ELECTRONIC DEVICE AND NEURAL NETWORK QUANTIZATION METHOD
Embodiments of the present application provide an electronic device and a neural network quantization method. The electronic device comprises a processor and a logic circuit. The processor is used to determine a first zero offset and a first quantization coefficient according to floating point data,...
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Patent |
| Sprache: | chi ; eng ; fre |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | ZHANG, Yihao LIU, Genshu XIAO, Yannan ZUO, Wenming |
| description | Embodiments of the present application provide an electronic device and a neural network quantization method. The electronic device comprises a processor and a logic circuit. The processor is used to determine a first zero offset and a first quantization coefficient according to floating point data, and a maximum value and minimum value of preset fixed point data, multiply the first quantization coefficient to obtain a second quantization coefficient, and multiply the first zero offset to obtain a second zero offset. The logic circuit is used to quantize data to be quantized by means of floating point multiplication and fixed point addition according to the second quantization coefficient and the second zero point offset, to obtain a first quantization result; and shift the first quantization result according to a preset fixed-point quantization coefficient to obtain a final quantization result of the data to be quantized. In the electronic device and the neural network quantization method of the embodiments |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_WO2023004799A1</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>WO2023004799A1</sourcerecordid><originalsourceid>FETCH-epo_espacenet_WO2023004799A13</originalsourceid><addsrcrecordid>eNrjZLBw9XF1Dgny9_N0VnBxDfN0dlVw9HNR8HMNDXL0AVIh4f5B3gqBoY5-IZ5RjiGe_n4Kvq4hHv4uPAysaYk5xam8UJqbQdnNNcTZQze1ID8-tbggMTk1L7UkPtzfyMDI2MDAxNzS0tHQmDhVABeIKZs</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>ELECTRONIC DEVICE AND NEURAL NETWORK QUANTIZATION METHOD</title><source>esp@cenet</source><creator>ZHANG, Yihao ; LIU, Genshu ; XIAO, Yannan ; ZUO, Wenming</creator><creatorcontrib>ZHANG, Yihao ; LIU, Genshu ; XIAO, Yannan ; ZUO, Wenming</creatorcontrib><description>Embodiments of the present application provide an electronic device and a neural network quantization method. The electronic device comprises a processor and a logic circuit. The processor is used to determine a first zero offset and a first quantization coefficient according to floating point data, and a maximum value and minimum value of preset fixed point data, multiply the first quantization coefficient to obtain a second quantization coefficient, and multiply the first zero offset to obtain a second zero offset. The logic circuit is used to quantize data to be quantized by means of floating point multiplication and fixed point addition according to the second quantization coefficient and the second zero point offset, to obtain a first quantization result; and shift the first quantization result according to a preset fixed-point quantization coefficient to obtain a final quantization result of the data to be quantized. In the electronic device and the neural network quantization method of the embodiments</description><language>chi ; eng ; fre</language><subject>CALCULATING ; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS ; COMPUTING ; COUNTING ; PHYSICS</subject><creationdate>2023</creationdate><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20230202&DB=EPODOC&CC=WO&NR=2023004799A1$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,780,885,25563,76418</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20230202&DB=EPODOC&CC=WO&NR=2023004799A1$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>ZHANG, Yihao</creatorcontrib><creatorcontrib>LIU, Genshu</creatorcontrib><creatorcontrib>XIAO, Yannan</creatorcontrib><creatorcontrib>ZUO, Wenming</creatorcontrib><title>ELECTRONIC DEVICE AND NEURAL NETWORK QUANTIZATION METHOD</title><description>Embodiments of the present application provide an electronic device and a neural network quantization method. The electronic device comprises a processor and a logic circuit. The processor is used to determine a first zero offset and a first quantization coefficient according to floating point data, and a maximum value and minimum value of preset fixed point data, multiply the first quantization coefficient to obtain a second quantization coefficient, and multiply the first zero offset to obtain a second zero offset. The logic circuit is used to quantize data to be quantized by means of floating point multiplication and fixed point addition according to the second quantization coefficient and the second zero point offset, to obtain a first quantization result; and shift the first quantization result according to a preset fixed-point quantization coefficient to obtain a final quantization result of the data to be quantized. In the electronic device and the neural network quantization method of the embodiments</description><subject>CALCULATING</subject><subject>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</subject><subject>COMPUTING</subject><subject>COUNTING</subject><subject>PHYSICS</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2023</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNrjZLBw9XF1Dgny9_N0VnBxDfN0dlVw9HNR8HMNDXL0AVIh4f5B3gqBoY5-IZ5RjiGe_n4Kvq4hHv4uPAysaYk5xam8UJqbQdnNNcTZQze1ID8-tbggMTk1L7UkPtzfyMDI2MDAxNzS0tHQmDhVABeIKZs</recordid><startdate>20230202</startdate><enddate>20230202</enddate><creator>ZHANG, Yihao</creator><creator>LIU, Genshu</creator><creator>XIAO, Yannan</creator><creator>ZUO, Wenming</creator><scope>EVB</scope></search><sort><creationdate>20230202</creationdate><title>ELECTRONIC DEVICE AND NEURAL NETWORK QUANTIZATION METHOD</title><author>ZHANG, Yihao ; LIU, Genshu ; XIAO, Yannan ; ZUO, Wenming</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_WO2023004799A13</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>chi ; eng ; fre</language><creationdate>2023</creationdate><topic>CALCULATING</topic><topic>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</topic><topic>COMPUTING</topic><topic>COUNTING</topic><topic>PHYSICS</topic><toplevel>online_resources</toplevel><creatorcontrib>ZHANG, Yihao</creatorcontrib><creatorcontrib>LIU, Genshu</creatorcontrib><creatorcontrib>XIAO, Yannan</creatorcontrib><creatorcontrib>ZUO, Wenming</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>ZHANG, Yihao</au><au>LIU, Genshu</au><au>XIAO, Yannan</au><au>ZUO, Wenming</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>ELECTRONIC DEVICE AND NEURAL NETWORK QUANTIZATION METHOD</title><date>2023-02-02</date><risdate>2023</risdate><abstract>Embodiments of the present application provide an electronic device and a neural network quantization method. The electronic device comprises a processor and a logic circuit. The processor is used to determine a first zero offset and a first quantization coefficient according to floating point data, and a maximum value and minimum value of preset fixed point data, multiply the first quantization coefficient to obtain a second quantization coefficient, and multiply the first zero offset to obtain a second zero offset. The logic circuit is used to quantize data to be quantized by means of floating point multiplication and fixed point addition according to the second quantization coefficient and the second zero point offset, to obtain a first quantization result; and shift the first quantization result according to a preset fixed-point quantization coefficient to obtain a final quantization result of the data to be quantized. In the electronic device and the neural network quantization method of the embodiments</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | chi ; eng ; fre |
| recordid | cdi_epo_espacenet_WO2023004799A1 |
| source | esp@cenet |
| subjects | CALCULATING COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS COMPUTING COUNTING PHYSICS |
| title | ELECTRONIC DEVICE AND NEURAL NETWORK QUANTIZATION METHOD |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-08T18%3A39%3A02IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-epo_EVB&rft_val_fmt=info:ofi/fmt:kev:mtx:patent&rft.genre=patent&rft.au=ZHANG,%20Yihao&rft.date=2023-02-02&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3EWO2023004799A1%3C/epo_EVB%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |