{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:C6RUBASQYQQFQ4DIJKLTWVQRZM","short_pith_number":"pith:C6RUBASQ","canonical_record":{"source":{"id":"2304.15010","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-04-28T17:59:25Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG","cs.MM"],"title_canon_sha256":"c3cc9569425a0dcc2b6287f8b1cc9dda45a8c9740ce9805e43e60a104fda39b6","abstract_canon_sha256":"44f870f9d7255528a47e73343d36a3f6ffeee3ff8d7d32938a2bb5d2da548e9a"},"schema_version":"1.0"},"canonical_sha256":"17a3408250c4205870684a973b5611cb1d7ab91dd3d6445d7076680005a5045b","source":{"kind":"arxiv","id":"2304.15010","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2304.15010","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"arxiv_version","alias_value":"2304.15010v1","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.15010","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"pith_short_12","alias_value":"C6RUBASQYQQF","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"C6RUBASQYQQFQ4DI","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"C6RUBASQ","created_at":"2026-05-18T12:33:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:C6RUBASQYQQFQ4DIJKLTWVQRZM","target":"record","payload":{"canonical_record":{"source":{"id":"2304.15010","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-04-28T17:59:25Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG","cs.MM"],"title_canon_sha256":"c3cc9569425a0dcc2b6287f8b1cc9dda45a8c9740ce9805e43e60a104fda39b6","abstract_canon_sha256":"44f870f9d7255528a47e73343d36a3f6ffeee3ff8d7d32938a2bb5d2da548e9a"},"schema_version":"1.0"},"canonical_sha256":"17a3408250c4205870684a973b5611cb1d7ab91dd3d6445d7076680005a5045b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:52.998450Z","signature_b64":"GfYysLXCjHQSnxbcZ/y5rdanhVl5tHtgYWmKtD2SGiQk9rBN4LDHK3GaoIyeWEjXB30vQKJprzHL+rSQuxvoAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"17a3408250c4205870684a973b5611cb1d7ab91dd3d6445d7076680005a5045b","last_reissued_at":"2026-05-17T23:38:52.997953Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:52.997953Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2304.15010","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:52Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1xtAtAu6+v+LkjZMcwBvLoyaxAqmioY/q7Vje3IB7p9JOD16dL0pdCoXPkVcpK8ka/pbTlbDF9uK1rVi0kuvCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T23:43:16.883069Z"},"content_sha256":"89a633b38f965df5b51561a859dde98fe72f53466d31fef29e2bbd4bd5d4ce08","schema_version":"1.0","event_id":"sha256:89a633b38f965df5b51561a859dde98fe72f53466d31fef29e2bbd4bd5d4ce08"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:C6RUBASQYQQFQ4DIJKLTWVQRZM","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"LLaMA-Adapter V2: Parameter-Efficient Visual Instruction Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"LLaMA-Adapter V2 turns LLaMA into an open-ended visual instruction follower by adding only 14 million parameters.","cross_cats":["cs.AI","cs.CL","cs.LG","cs.MM"],"primary_cat":"cs.CV","authors_text":"Aojun Zhou, Conghui He, Hongsheng Li, Jiaming Han, Pan Lu, Peng Gao, Renrui Zhang, Shijie Geng, Wei Zhang, Xiangyu Yue, Yu Qiao, Ziyi Lin","submitted_at":"2023-04-28T17:59:25Z","abstract_excerpt":"How to efficiently transform large language models (LLMs) into instruction followers is recently a popular research direction, while training LLM for multi-modal reasoning remains less explored. Although the recent LLaMA-Adapter demonstrates the potential to handle visual inputs with LLMs, it still cannot generalize well to open-ended visual instructions and lags behind GPT-4. In this paper, we present LLaMA-Adapter V2, a parameter-efficient visual instruction model. Specifically, we first augment LLaMA-Adapter by unlocking more learnable parameters (e.g., norm, bias and scale), which distribu"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Compared to the original LLaMA-Adapter, our LLaMA-Adapter V2 can perform open-ended multi-modal instructions by merely introducing 14M parameters over LLaMA.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the early-fusion placement and the disjoint-parameter joint training will continue to prevent task interference and maintain generalization when the instruction data distribution shifts or when larger base models are used.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LLaMA-Adapter V2 achieves open-ended visual instruction following in LLMs by unlocking more parameters, early fusion of visual tokens, and joint training on disjoint parameter groups with only 14M added parameters.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"LLaMA-Adapter V2 turns LLaMA into an open-ended visual instruction follower by adding only 14 million parameters.