{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:UUJI3VCD7C2B7TL7EFFQAJ6BIM","short_pith_number":"pith:UUJI3VCD","schema_version":"1.0","canonical_sha256":"a5128dd443f8b41fcd7f214b0027c1432852a98772cd0d9ff07c4f85d33c8cbd","source":{"kind":"arxiv","id":"2306.11695","version":3},"attestation_state":"computed","paper":{"title":"A Simple and Effective Pruning Approach for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Wanda prunes large language models by removing weights whose magnitudes times input activations are smallest, with no retraining required.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anna Bair, J. Zico Kolter, Mingjie Sun, Zhuang Liu","submitted_at":"2023-06-20T17:18:20Z","abstract_excerpt":"As their size increases, Large Languages Models (LLMs) are natural candidates for network pruning methods: approaches that drop a subset of network weights while striving to preserve performance. Existing methods, however, require either retraining, which is rarely affordable for billion-scale LLMs, or solving a weight reconstruction problem reliant on second-order information, which may also be computationally expensive. In this paper, we introduce a novel, straightforward yet effective pruning method, termed Wanda (Pruning by Weights and activations), designed to induce sparsity in pretraine"},"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":true},"canonical_record":{"source":{"id":"2306.11695","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-06-20T17:18:20Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"9b92b75b9a1fafea9f2ef9463f06a2857e2f8610d84409b6122816b94f657316","abstract_canon_sha256":"8bd7eefa2e525be9a8735c1df727f097e7d9a424979146ae0529209eab22af9e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:47.337163Z","signature_b64":"1DO01t+kNFYiimAF7CW+dfNqHLibUZ40OlsYIzo4lYbi52M4Lw9XXYtLQh9p8sTLLEd8MZ4u5EXVS1KaQ9wkAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a5128dd443f8b41fcd7f214b0027c1432852a98772cd0d9ff07c4f85d33c8cbd","last_reissued_at":"2026-05-17T23:38:47.336614Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:47.336614Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Simple and Effective Pruning Approach for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Wanda prunes large language models by removing weights whose magnitudes times input activations are smallest, with no retraining required.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anna Bair, J. Zico Kolter, Mingjie Sun, Zhuang Liu","submitted_at":"2023-06-20T17:18:20Z","abstract_excerpt":"As their size increases, Large Languages Models (LLMs) are natural candidates for network pruning methods: approaches that drop a subset of network weights while striving to preserve performance. Existing methods, however, require either retraining, which is rarely affordable for billion-scale LLMs, or solving a weight reconstruction problem reliant on second-order information, which may also be computationally expensive. In this paper, we introduce a novel, straightforward yet effective pruning method, termed Wanda (Pruning by Weights and activations), designed to induce sparsity in pretraine"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Wanda significantly outperforms the established baseline of magnitude pruning and performs competitively against recent methods involving intensive weight update on LLaMA and LLaMA-2 across various language benchmarks, with no retraining or weight update required.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The central assumption is that the product of weight magnitude and input activation reliably identifies weights whose removal will least affect model performance, motivated by emergent large-magnitude features but without a formal proof that this criterion is optimal or generalizes beyond the tested models.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Wanda prunes pretrained LLMs by dropping weights with smallest magnitude times activation values per output channel, with no retraining needed and better results than magnitude pruning on language benchmarks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Wanda prunes large language models by removing weights whose magnitudes times input activations are smallest, with no retraining required.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"71016ab615440db7754d7869f62659141955fc8c09bbc5ce2165b455abfe444f"},"source":{"id":"2306.11695","kind":"arxiv","version":3},"verdict":{"id":"7397a1c0-a27e-43f8-ac96-b7dd12a0aad6","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T16:08:25.620116Z","strongest_claim":"Wanda significantly outperforms the established baseline of magnitude pruning and performs competitively against recent methods involving intensive weight update on LLaMA and LLaMA-2 across various language benchmarks, with no retraining or weight update required.","one_line_summary":"Wanda prunes pretrained LLMs by dropping weights with smallest magnitude times activation values per output channel, with no retraining needed and better results than magnitude pruning on language benchmarks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The central assumption is that the product of weight magnitude and input activation reliably identifies weights whose removal will least affect model performance, motivated by emergent large-magnitude features but without a formal proof that this criterion is optimal or generalizes beyond the tested models.","pith_extraction_headline":"Wanda prunes large language models by removing weights whose magnitudes times input activations are smallest, with no retraining required."