Learned step size quantization
Learned step size quantization in artificial neural network is provided. In various embodiments, a system comprises an artificial neural network and a computing node. The artificial neural network comprises: a quantizer having a configurable step size, the quantizer adapted to receive a plurality of...
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Patent |
| Sprache: | eng |
| 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 | Appuswamy, Rathinakumar McKinstry, Jeffrey L Esser, Steve Bablani, Deepika Modha, Dharmendra S |
| description | Learned step size quantization in artificial neural network is provided. In various embodiments, a system comprises an artificial neural network and a computing node. The artificial neural network comprises: a quantizer having a configurable step size, the quantizer adapted to receive a plurality of input values and quantize the plurality of input values according to the configurable step size to produce a plurality of quantized input values, at least one matrix multiplier configured to receive the plurality of quantized input values from the quantizer and to apply a plurality of weights to the quantized input values to determine a plurality of output values having a first precision, and a multiplier configured to scale the output values to a second precision. The computing node is operatively coupled to the artificial neural network and is configured to: provide training input data to the artificial neural network, and optimize the configurable step size based on a gradient through the quantizer and the training input data. |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_US11823054B2</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>US11823054B2</sourcerecordid><originalsourceid>FETCH-epo_espacenet_US11823054B23</originalsourceid><addsrcrecordid>eNrjZJDzSU0syktNUSguSS1QKM6sSlUoLE3MK8msSizJzM_jYWBNS8wpTuWF0twMim6uIc4euqkF-fGpxQWJyal5qSXxocGGhhZGxgamJk5GxsSoAQASiiTi</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>Learned step size quantization</title><source>esp@cenet</source><creator>Appuswamy, Rathinakumar ; McKinstry, Jeffrey L ; Esser, Steve ; Bablani, Deepika ; Modha, Dharmendra S</creator><creatorcontrib>Appuswamy, Rathinakumar ; McKinstry, Jeffrey L ; Esser, Steve ; Bablani, Deepika ; Modha, Dharmendra S</creatorcontrib><description>Learned step size quantization in artificial neural network is provided. In various embodiments, a system comprises an artificial neural network and a computing node. The artificial neural network comprises: a quantizer having a configurable step size, the quantizer adapted to receive a plurality of input values and quantize the plurality of input values according to the configurable step size to produce a plurality of quantized input values, at least one matrix multiplier configured to receive the plurality of quantized input values from the quantizer and to apply a plurality of weights to the quantized input values to determine a plurality of output values having a first precision, and a multiplier configured to scale the output values to a second precision. The computing node is operatively coupled to the artificial neural network and is configured to: provide training input data to the artificial neural network, and optimize the configurable step size based on a gradient through the quantizer and the training input data.</description><language>eng</language><subject>CALCULATING ; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS ; COMPUTING ; COUNTING ; ELECTRIC DIGITAL DATA PROCESSING ; 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=20231121&DB=EPODOC&CC=US&NR=11823054B2$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,776,881,25542,76290</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20231121&DB=EPODOC&CC=US&NR=11823054B2$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>Appuswamy, Rathinakumar</creatorcontrib><creatorcontrib>McKinstry, Jeffrey L</creatorcontrib><creatorcontrib>Esser, Steve</creatorcontrib><creatorcontrib>Bablani, Deepika</creatorcontrib><creatorcontrib>Modha, Dharmendra S</creatorcontrib><title>Learned step size quantization</title><description>Learned step size quantization in artificial neural network is provided. In various embodiments, a system comprises an artificial neural network and a computing node. The artificial neural network comprises: a quantizer having a configurable step size, the quantizer adapted to receive a plurality of input values and quantize the plurality of input values according to the configurable step size to produce a plurality of quantized input values, at least one matrix multiplier configured to receive the plurality of quantized input values from the quantizer and to apply a plurality of weights to the quantized input values to determine a plurality of output values having a first precision, and a multiplier configured to scale the output values to a second precision. The computing node is operatively coupled to the artificial neural network and is configured to: provide training input data to the artificial neural network, and optimize the configurable step size based on a gradient through the quantizer and the training input data.</description><subject>CALCULATING</subject><subject>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</subject><subject>COMPUTING</subject><subject>COUNTING</subject><subject>ELECTRIC DIGITAL DATA PROCESSING</subject><subject>PHYSICS</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2023</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNrjZJDzSU0syktNUSguSS1QKM6sSlUoLE3MK8msSizJzM_jYWBNS8wpTuWF0twMim6uIc4euqkF-fGpxQWJyal5qSXxocGGhhZGxgamJk5GxsSoAQASiiTi</recordid><startdate>20231121</startdate><enddate>20231121</enddate><creator>Appuswamy, Rathinakumar</creator><creator>McKinstry, Jeffrey L</creator><creator>Esser, Steve</creator><creator>Bablani, Deepika</creator><creator>Modha, Dharmendra S</creator><scope>EVB</scope></search><sort><creationdate>20231121</creationdate><title>Learned step size quantization</title><author>Appuswamy, Rathinakumar ; McKinstry, Jeffrey L ; Esser, Steve ; Bablani, Deepika ; Modha, Dharmendra S</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_US11823054B23</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>eng</language><creationdate>2023</creationdate><topic>CALCULATING</topic><topic>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</topic><topic>COMPUTING</topic><topic>COUNTING</topic><topic>ELECTRIC DIGITAL DATA PROCESSING</topic><topic>PHYSICS</topic><toplevel>online_resources</toplevel><creatorcontrib>Appuswamy, Rathinakumar</creatorcontrib><creatorcontrib>McKinstry, Jeffrey L</creatorcontrib><creatorcontrib>Esser, Steve</creatorcontrib><creatorcontrib>Bablani, Deepika</creatorcontrib><creatorcontrib>Modha, Dharmendra S</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Appuswamy, Rathinakumar</au><au>McKinstry, Jeffrey L</au><au>Esser, Steve</au><au>Bablani, Deepika</au><au>Modha, Dharmendra S</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>Learned step size quantization</title><date>2023-11-21</date><risdate>2023</risdate><abstract>Learned step size quantization in artificial neural network is provided. In various embodiments, a system comprises an artificial neural network and a computing node. The artificial neural network comprises: a quantizer having a configurable step size, the quantizer adapted to receive a plurality of input values and quantize the plurality of input values according to the configurable step size to produce a plurality of quantized input values, at least one matrix multiplier configured to receive the plurality of quantized input values from the quantizer and to apply a plurality of weights to the quantized input values to determine a plurality of output values having a first precision, and a multiplier configured to scale the output values to a second precision. The computing node is operatively coupled to the artificial neural network and is configured to: provide training input data to the artificial neural network, and optimize the configurable step size based on a gradient through the quantizer and the training input data.</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_epo_espacenet_US11823054B2 |
| source | esp@cenet |
| subjects | CALCULATING COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS COMPUTING COUNTING ELECTRIC DIGITAL DATA PROCESSING PHYSICS |
| title | Learned step size quantization |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-04T01%3A23%3A52IST&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=Appuswamy,%20Rathinakumar&rft.date=2023-11-21&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3EUS11823054B2%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 |