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c9d348bc3257a2da07569ba9ae3472017e71220e651ebc57c1ec811a410510be"},"source":{"id":"2304.15010","kind":"arxiv","version":1},"verdict":{"id":"4ed12321-98b1-4490-beda-aebc663e4e58","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T08:36:50.707814Z","strongest_claim":"Compared to the original LLaMA-Adapter, our LLaMA-Adapter V2 can perform open-ended multi-modal instructions by merely introducing 14M parameters over LLaMA.","one_line_summary":"LLaMA-Adapter V2 achieves open-ended visual instruction following in LLMs by unlocking more parameters, early fusion of visual tokens, and joint training on disjoint parameter groups with only 14M added parameters.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the early-fusion placement and the disjoint-parameter joint training will continue to prevent task interference and maintain generalization when the instruction data distribution shifts or when larger base models are used.","pith_extraction_headline":"LLaMA-Adapter V2 turns LLaMA into an open-ended visual instruction follower by adding only 14 million parameters."},"references":{"count":79,"sample":[{"doi":"","year":null,"title":"https://sharegpt.com/","work_id":"42263fdc-d42c-4b2a-8562-4f2dc47ecf6c","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Flamingo: a visual language model for few-shot learning","work_id":"31e3af5c-9fec-43d9-b533-5bb70172dd15","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Bottom-up and top-down attention for image captioning and visual question answering","work_id":"40ff759e-80ef-4b6f-86f5-11ea4321f5c8","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1901,"title":"Lan- guage models are few-shot learners","work_id":"5b23bebc-10b7-4150-9a97-e3f37825079e","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Conceptual 12m: Pushing web-scale image-text pre- training to recognize long-tail visual concepts","work_id":"c9a39a05-4f9a-45e3-92e2-f310468325af","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":79,"snapshot_sha256":"dd7cf17851379fca4c7b1bc3da5a359dac627288662817514af94575b0695681","internal_anchors":19},"formal_canon":{"evidence_count":2,"snapshot_sha256":"6b6edd9601be9795336ad80bbde7cdb959d9fbaefb4514280dc3923e30cfe686"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"4ed12321-98b1-4490-beda-aebc663e4e58"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:52Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UXdV0DHS24XaSsC5hUo7E93AeTK0lBYibl12FGOqFlYkHKNxHnftEvqukFtOBCxmwJXEvVVSBCyaKaUP/JT3Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T23:43:16.883675Z"},"content_sha256":"28a20f906dcb26de05f63e11759e383334abddd94f7bef1f69a3cb9a65bcd93e","schema_version":"1.0","event_id":"sha256:28a20f906dcb26de05f63e11759e383334abddd94f7bef1f69a3cb9a65bcd93e"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/C6RUBASQYQQFQ4DIJKLTWVQRZM/bundle.json","state_url":"https://pith.science/pith/C6RUBASQYQQFQ4DIJKLTWVQRZM/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/C6RUBASQYQQFQ4DIJKLTWVQRZM/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":"2026-05-27T23:43:16Z","links":{"resolver":"https://pith.science/pith/C6RUBASQYQQFQ4DIJKLTWVQRZM","bundle":"https://pith.science/pith/C6RUBASQYQQFQ4DIJKLTWVQRZM/bundle.json","state":"https://pith.science/pith/C6RUBASQYQQFQ4DIJKLTWVQRZM/state.json","well_known_bundle":"https://pith.science/.well-known/pith/C6RUBASQYQQFQ4DIJKLTWVQRZM/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:C6RUBASQYQQFQ4DIJKLTWVQRZM","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":"44f870f9d7255528a47e73343d36a3f6ffeee3ff8d7d32938a2bb5d2da548e9a","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":"2023-04-28T17:59:25Z","title_canon_sha256":"c3cc9569425a0dcc2b6287f8b1cc9dda45a8c9740ce9805e43e60a104fda39b6"},"schema_version":"1.0","source":{"id":"2304.15010","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2304.15010","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"arxiv_version","alias_value":"2304.15010v1","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.15010","created_at":"2026-05-17T23:38:52Z"},{"alias_kind":"pith_short_12","alias_value":"C6RUBASQYQQF","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"C6RUBASQYQQFQ4DI","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"C6RUBASQ","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:28a20f906dcb26de05f63e11759e383334abddd94f7bef1f69a3cb9a65bcd93e","target":"graph","created_at":"2026-05-17T23:38:52Z","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":"Compared to the original LLaMA-Adapter, our LLaMA-Adapter V2 can perform open-ended multi-modal instructions by merely introducing 14M parameters over LLaMA."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the early-fusion placement and the disjoint-parameter joint training will continue to prevent task interference and maintain generalization when the instruction data distribution shifts or when larger base models are used."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"LLaMA-Adapter V2 achieves open-ended visual instruction following in LLMs by unlocking more parameters, early fusion of visual tokens, and joint training on disjoint parameter groups with only 14M added parameters."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"LLaMA-Adapter V2 turns LLaMA into an open-ended visual instruction follower by adding only 14 million parameters."