[B! æ©Ÿæ¢°å¦ç¿’][google] fm315ã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

fm315 id:fm315

æ©Ÿæ¢°å¦ç¿’ã¨googleã«é–¢ã™ã‚‹fm315ã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (7)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

Google Colab ã§ PEFT ã«ã‚ˆã‚‹å¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ«ã®ãƒ•ã‚¡ã‚¤ãƒ³ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’è©¦ã™ï½œnpaka
ã€ŒGoogle Colabã€ã§ ã€ŒPEFTã€ã«ã‚ˆã‚‹å¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ«ã®ãƒ•ã‚¡ã‚¤ãƒ³ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’è©¦ã—ãŸã®ã§ã€ã¾ã¨ã‚ã¾ã—ãŸã€‚ 1. PEFTã€ŒPEFTã€(Parameter-Efficient Fine-Tuning)ã¯ã€ãƒ¢ãƒ‡ãƒ«ã®å…¨ä½“ã®ãƒ•ã‚¡ã‚¤ãƒ³ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ãªã—ã«ã€äº‹å‰å¦ç¿’æ¸ˆã¿ã®è¨€èªžãƒ¢ãƒ‡ãƒ«ã‚’ã•ã¾ã–ã¾ãªä¸‹æµã‚¿ã‚¹ã‚¯ã«é©å¿œã•ã›ã‚‹ã“ã¨ãŒã§ãã‚‹ãƒ‘ãƒƒã‚±ãƒ¼ã‚¸ã§ã™ã€‚ å¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ«ã®ãƒ•ã‚¡ã‚¤ãƒ³ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã¯ã€å¤šãã®å ´åˆã€æ³•å¤–ãªã‚³ã‚¹ãƒˆãŒã‹ã‹ã‚Šã¾ã™ãŒã€ã€ŒPEFTã€ã¯å°‘æ•°ã®ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿ã®ã¿ã‚’ãƒ•ã‚¡ã‚¤ãƒ³ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã™ã‚‹ãŸã‚ã€è¨ˆç®—ã‚³ã‚¹ãƒˆã¨ã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸ ã‚³ã‚¹ãƒˆãŒå¤§å¹…ã«å‰Šæ¸›ã§ãã€ã•ã‚‰ã«ã¯ã€å®Œå…¨ãªãƒ•ã‚¡ã‚¤ãƒ³ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã«åŒ¹æ•µã™ã‚‹ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ã‚’å®Ÿç¾ã—ã¾ã™ã€‚ ç¾åœ¨ã‚µãƒãƒ¼ãƒˆã—ã¦ã„ã‚‹æ‰‹æ³•ã¯ã€æ¬¡ã®4ã¤ã§ã™ã€‚ ãƒ»LoRA ãƒ»Prefix Tuning ãƒ»P-Tuning ãƒ»Prompt Tuning 2. Colabã§ã®å®Ÿè¡ŒGoogle Colab
fm315 2023/02/23
google

performance

æ©Ÿæ¢°å¦ç¿’

ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°
ãƒªãƒ³ã‚¯
ã‚´ãƒ¼ãƒ«ãƒ‡ãƒ³ã‚¦ã‚£ãƒ¼ã‚¯ãªã®ã§Google App Engineã«æ©Ÿæ¢°å¦ç¿’ã®ãƒ¢ãƒ‡ãƒ«ã‚’deployã—ã¦éŠã‚“ã§ã¿ãŸ (using TensorFlow 2.1) | DevelopersIO
ã“ã‚“ã«ã¡ã¯ã€ã‚´ãƒ¼ãƒ«ãƒ‡ãƒ³ã‚¦ã‚£ãƒ¼ã‚¯ã¯ã—ã£ã‹ã‚Šã‚¹ãƒ†ã‚¤ãƒ›ãƒ¼ãƒ ã—ã¦ã„ã‚‹Mr.Moã§ã™ã€‚ GWã«ä½•ã‚’ã‚„ã‚ã†ã‹ãªãƒ¼ã¨æ€ã£ã¦ã„ãŸã‚‰ç›®ã®å‰ã«App Engineã¨æ©Ÿæ¢°å¦ç¿’ã®ãƒ¢ãƒ‡ãƒ«ãŒã‚ã£ãŸã®ã§éŠã‚“ã§ã¿ã¾ã—ãŸï¼ Google App Engineã¨ã¯ ä¸‹è¨˜ã«è©³ã—ã„è¨˜äº‹ãŒã‚ã‚‹ã®ã§ã”å‚ç…§ãã ã•ã„ï¼ https://docs.google.com/presentation/d/1trxz4_ksRyHdj2cAGVKiy3gtx7q1-AE06XbzLraBD44/htmlpresent https://dev.classmethod.jp/articles/gae-webapp/ ãªãœGAEãªã®ã‹ï¼Ÿ App Engineç„¡ã—ã§ã¯ç”Ÿãã¦ã„ã‘ãªã„ä½“ã«ãªã£ã¦ã„ã‚‹ã‹ã‚‰ã§ã™ã€‚ï¼ˆç¬‘ï¼‰ ã‚ã¨ã¯ã‚„ã£ã±ã‚Šç„¡æ–™ã§å‹•ã‹ã›ãªã„ã‹ã¨æ€ã„ã¾ã—ã¦... æœ€ä½Žé™å¿…è¦ãªã‚‚ã® Google Cloud Platformã«ãƒã‚°ã‚¤ãƒ³ã§ãã‚‹ã‚¢ã‚«ã‚¦ãƒ³ãƒˆ æ©Ÿæ¢°
fm315 2020/05/05
google

