Adaptive speculative decoding
Examples herein relate to decoding tokens using speculative decoding operations to decode tokens at an offset from a token decoded by a sequential decoding operation. At a checkpoint, a determination is made as to whether tokens to be decoded by the sequential and speculative decoding operations ali...
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 | Satpathy, Sudhir K Suresh, Vikram B Mathew, Sanu K |
| description | Examples herein relate to decoding tokens using speculative decoding operations to decode tokens at an offset from a token decoded by a sequential decoding operation. At a checkpoint, a determination is made as to whether tokens to be decoded by the sequential and speculative decoding operations align. If there is alignment, the speculatively decoded tokens after a discard window are committed and made available for access. If there is not alignment, the speculatively decoded tokens are discarded. A miss in alignment and a fullness level of a buffer that stores speculatively decoded tokens are assessed to determine a next offset level for a start of speculative decoding. A size of a discard window can be set using a relationship based on the offset level to improve buffer utilization and to attempt to improve changes of alignments. |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_US10911063B2</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>US10911063B2</sourcerecordid><originalsourceid>FETCH-epo_espacenet_US10911063B23</originalsourceid><addsrcrecordid>eNrjZJB1TEksKMksS1UoLkhNLs1JBLNTUpPzUzLz0nkYWNMSc4pTeaE0N4Oim2uIs4duakF-fGpxQWJyal5qSXxosKGBpaGhgZmxk5ExMWoA5Uckcg</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>Adaptive speculative decoding</title><source>esp@cenet</source><creator>Satpathy, Sudhir K ; Suresh, Vikram B ; Mathew, Sanu K</creator><creatorcontrib>Satpathy, Sudhir K ; Suresh, Vikram B ; Mathew, Sanu K</creatorcontrib><description>Examples herein relate to decoding tokens using speculative decoding operations to decode tokens at an offset from a token decoded by a sequential decoding operation. At a checkpoint, a determination is made as to whether tokens to be decoded by the sequential and speculative decoding operations align. If there is alignment, the speculatively decoded tokens after a discard window are committed and made available for access. If there is not alignment, the speculatively decoded tokens are discarded. A miss in alignment and a fullness level of a buffer that stores speculatively decoded tokens are assessed to determine a next offset level for a start of speculative decoding. A size of a discard window can be set using a relationship based on the offset level to improve buffer utilization and to attempt to improve changes of alignments.</description><language>eng</language><subject>BASIC ELECTRONIC CIRCUITRY ; CODE CONVERSION IN GENERAL ; CODING ; DECODING ; ELECTRICITY</subject><creationdate>2021</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=20210202&DB=EPODOC&CC=US&NR=10911063B2$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,780,885,25562,76317</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20210202&DB=EPODOC&CC=US&NR=10911063B2$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>Satpathy, Sudhir K</creatorcontrib><creatorcontrib>Suresh, Vikram B</creatorcontrib><creatorcontrib>Mathew, Sanu K</creatorcontrib><title>Adaptive speculative decoding</title><description>Examples herein relate to decoding tokens using speculative decoding operations to decode tokens at an offset from a token decoded by a sequential decoding operation. At a checkpoint, a determination is made as to whether tokens to be decoded by the sequential and speculative decoding operations align. If there is alignment, the speculatively decoded tokens after a discard window are committed and made available for access. If there is not alignment, the speculatively decoded tokens are discarded. A miss in alignment and a fullness level of a buffer that stores speculatively decoded tokens are assessed to determine a next offset level for a start of speculative decoding. A size of a discard window can be set using a relationship based on the offset level to improve buffer utilization and to attempt to improve changes of alignments.</description><subject>BASIC ELECTRONIC CIRCUITRY</subject><subject>CODE CONVERSION IN GENERAL</subject><subject>CODING</subject><subject>DECODING</subject><subject>ELECTRICITY</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2021</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNrjZJB1TEksKMksS1UoLkhNLs1JBLNTUpPzUzLz0nkYWNMSc4pTeaE0N4Oim2uIs4duakF-fGpxQWJyal5qSXxosKGBpaGhgZmxk5ExMWoA5Uckcg</recordid><startdate>20210202</startdate><enddate>20210202</enddate><creator>Satpathy, Sudhir K</creator><creator>Suresh, Vikram B</creator><creator>Mathew, Sanu K</creator><scope>EVB</scope></search><sort><creationdate>20210202</creationdate><title>Adaptive speculative decoding</title><author>Satpathy, Sudhir K ; Suresh, Vikram B ; Mathew, Sanu K</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_US10911063B23</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>eng</language><creationdate>2021</creationdate><topic>BASIC ELECTRONIC CIRCUITRY</topic><topic>CODE CONVERSION IN GENERAL</topic><topic>CODING</topic><topic>DECODING</topic><topic>ELECTRICITY</topic><toplevel>online_resources</toplevel><creatorcontrib>Satpathy, Sudhir K</creatorcontrib><creatorcontrib>Suresh, Vikram B</creatorcontrib><creatorcontrib>Mathew, Sanu K</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Satpathy, Sudhir K</au><au>Suresh, Vikram B</au><au>Mathew, Sanu K</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>Adaptive speculative decoding</title><date>2021-02-02</date><risdate>2021</risdate><abstract>Examples herein relate to decoding tokens using speculative decoding operations to decode tokens at an offset from a token decoded by a sequential decoding operation. At a checkpoint, a determination is made as to whether tokens to be decoded by the sequential and speculative decoding operations align. If there is alignment, the speculatively decoded tokens after a discard window are committed and made available for access. If there is not alignment, the speculatively decoded tokens are discarded. A miss in alignment and a fullness level of a buffer that stores speculatively decoded tokens are assessed to determine a next offset level for a start of speculative decoding. A size of a discard window can be set using a relationship based on the offset level to improve buffer utilization and to attempt to improve changes of alignments.</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_epo_espacenet_US10911063B2 |
| source | esp@cenet |
| subjects | BASIC ELECTRONIC CIRCUITRY CODE CONVERSION IN GENERAL CODING DECODING ELECTRICITY |
| title | Adaptive speculative decoding |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-14T09%3A30%3A17IST&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=Satpathy,%20Sudhir%20K&rft.date=2021-02-02&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3EUS10911063B2%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 |