}],"snapshot_sha256":"c9d348bc3257a2da07569ba9ae3472017e71220e651ebc57c1ec811a410510be"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"6b6edd9601be9795336ad80bbde7cdb959d9fbaefb4514280dc3923e30cfe686"},"paper":{"abstract_excerpt":"How to efficiently transform large language models (LLMs) into instruction followers is recently a popular research direction, while training LLM for multi-modal reasoning remains less explored. Although the recent LLaMA-Adapter demonstrates the potential to handle visual inputs with LLMs, it still cannot generalize well to open-ended visual instructions and lags behind GPT-4. In this paper, we present LLaMA-Adapter V2, a parameter-efficient visual instruction model. Specifically, we first augment LLaMA-Adapter by unlocking more learnable parameters (e.g., norm, bias and scale), which distribu","authors_text":"Aojun Zhou, Conghui He, Hongsheng Li, Jiaming Han, Pan Lu, Peng Gao, Renrui Zhang, Shijie Geng, Wei Zhang, Xiangyu Yue, Yu Qiao, Ziyi Lin","cross_cats":["cs.AI","cs.CL","cs.LG","cs.MM"],"headline":"LLaMA-Adapter V2 turns LLaMA into an open-ended visual instruction follower by adding only 14 million parameters.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-04-28T17:59:25Z","title":"LLaMA-Adapter V2: Parameter-Efficient Visual Instruction Model"},"references":{"count":79,"internal_anchors":19,"resolved_work":79,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"https://sharegpt.com/","work_id":"42263fdc-d42c-4b2a-8562-4f2dc47ecf6c","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Flamingo: a visual language model for few-shot learning","work_id":"31e3af5c-9fec-43d9-b533-5bb70172dd15","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Bottom-up and top-down attention for image captioning and visual question answering","work_id":"40ff759e-80ef-4b6f-86f5-11ea4321f5c8","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Lan- guage models are few-shot learners","work_id":"5b23bebc-10b7-4150-9a97-e3f37825079e","year":1901},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Conceptual 12m: Pushing web-scale image-text pre- training to recognize long-tail visual concepts","work_id":"c9a39a05-4f9a-45e3-92e2-f310468325af","year":2021}],"snapshot_sha256":"dd7cf17851379fca4c7b1bc3da5a359dac627288662817514af94575b0695681"},"source":{"id":"2304.15010","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T08:36:50.707814Z","id":"4ed12321-98b1-4490-beda-aebc663e4e58","model_set":{"reader":"grok-4.3"},"one_line_summary":"LLaMA-Adapter V2 achieves open-ended visual instruction following in LLMs by unlocking more parameters, early fusion of visual tokens, and joint training on disjoint parameter groups with only 14M added parameters.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"LLaMA-Adapter V2 turns LLaMA into an open-ended visual instruction follower by adding only 14 million parameters.","strongest_claim":"Compared to the original LLaMA-Adapter, our LLaMA-Adapter V2 can perform open-ended multi-modal instructions by merely introducing 14M parameters over LLaMA.","weakest_assumption":"That the early-fusion placement and the disjoint-parameter joint training will continue to prevent task interference and maintain generalization when the instruction data distribution shifts or when larger base models are used."}},"verdict_id":"4ed12321-98b1-4490-beda-aebc663e4e58"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:89a633b38f965df5b51561a859dde98fe72f53466d31fef29e2bbd4bd5d4ce08","target":"record","created_at":"2026-05-17T23:38:52Z","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":"44f870f9d7255528a47e73343d36a3f6ffeee3ff8d7d32938a2bb5d2da548e9a","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":"2023-04-28T17:59:25Z","title_canon_sha256":"c3cc9569425a0dcc2b6287f8b1cc9dda45a8c9740ce9805e43e60a104fda39b6"},"schema_version":"1.0","source":{"id":"2304.15010","kind":"arxiv","version":1}},"canonical_sha256":"17a3408250c4205870684a973b5611cb1d7ab91dd3d6445d7076680005a5045b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"17a3408250c4205870684a973b5611cb1d7ab91dd3d6445d7076680005a5045b","first_computed_at":"2026-05-17T23:38:52.997953Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:52.997953Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"GfYysLXCjHQSnxbcZ/y5rdanhVl5tHtgYWmKtD2SGiQk9rBN4LDHK3GaoIyeWEjXB30vQKJprzHL+rSQuxvoAw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:52.998450Z","signed_message":"canonical_sha256_bytes"},"source_id":"2304.15010","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:89a633b38f965df5b51561a859dde98fe72f53466d31fef29e2bbd4bd5d4ce08","sha256:28a20f906dcb26de05f63e11759e383334abddd94f7bef1f69a3cb9a65bcd93e"],"state_sha256":"8e6c593da19e088a00120d7d573cb3e37586bb3f38c24289b04f371a1df797ef"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ywS6ExDPhkSR7uEhJtpbgzbja3tPPO5tKxr5Ti61mNaJLduYOh8FXwWUXBVRF5MbvlbqivQE7uhJesVbK0bTBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-27T23:43:16.887248Z","bundle_sha256":"e3eb9610d731e1a69c64ea5e59aada47eb5d3d7487eef47821cabf4a92eb09c5"}}