å¦ç¿’

æ©Ÿæ¢°å¦ç¿’

Python

Google Cloud Platform
ãƒªãƒ³ã‚¯
åŸºæœ¬ã‹ã‚‰å¦ã¶ ãƒ“ãƒƒã‚°ãƒ‡ãƒ¼ã‚¿ / ãƒ‡ãƒ¼ã‚¿åˆ†æž / æ©Ÿæ¢°å¦ç¿’ ã‚µãƒ¼ãƒ“ã‚¹ç¾¤
2017-11-22 wed. ç¬¬ 2 å›ž Google Cloud INSIDE Games & Apps Google Cloud ãƒ‡ãƒ¼ã‚¿ã‚¢ãƒŠãƒªãƒ†ã‚£ã‚¯ã‚¹ãƒ†ã‚¯ãƒ‹ã‚«ãƒ«ã‚¹ãƒšã‚·ãƒ£ãƒªã‚¹ãƒˆ ä¸‹ç”° å€«å¤§ ã®ç™»å£‡ã‚¹ãƒ©ã‚¤ãƒ‰ã§ã™ã€‚Read less
fm315 2017/12/21
æ©Ÿæ¢°å¦ç¿’

Google
ãƒªãƒ³ã‚¯
TensorFlowã§ã†ãªãŽã¨å¯¿å¸ã‚’åˆ¤å®šã™ã‚‹Androidã‚¢ãƒ—ãƒªã‚’ä½œã£ã¦ã¿ãŸï¼Ž - Qiita
tensorflowã‚’ä½¿ã£ã¦ã†ãªãŽã¨å¯¿å¸ã®åˆ¤åˆ¥ã™ã‚‹Classifierï¼ˆåˆ†é¡žå™¨ï¼‰ã‚’ä½œã£ã¦ã¿ã¾ã—ãŸï¼Ž ï¼ˆä»Šã®å£ç¯€ã«æ—¬ãªã‚‚ã®ã‚’åˆ†é¡žã—ãŸã„ã¨ã„ã†ã®ãŒä»Šå›žã®è¨˜äº‹ã®ãã£ã‹ã‘ã§ã™ï¼Žï¼‰ ãŠãŠã¾ã‹ãªæµã‚Œã¨ã—ã¦ã¯ï¼Œã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿ã§ãƒ¢ãƒ‡ãƒ«ã®å¦ç¿’ã‚’è¡Œã£ã¦å®Œæˆã—ãŸãƒ¢ãƒ‡ãƒ«ã‚’androidã«å…¥ã‚Œã¦ä½¿ç”¨ã™ã‚‹ã¨ã„ã†æ„Ÿã˜ã§ã™ï¼Ž è¨˜äº‹ã®å†…å®¹ã¯ä»»å¤©å ‚ã®ã‚ãƒ£ãƒ©ã‚’åˆ†é¡žã™ã‚‹ã“ã®è¨˜äº‹ã‚’ã‹ãªã‚Šå‚è€ƒã«ã•ã›ã¦ã‚‚ã‚‰ã„ã¾ã—ãŸï¼Žè¨˜äº‹ã§ã¯å¤šã‚¯ãƒ©ã‚¹ã§ã™ãŒï¼Œä»Šå›žã¯tensorflowã®ç†è§£ã‚’æ·±ã‚ã‚‹ã“ã¨ãŒç›®çš„ãªã®ã§ï¼Œï¼’ã‚¯ãƒ©ã‚¹åˆ†é¡žå•é¡Œã¨ã—ã¾ã—ãŸï¼Ž (ã‚‚ã¡ã‚ã‚“ï¼’ã¤ä»¥ä¸Šã®ã‚‚ã®ã‚’å¦ç¿’ã•ã›ã‚‹ã“ã¨ã¯å¯èƒ½ã§ã™ï¼Žï¼“å€‹ã§ã‚‚ï¼‘ï¼ï¼å€‹ã§ã‚‚ï¼’ï¼ï¼å€‹ã§ã‚‚ï¼Žï¼Žï¼Ž) ã§ã¯æ—©é€Ÿã‚„ã£ã¦ã„ãã¾ã—ã‚‡ã†ï¼ ##ã¾ãšã¯åˆ†é¡žã—ãŸã„ã‚‚ã®ã®ç”»åƒã‚’é›†ã‚ã¾ã—ã‚‡ã†ï¼ ã ã„ãŸã„ï¼”ï¼ã€œï¼–ï¼æžšãã‚‰ã„ã ã¨ã„ã„ã§ã™ãï¼Žæœ€ä½Žã§ã‚‚ï¼“ï¼æžšã¯å¿…è¦ã§ã™ï¼Ž å¤šã„åˆ†ã«ã¯å•é¡Œã¯ãªã„ã®ãŒæ·±å±¤å¦ç¿’ã§ã™ãŒï¼Œæ¥µç«¯ã«å¤šã„æžšæ•°ï¼’ï¼ï¼ï¼ï¼æžšã¨ã‹ã ã¨
fm315 2017/08/20
android

