{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/massive-language-models-can-be-accurately","title":"SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot","arxiv_id":"2301.00774","date":"2023-01-02","proceeding":null,"authors":["Elias Frantar","Dan Alistarh"],"abstract":"We show for the first time that large-scale generative pretrained transformer (GPT) family models can be pruned to at least 50% sparsity in one-shot, without any retraining, at minimal loss of accuracy. This is achieved via a new pruning method called SparseGPT, specifically designed to work efficiently and accurately on massive GPT-family models. We can execute SparseGPT on the largest available open-source models, OPT-175B and BLOOM-176B, in under 4.5 hours, and can reach 60% unstructured sparsity with negligible increase in perplexity: remarkably, more than 100 billion weights from these models can be ignored at inference time. SparseGPT generalizes to semi-structured (2:4 and 4:8) patterns, and is compatible with weight quantization approaches. The code is available at: https://github.com/IST-DASLab/sparsegpt.","url_abs":"https://arxiv.org/abs/2301.00774v3","url_pdf":"https://arxiv.org/pdf/2301.00774v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"massive-language-models-can-be-accurately","repo_url":"https://github.com/ist-daslab/sparsegpt","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"massive-language-models-can-be-accurately","repo_url":"https://github.com/baithebest/adagp","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"massive-language-models-can-be-accurately","repo_url":"https://github.com/baithebest/sparsellm","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"massive-language-models-can-be-accurately","repo_url":"https://github.com/eth-easl/deltazip","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"massive-language-models-can-be-accurately","repo_url":"https://github.com/nvidia/tensorrt-model-optimizer","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"massive-language-models-can-be-accurately","repo_url":"https://github.com/nvlabs/maskllm","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"common-sense-reasoning","task_name":"Common Sense Reasoning"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"quantization","task_name":"Quantization"},{"task_slug":"question-answering","task_name":"Question Answering"}],"methods":[{"method_slug":"pruning","method_name":"Pruning"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/common-sense-reasoning-on-arc-challenge","task":"Common Sense Reasoning","dataset":"ARC (Challenge)","model":"OPT-175B","rank_in_archive_order":44,"of":54,"metrics":{"Accuracy":"43.94"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-challenge","task":"Common Sense Reasoning","dataset":"ARC (Challenge)","model":"SparseGPT (175B, 50% Sparsity)","rank_in_archive_order":46,"of":54,"metrics":{"Accuracy":"41.3"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-challenge","task":"Common Sense Reasoning","dataset":"ARC (Challenge)","model":"SparseGPT (175B, 4:8 Sparsity)","rank_in_archive_order":47,"of":54,"metrics":{"Accuracy":"39.85"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-challenge","task":"Common Sense Reasoning","dataset":"ARC (Challenge)","model":"SparseGPT (175B, 2:4 Sparsity)","rank_in_archive_order":48,"of":54,"metrics":{"Accuracy":"38.99"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-challenge","task":"Common Sense Reasoning","dataset":"ARC (Challenge)","model":"OPT-175B (50% Sparsity)","rank_in_archive_order":54,"of":54,"metrics":{"Accuracy":"25.6"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-easy","task":"Common Sense Reasoning","dataset":"ARC (Easy)","model":"OPT-175B","rank_in_archive_order":31,"of":47,"metrics":{"Accuracy":"71.04"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-easy","task":"Common Sense Reasoning","dataset":"ARC (Easy)","model":"SparseGPT 175B (50% sparsity)","rank_in_archive_order":36,"of":47,"metrics":{"Accuracy":"69.65"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-easy","task":"Common Sense Reasoning","dataset":"ARC (Easy)","model":"SparseGPT (175B, 4:8 Sparsity)","rank_in_archive_order":39,"of":47,"metrics":{"Accuracy":"68.35"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-easy","task":"Common Sense Reasoning","dataset":"ARC (Easy)","model":"SparseGPT 175B (2:4 sparsity)","rank_in_archive_order":41,"of":47,"metrics":{"Accuracy":"67.08"},"uses_additional_data":false},{"leaderboard":"/sota/common-sense-reasoning-on-arc-easy","task":"Common Sense Reasoning","dataset":"ARC (Easy)","model":"OPT 175B (50% Sparsity)","rank_in_archive_order":47,"of":47,"metrics":{"Accuracy":"28.03"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-lambada","task":"Language Modelling","dataset":"LAMBADA","model":"SparseGPT (175B, 2:4 Sparsity)","rank_in_archive_order":13,"of":37,"metrics":{"Accuracy":"79.47"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-lambada","task":"Language Modelling","dataset":"LAMBADA","model":"SparseGPT (175B, 4:8 Sparsity)","rank_in_archive_order":14,"of":37,"metrics":{"Accuracy":"78.77"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-lambada","task":"Language Modelling","dataset":"LAMBADA","model":"SparseGPT (175B, 50% Sparsity)","rank_in_archive_order":17,"of":37,"metrics":{"Accuracy":"76.51"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-lambada","task":"Language Modelling","dataset":"LAMBADA","model":"OPT-175B","rank_in_archive_order":19,"of":37,"metrics":{"Accuracy":"75.59"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-lambada","task":"Language Modelling","dataset":"LAMBADA","model":"OPT-175B (50% Sparsity)","rank_in_archive_order":33,"of":37,"metrics":{"Accuracy":"0.02"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-wikitext-2","task":"Language Modelling","dataset":"WikiText-2","model":"SparseGPT (175B, 50% Sparsity)","rank_in_archive_order":1,"of":38,"metrics":{"Test perplexity":"8.21"},"uses_additional_data":true},{"leaderboard":"/sota/language-modelling-on-wikitext-2","task":"Language Modelling","dataset":"WikiText-2","model":"OPT-175B","rank_in_archive_order":2,"of":38,"metrics":{"Test perplexity":"8.34"},"uses_additional_data":true},{"leaderboard":"/sota/language-modelling-on-wikitext-2","task":"Language Modelling","dataset":"WikiText-2","model":"SparseGPT (175B, 4:8 Sparsity)","rank_in_archive_order":3,"of":38,"metrics":{"Test perplexity":"8.45"},"uses_additional_data":true},{"leaderboard":"/sota/language-modelling-on-wikitext-2","task":"Language Modelling","dataset":"WikiText-2","model":"SparseGPT (175B, 2:4 Sparsity)","rank_in_archive_order":4,"of":38,"metrics":{"Test perplexity":"8.73"},"uses_additional_data":true},{"leaderboard":"/sota/language-modelling-on-wikitext-2","task":"Language Modelling","dataset":"WikiText-2","model":"OPT-175B (50% Sparsity)","rank_in_archive_order":38,"of":38,"metrics":{"Test perplexity":"234.77"},"uses_additional_data":true},{"leaderboard":"/sota/question-answering-on-piqa","task":"Question Answering","dataset":"PIQA","model":"OPT-175B","rank_in_archive_order":28,"of":67,"metrics":{"Accuracy":"81.07"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-piqa","task":"Question Answering","dataset":"PIQA","model":"SparseGPT 175B (50% Sparsity)","rank_in_archive_order":30,"of":67,"metrics":{"Accuracy":"80.63"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-piqa","task":"Question Answering","dataset":"PIQA","model":"SparseGPT 175B (4:8 Sparsity)","rank_in_archive_order":35,"of":67,"metrics":{"Accuracy":"79.54"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-piqa","task":"Question Answering","dataset":"PIQA","model":"SparseGPT 175B (2:4 Sparsity)","rank_in_archive_order":36,"of":67,"metrics":{"Accuracy":"79.54"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-piqa","task":"Question Answering","dataset":"PIQA","model":"OPT-175B (50% Sparsity)","rank_in_archive_order":66,"of":67,"metrics":{"Accuracy":"54.73"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-storycloze","task":"Question Answering","dataset":"StoryCloze","model":"OPT-175B","rank_in_archive_order":10,"of":23,"metrics":{"Accuracy":"79.82"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-storycloze","task":"Question Answering","dataset":"StoryCloze","model":"SparseGPT (175B, 50% Sparsity)","rank_in_archive_order":11,"of":23,"metrics":{"Accuracy":"78.87"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-storycloze","task":"Question Answering","dataset":"StoryCloze","model":"SparseGPT (175B, 4:8 Sparsity)","rank_in_archive_order":14,"of":23,"metrics":{"Accuracy":"77.02"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-storycloze","task":"Question Answering","dataset":"StoryCloze","model":"SparseGPT (175B, 2:4 Sparsity)","rank_in_archive_order":16,"of":23,"metrics":{"Accuracy":"76.19"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-storycloze","task":"Question Answering","dataset":"StoryCloze","model":"OPT-175B (50% Sparsity)","rank_in_archive_order":23,"of":23,"metrics":{"Accuracy":"47.10"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2301.00774","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2301.00774"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/baithebest/sparsellm","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/baithebest/adagp","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/eth-easl/deltazip","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nvlabs/maskllm","reach":{"status":"ok","spdx":"NOASSERTION"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ist-daslab/sparsegpt","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nvidia/tensorrt-model-optimizer","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"ran_draft_wrong":2,"unverified":10},"by_repo_kind":{"official":{"samples":3,"ran":1,"repositories":1},"listed":{"samples":8,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":9,"samples":[{"code_sha256_prefix":"cd666c16d4aaa66d","entry":"get_bloom","repo":"ist-daslab/sparsegpt","repo_kind":"official","path":"bloom.py","file_url":"https://github.com/ist-daslab/sparsegpt/blob/HEAD/bloom.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"cd666c16d4aaa66d"}},{"code_sha256_prefix":"887f43fe08a6472d","entry":"get_opt","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"887f43fe08a6472d"}},{"code_sha256_prefix":"5076786563dd8499","entry":"bloom_sequential","repo":"ist-daslab/sparsegpt","repo_kind":"official","path":"bloom.py","file_url":"https://github.com/ist-daslab/sparsegpt/blob/HEAD/bloom.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"5076786563dd8499"}},{"code_sha256_prefix":"f1c26f01bf2f39a9","entry":"get_language_model_from_vl","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/export/model_utils.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/export/model_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"f1c26f01bf2f39a9"}},{"code_sha256_prefix":"bc6cfb190b953479","entry":"get_model_type","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/export/model_utils.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/export/model_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"bc6cfb190b953479"}},{"code_sha256_prefix":"140c6ce5087cbbce","entry":"is_multimodal_model","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/export/model_utils.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/export/model_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"140c6ce5087cbbce"}},{"code_sha256_prefix":"bf403ae47e912dc1","entry":"opt_sequential","repo":"ist-daslab/sparsegpt","repo_kind":"official","path":"opt.py","file_url":"https://github.com/ist-daslab/sparsegpt/blob/HEAD/opt.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"bf403ae47e912dc1"}},{"code_sha256_prefix":"f36d24bde785f7cc","entry":"pad_weights","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/export/trtllm/model_config_utils.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/export/trtllm/model_config_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"f36d24bde785f7cc"}},{"code_sha256_prefix":"1c19ab0c6ad9a45c","entry":"postprocess_amax","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/quantization/model_quant.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/quantization/model_quant.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"1c19ab0c6ad9a45c"}},{"code_sha256_prefix":"e9eca650c1a383f2","entry":"prepare_enc_dec_export_dir","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/export/trtllm/tensorrt_llm_utils.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/export/trtllm/tensorrt_llm_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"e9eca650c1a383f2"}},{"code_sha256_prefix":"e3133ddfe52e03a0","entry":"student_input_bypass_fwd_hook","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/distill/layerwise_distillation_model.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/distill/layerwise_distillation_model.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"e3133ddfe52e03a0"}},{"code_sha256_prefix":"54d01cbd62cf758a","entry":"svd","repo":"nvidia/tensorrt-model-optimizer","repo_kind":"listed","path":"modelopt/torch/quantization/model_calib.py","file_url":"https://github.com/nvidia/tensorrt-model-optimizer/blob/HEAD/modelopt/torch/quantization/model_calib.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":false,"mcp_get_code":{"code_sha256":"54d01cbd62cf758a"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}