canonical_sha256_bytes"},"source_id":"2303.16199","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c1cf0770929562ff4e1cac2066ab440d8a2ad8d2a26c69d8092657c379443fc7","sha256:61116247afe57e9d38ab998a0c8d53264d2254ec25159a9e4948740efbd6214e"],"state_sha256":"2df89d47b7ed28a43b7944a8e71faf65e9f836423c76b8d8a544b3fb56a1f254"}