{"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/pact-parameterized-clipping-activation-for","title":"PACT: Parameterized Clipping Activation for Quantized Neural Networks","arxiv_id":"1805.06085","date":"2018-05-16","proceeding":"ICLR 2018 1","authors":["Jungwook Choi","Zhuo Wang","Swagath Venkataramani","Pierce I-Jen Chuang","Vijayalakshmi Srinivasan","Kailash Gopalakrishnan"],"abstract":"Deep learning algorithms achieve high classification accuracy at the expense\nof significant computation cost. To address this cost, a number of quantization\nschemes have been proposed - but most of these techniques focused on quantizing\nweights, which are relatively smaller in size compared to activations. This\npaper proposes a novel quantization scheme for activations during training -\nthat enables neural networks to work well with ultra low precision weights and\nactivations without any significant accuracy degradation. This technique,\nPArameterized Clipping acTivation (PACT), uses an activation clipping parameter\n$\\alpha$ that is optimized during training to find the right quantization\nscale. PACT allows quantizing activations to arbitrary bit precisions, while\nachieving much better accuracy relative to published state-of-the-art\nquantization schemes. We show, for the first time, that both weights and\nactivations can be quantized to 4-bits of precision while still achieving\naccuracy comparable to full precision networks across a range of popular models\nand datasets. We also show that exploiting these reduced-precision\ncomputational units in hardware can enable a super-linear improvement in\ninferencing performance due to a significant reduction in the area of\naccelerator compute engines coupled with the ability to retain the quantized\nmodel and activation data in on-chip memories.","url_abs":"http://arxiv.org/abs/1805.06085v2","url_pdf":"http://arxiv.org/pdf/1805.06085v2.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":"pact-parameterized-clipping-activation-for","repo_url":"https://github.com/KwangHoonAn/PACT","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pact-parameterized-clipping-activation-for","repo_url":"https://github.com/cornell-zhang/dnn-gating","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pact-parameterized-clipping-activation-for","repo_url":"https://github.com/PaddlePaddle/PaddleOCR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"paddle","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"quantization","task_name":"Quantization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1805.06085","atlas_url":"https://app.syntology.ai/?focus=1805.06085","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1805.06085"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+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/PaddlePaddle/PaddleOCR","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/KwangHoonAn/PACT","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/cornell-zhang/dnn-gating","reach":null}],"summary":{"unverified":2},"by_repo_kind":{"listed":{"samples":1,"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":1,"samples":[{"code_sha256_prefix":"f0c9a29156911331","entry":"accuracy","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"f0c9a29156911331"}},{"code_sha256_prefix":"b1460631afc194fb","entry":"validate","repo":"KwangHoonAn/PACT","repo_kind":"listed","path":"trainer.py","file_url":"https://github.com/KwangHoonAn/PACT/blob/HEAD/trainer.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":"b1460631afc194fb"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}