Model training method and device and readable storage medium
The embodiment of the invention relates to a model training method and device and a readable storage medium, and the method comprises the steps: obtaining a sample data set corresponding to a target task, a pre-trained teacher model and an ith initial student model; i-th channel pruning is carried o...
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Patent |
| Sprache: | chi ; 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 | ZENG HAIEN |
| description | The embodiment of the invention relates to a model training method and device and a readable storage medium, and the method comprises the steps: obtaining a sample data set corresponding to a target task, a pre-trained teacher model and an ith initial student model; i-th channel pruning is carried out on the i-th initial student model, a student model after i-th channel pruning is obtained, and the initial value of i is 1; knowledge distillation is carried out according to the sample data set, the teacher model and the student model after ith channel pruning, an (i + 1) th initial student model is obtained, and the compression ratio between the (i + 1) th initial student model and the ith initial student model is equal to a preset ith compression ratio; and updating i = i + 1, returning to execute the ith channel pruning on the ith initial student model until the updated i is greater than a preset threshold N, and obtaining a target student model. Step-by-step compression is realized through successive prunin |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_CN115511071A</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>CN115511071A</sourcerecordid><originalsourceid>FETCH-epo_espacenet_CN115511071A3</originalsourceid><addsrcrecordid>eNrjZLDxzU9JzVEoKUrMzMvMS1fITS3JyE9RSMxLUUhJLctMTgUzi1ITUxKTclIVikvyixLTU4HKUjJLc3kYWNMSc4pTeaE0N4Oim2uIs4duakF-fGpxQWJyal5qSbyzn6GhqamhoYG5oaMxMWoAt_cvWQ</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>Model training method and device and readable storage medium</title><source>esp@cenet</source><creator>ZENG HAIEN</creator><creatorcontrib>ZENG HAIEN</creatorcontrib><description>The embodiment of the invention relates to a model training method and device and a readable storage medium, and the method comprises the steps: obtaining a sample data set corresponding to a target task, a pre-trained teacher model and an ith initial student model; i-th channel pruning is carried out on the i-th initial student model, a student model after i-th channel pruning is obtained, and the initial value of i is 1; knowledge distillation is carried out according to the sample data set, the teacher model and the student model after ith channel pruning, an (i + 1) th initial student model is obtained, and the compression ratio between the (i + 1) th initial student model and the ith initial student model is equal to a preset ith compression ratio; and updating i = i + 1, returning to execute the ith channel pruning on the ith initial student model until the updated i is greater than a preset threshold N, and obtaining a target student model. Step-by-step compression is realized through successive prunin</description><language>chi ; eng</language><subject>CALCULATING ; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS ; COMPUTING ; COUNTING ; PHYSICS</subject><creationdate>2022</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=20221223&DB=EPODOC&CC=CN&NR=115511071A$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,776,881,25542,76289</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20221223&DB=EPODOC&CC=CN&NR=115511071A$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>ZENG HAIEN</creatorcontrib><title>Model training method and device and readable storage medium</title><description>The embodiment of the invention relates to a model training method and device and a readable storage medium, and the method comprises the steps: obtaining a sample data set corresponding to a target task, a pre-trained teacher model and an ith initial student model; i-th channel pruning is carried out on the i-th initial student model, a student model after i-th channel pruning is obtained, and the initial value of i is 1; knowledge distillation is carried out according to the sample data set, the teacher model and the student model after ith channel pruning, an (i + 1) th initial student model is obtained, and the compression ratio between the (i + 1) th initial student model and the ith initial student model is equal to a preset ith compression ratio; and updating i = i + 1, returning to execute the ith channel pruning on the ith initial student model until the updated i is greater than a preset threshold N, and obtaining a target student model. Step-by-step compression is realized through successive prunin</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>2022</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNrjZLDxzU9JzVEoKUrMzMvMS1fITS3JyE9RSMxLUUhJLctMTgUzi1ITUxKTclIVikvyixLTU4HKUjJLc3kYWNMSc4pTeaE0N4Oim2uIs4duakF-fGpxQWJyal5qSbyzn6GhqamhoYG5oaMxMWoAt_cvWQ</recordid><startdate>20221223</startdate><enddate>20221223</enddate><creator>ZENG HAIEN</creator><scope>EVB</scope></search><sort><creationdate>20221223</creationdate><title>Model training method and device and readable storage medium</title><author>ZENG HAIEN</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_CN115511071A3</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>chi ; eng</language><creationdate>2022</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>ZENG HAIEN</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>ZENG HAIEN</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>Model training method and device and readable storage medium</title><date>2022-12-23</date><risdate>2022</risdate><abstract>The embodiment of the invention relates to a model training method and device and a readable storage medium, and the method comprises the steps: obtaining a sample data set corresponding to a target task, a pre-trained teacher model and an ith initial student model; i-th channel pruning is carried out on the i-th initial student model, a student model after i-th channel pruning is obtained, and the initial value of i is 1; knowledge distillation is carried out according to the sample data set, the teacher model and the student model after ith channel pruning, an (i + 1) th initial student model is obtained, and the compression ratio between the (i + 1) th initial student model and the ith initial student model is equal to a preset ith compression ratio; and updating i = i + 1, returning to execute the ith channel pruning on the ith initial student model until the updated i is greater than a preset threshold N, and obtaining a target student model. Step-by-step compression is realized through successive prunin</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | chi ; eng |
| recordid | cdi_epo_espacenet_CN115511071A |
| source | esp@cenet |
| subjects | CALCULATING COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS COMPUTING COUNTING PHYSICS |
| title | Model training method and device and readable storage medium |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-10T07%3A50%3A34IST&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=ZENG%20HAIEN&rft.date=2022-12-23&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3ECN115511071A%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 |