Google

Transflow

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
ï¼»é€Ÿå ±ï¼½Googleã€æ©Ÿæ¢°å¦ç¿’ã«æœ€é©åŒ–ã—ãŸã€ŒCloud TPUã€ç™ºè¡¨ã€‚ã€Œã‚ã‚Œã‚ã‚Œã‚Œã¯ã„ã¾ã€AIãƒ•ã‚¡ãƒ¼ã‚¹ãƒˆãƒ‡ãƒ¼ã‚¿ã‚»ãƒ³ã‚¿ãƒ¼ã®æ§‹ç¯‰ã«å–ã‚Šã‹ã‹ã£ã¦ã„ã‚‹ã€ã€‚Google I/O 2017
ï¼»é€Ÿå ±ï¼½Googleã€æ©Ÿæ¢°å¦ç¿’ã«æœ€é©åŒ–ã—ãŸã€ŒCloud TPUã€ç™ºè¡¨ã€‚ã€Œã‚ã‚Œã‚ã‚Œã‚Œã¯ã„ã¾ã€AIãƒ•ã‚¡ãƒ¼ã‚¹ãƒˆãƒ‡ãƒ¼ã‚¿ã‚»ãƒ³ã‚¿ãƒ¼ã®æ§‹ç¯‰ã«å–ã‚Šã‹ã‹ã£ã¦ã„ã‚‹ã€ã€‚Google I/O 2017 Googleã¯5æœˆ17æ—¥ï¼ˆæ—¥æœ¬æ™‚é–“5æœˆ18æ—¥æœªæ˜Žï¼‰ã‹ã‚‰é–‹å‚¬ä¸ã®å¹´æ¬¡ã‚¤ãƒ™ãƒ³ãƒˆã€ŒGoogle I/O 2017ã€ã«ã¦ã€æ©Ÿæ¢°å¦ç¿’ã«æœ€é©åŒ–ã—ãŸç¬¬äºŒä¸–ä»£ã®TensorFlow Processing Unitï¼ˆTPUï¼‰ã¨ãã‚Œã‚’ç”¨ã„ãŸã‚¯ãƒ©ã‚¦ãƒ‰ã‚µãƒ¼ãƒ“ã‚¹ã€ŒCloud TPUã€ã‚’ç™ºè¡¨ã—ã¾ã—ãŸã€‚ ã‚ã‚Œã‚ã‚Œã¯ã„ã¾ã€AIãƒ•ã‚¡ãƒ¼ã‚¹ãƒˆãƒ‡ãƒ¼ã‚¿ã‚»ãƒ³ã‚¿ãƒ¼ã‚’ä½œã£ã¦ã„ã‚‹ Google CEO Sundar Pichaiæ°ã€‚ ã‚ã‚Œã‚ã‚ŒãŒæ¤œç´¢äº‹æ¥ã‚’å§‹ã‚ãŸã¨ãã€ã“ã‚Œã‚’ã‚¹ã‚±ãƒ¼ãƒ«ã•ã›ã‚‹ãŸã‚ã«ã¯ãã‚Œã¾ã§ã®è¨ˆç®—æ©Ÿã®ã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£ã‚’è¦‹ç›´ã™å¿…è¦ãŒã‚ã£ãŸã€‚ã ã‹ã‚‰ã“ãã‚ã‚Œã‚ã‚Œã¯ãã®ãŸã‚ã«ãƒ‡ãƒ¼ã‚¿ã‚»ãƒ³ã‚¿ãƒ¼ã‚’æœ€åˆã‹ã‚‰æ§‹ç¯‰ã™ã‚‹ã“ã¨ã«ãªã£ãŸã®ã ã€‚ ãã—ã¦ã„ã¾ã€æ©Ÿæ¢°å¦ç¿’ã‚„AIã®é€²åŒ–ã¯ãµãŸ
fm315 2017/05/25
AI

