{"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/microscaling-data-formats-for-deep-learning","title":"Microscaling Data Formats for Deep Learning","arxiv_id":"2310.10537","date":"2023-10-16","proceeding":null,"authors":["Bita Darvish Rouhani","Ritchie Zhao","Ankit More","Mathew Hall","Alireza Khodamoradi","Summer Deng","Dhruv Choudhary","Marius Cornea","Eric Dellinger","Kristof Denolf","Stosic Dusan","Venmugil Elango","Maximilian Golub","Alexander Heinecke","Phil James-Roxby","Dharmesh Jani","Gaurav Kolhe","Martin Langhammer","Ada Li","Levi Melnick","Maral Mesmakhosroshahi","Andres Rodriguez","Michael Schulte","Rasoul Shafipour","Lei Shao","Michael Siu","Pradeep Dubey","Paulius Micikevicius","Maxim Naumov","Colin Verrilli","Ralph Wittig","Doug Burger","Eric Chung"],"abstract":"Narrow bit-width data formats are key to reducing the computational and storage costs of modern deep learning applications. This paper evaluates Microscaling (MX) data formats that combine a per-block scaling factor with narrow floating-point and integer types for individual elements. MX formats balance the competing needs of hardware efficiency, model accuracy, and user friction. Empirical results on over two dozen benchmarks demonstrate practicality of MX data formats as a drop-in replacement for baseline FP32 for AI inference and training with low user friction. We also show the first instance of training generative language models at sub-8-bit weights, activations, and gradients with minimal accuracy loss and no modifications to the training recipe.","url_abs":"https://arxiv.org/abs/2310.10537v3","url_pdf":"https://arxiv.org/pdf/2310.10537v3.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":"microscaling-data-formats-for-deep-learning","repo_url":"https://github.com/microsoft/microxcaling","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"microscaling-data-formats-for-deep-learning","repo_url":"https://github.com/rocm/tensorcast","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"friction","task_name":"Friction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2310.10537","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2310.10537"}},"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/rocm/tensorcast","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/microsoft/microxcaling","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":3,"unverified":15},"by_repo_kind":{"official":{"samples":15,"ran":3,"repositories":1},"listed":{"samples":3,"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":0,"samples":[{"code_sha256_prefix":"165ecb97772b0c77","entry":"conv_weight","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/convolution.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/convolution.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"165ecb97772b0c77"}},{"code_sha256_prefix":"c99ae440924911d3","entry":"end_index","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/adaptive_avg_pooling.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/adaptive_avg_pooling.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"c99ae440924911d3"}},{"code_sha256_prefix":"910795c444d0d80e","entry":"start_index","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/adaptive_avg_pooling.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/adaptive_avg_pooling.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"910795c444d0d80e"}},{"code_sha256_prefix":"5d134e37b84f1a48","entry":"adaptive_avg_pool2d","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/adaptive_avg_pooling.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/adaptive_avg_pooling.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5d134e37b84f1a48"}},{"code_sha256_prefix":"d3704f66f62c1c7c","entry":"batch_norm","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/batchnorm.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/batchnorm.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"d3704f66f62c1c7c"}},{"code_sha256_prefix":"b570ce1bd30b22ae","entry":"bmm","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/bmm.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/bmm.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"b570ce1bd30b22ae"}},{"code_sha256_prefix":"4411ca6f7e032435","entry":"check_literal","repo":"rocm/tensorcast","repo_kind":"listed","path":"tcast/utils.py","file_url":"https://github.com/rocm/tensorcast/blob/HEAD/tcast/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4411ca6f7e032435"}},{"code_sha256_prefix":"a43498acc45e8747","entry":"conv1d","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/convolution.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/convolution.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"a43498acc45e8747"}},{"code_sha256_prefix":"a64d28074918da0a","entry":"conv2d","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/convolution.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/convolution.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"a64d28074918da0a"}},{"code_sha256_prefix":"1a3253edf2298c8a","entry":"group_norm","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/groupnorm.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/groupnorm.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1a3253edf2298c8a"}},{"code_sha256_prefix":"9adc30f9c45093d8","entry":"is_installed","repo":"rocm/tensorcast","repo_kind":"listed","path":"tcast/utils.py","file_url":"https://github.com/rocm/tensorcast/blob/HEAD/tcast/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"9adc30f9c45093d8"}},{"code_sha256_prefix":"401cf9eddca5daee","entry":"is_power_of_2","repo":"rocm/tensorcast","repo_kind":"listed","path":"tcast/utils.py","file_url":"https://github.com/rocm/tensorcast/blob/HEAD/tcast/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"401cf9eddca5daee"}},{"code_sha256_prefix":"494a1f0c452b05c6","entry":"layer_norm","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/layernorm.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/layernorm.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"494a1f0c452b05c6"}},{"code_sha256_prefix":"5ad2a3b676e12813","entry":"linear","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/linear.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/linear.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5ad2a3b676e12813"}},{"code_sha256_prefix":"e71c7c6674717c77","entry":"quantize_elemwise_op","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/elemwise_ops.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/elemwise_ops.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"e71c7c6674717c77"}},{"code_sha256_prefix":"4490c20d2c5ae00e","entry":"relu","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/activations.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/activations.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4490c20d2c5ae00e"}},{"code_sha256_prefix":"bb051d6d1c91807f","entry":"sigmoid","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/activations.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/activations.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"bb051d6d1c91807f"}},{"code_sha256_prefix":"977b4767b78121d5","entry":"tanh","repo":"microsoft/microxcaling","repo_kind":"official","path":"mx/activations.py","file_url":"https://github.com/microsoft/microxcaling/blob/HEAD/mx/activations.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"977b4767b78121d5"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}