},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"068cdd3ac91ece8e10f3c596f5b95a706ee3ae5f90a5009f3368649bfbbe5656"},"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":"2306.11695","created_at":"2026-05-17T23:38:47.336678+00:00"},{"alias_kind":"arxiv_version","alias_value":"2306.11695v3","created_at":"2026-05-17T23:38:47.336678+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.11695","created_at":"2026-05-17T23:38:47.336678+00:00"},{"alias_kind":"pith_short_12","alias_value":"UUJI3VCD7C2B","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"UUJI3VCD7C2B7TL7","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"UUJI3VCD","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":43,"internal_anchor_count":43,"sample":[{"citing_arxiv_id":"2509.12958","citing_title":"Forget What's Sensitive, Remember What Matters: Token-Level Differential Privacy in Memory Sculpting for Continual Learning","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2310.02277","citing_title":"Junk DNA Hypothesis: Pruning Small Pre-Trained Weights Irreversibly and Monotonically Impairs \"Difficult\" Downstream Tasks in LLMs","ref_index":49,"is_internal_anchor":true},{"citing_arxiv_id":"2412.00069","citing_title":"Condense, Don't Just Prune: Enhancing Efficiency and Performance in MoE Layer Pruning","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2503.08223","citing_title":"Will LLMs Scaling Hit the Wall? Breaking Barriers via Distributed Resources on Massive Edge Devices","ref_index":168,"is_internal_anchor":true},{"citing_arxiv_id":"2505.12942","citing_title":"A3 : an Analytical Low-Rank Approximation Framework for Attention","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2505.16120","citing_title":"LLM-Powered AI Agent Systems and Their Applications in Industry","ref_index":88,"is_internal_anchor":true},{"citing_arxiv_id":"2505.17138","citing_title":"RAP: Runtime Adaptive Pruning for LLM Inference","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2512.12744","citing_title":"Resting Neurons, Active Insights: Robustifying Activation Sparsity in LLMs via Spontaneity","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14738","citing_title":"TAPIOCA: Why Task- Aware Pruning Improves OOD model Capability","ref_index":77,"is_internal_anchor":true},{"citing_arxiv_id":"2509.25606","citing_title":"Effective Model Pruning: Measure The Redundancy of Model Components","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10933","citing_title":"DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2307.06435","citing_title":"A Comprehensive Overview of Large Language Models","ref_index":265,"is_internal_anchor":true},{"citing_arxiv_id":"2506.12876","citing_title":"MaskPro: Linear-Space Probabilistic Learning for Strict (N:M)-Sparsity on LLMs","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2506.13727","citing_title":"Attribution-Guided Pruning for Insight and Control: Circuit Discovery and Targeted Correction in Small-scale LLMs","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2509.12464","citing_title":"Reasoning Models Can be Accurately Pruned Via Chain-of-Thought Reconstruction","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2509.22075","citing_title":"CoSpaDi: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2510.09696","citing_title":"Vanishing Contributions: A Unified Framework for Smooth and Iterative Model Compression","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2511.14582","citing_title":"OmniZip: Audio-Guided Dynamic Token Compression for Fast Omnimodal Large Language Models","ref_index":39,"is_internal_anchor":true},{"citing_arxiv_id":"2306.14048","citing_title":"H$_2$O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2511.22099","citing_title":"Decomposed Trust: Privacy, Adversarial Robustness, Ethics, and Fairness in Low-Rank LLMs","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2401.05459","citing_title":"Personal LLM Agents: Insights and Survey about the Capability, Efficiency and Security","ref_index":255,"is_internal_anchor":true},{"citing_arxiv_id":"2512.12744","citing_title":"Resting Neurons, Active Insights: Robustifying Activation Sparsity in LLMs via Spontaneity","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2512.22671","citing_title":"Fragile Knowledge, Robust Instruction-Following: The Width Pruning Dichotomy in Llama-3.2","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2601.05524","citing_title":"Double: Breaking the Acceleration Limit via Double Retrieval Speculative Parallelism","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2312.16886","citing_title":"MobileVLM : A Fast, Strong and Open Vision Language Assistant for Mobile Devices","ref_index":110,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM","json":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM.json","graph_json":"https://pith.science/api/pith-number/UUJI3VCD7C2B7TL7EFFQAJ6BIM/graph.json","events_json":"https://pith.science/api/pith-number/UUJI3VCD7C2B7TL7EFFQAJ6BIM/events.json","paper":"https://pith.science/paper/UUJI3VCD"},"agent_actions":{"view_html":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM","download_json":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM.json","view_paper":"https://pith.science/paper/UUJI3VCD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2306.11695&json=true","fetch_graph":"https://pith.science/api/pith-number/UUJI3VCD7C2B7TL7EFFQAJ6BIM/graph.json","fetch_events":"https://pith.science/api/pith-number/UUJI3VCD7C2B7TL7EFFQAJ6BIM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM/action/storage_attestation","attest_author":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM/action/author_attestation","sign_citation":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM/action/citation_signature","submit_replication":"https://pith.science/pith/UUJI3VCD7C2B7TL7EFFQAJ6BIM/action/replication_record"}},"created_at":"2026-05-17T23:38:47.336678+00:00","updated_at":"2026-05-17T23:38:47.336678+00:00"}