cloud

Google

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
ï¼»é€Ÿå ±ï¼½Googleã€ã‚¹ãƒžãƒ¼ãƒˆãƒ•ã‚©ãƒ³å‘ã‘ã®æ©Ÿæ¢°å¦ç¿’ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã€ŒTensorFlow Liteã€ç™ºè¡¨ã€‚DSPã‚’ã‚¢ã‚¯ã‚»ãƒ©ãƒ¬ãƒ¼ã‚¿ã¨ã—ã¦åˆ©ç”¨ã€‚Google I/O 2017
ï¼»é€Ÿå ±ï¼½Googleã€ã‚¹ãƒžãƒ¼ãƒˆãƒ•ã‚©ãƒ³å‘ã‘ã®æ©Ÿæ¢°å¦ç¿’ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã€ŒTensorFlow Liteã€ç™ºè¡¨ã€‚DSPã‚’ã‚¢ã‚¯ã‚»ãƒ©ãƒ¬ãƒ¼ã‚¿ã¨ã—ã¦åˆ©ç”¨ã€‚Google I/O 2017 Googleã¯5æœˆ17æ—¥ï¼ˆæ—¥æœ¬æ™‚é–“5æœˆ18æ—¥æœªæ˜Žï¼‰ã‹ã‚‰é–‹å‚¬ä¸ã®å¹´æ¬¡ã‚¤ãƒ™ãƒ³ãƒˆã€ŒGoogle I/O 2017ã€ã«ã¦ã€ã‚¹ãƒžãƒ¼ãƒˆãƒ•ã‚©ãƒ³ãªã©ã®ãƒ‡ãƒã‚¤ã‚¹ã§å®Ÿè¡Œã™ã‚‹ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³å‘ã‘ã®æ©Ÿæ¢°å¦ç¿’ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã€ŒTensorFlow Liteã€ã‚’ç™ºè¡¨ã—ã¾ã—ãŸã€‚ ã€ŒTensorFlow Liteã¯ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³å‘ã‘ã®ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã§ã€é«˜é€Ÿã‹ã¤è»½é‡ã§ã‚ã‚ŠãªãŒã‚‰ã€Continentã‚„LSTMã¨ã„ã£ãŸæœ€æ–°ã®æŠ€è¡“ã‚’åˆ©ç”¨å¯èƒ½ã ã€ï¼ˆDave Burkeæ°ã€Vice President Enginieering, Androidï¼‰ TensorFlow LiteåŒæ™‚ã«ã€æ©Ÿæ¢°å¦ç¿’ãªã©ã§ä½¿ã‚ã‚Œã‚‹ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®è¨ˆç®—ã‚’DSPã«ã‚ˆã£ã¦ãƒãƒ¼ãƒ‰ã‚¦ã‚§ã‚¢ã‚¢ã‚¯ã‚»ãƒ©ãƒ¬ãƒ¼ã‚·ãƒ§
fm315 2017/05/25
Google

deeplearning

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
Google ã® Tensor Processing Unit (TPU) ã§æ©Ÿæ¢°å¦ç¿’ãŒ 30 å€é€Ÿããªã‚‹ãƒ¡ã‚«ãƒ‹ã‚ºãƒ | Google Cloud å…¬å¼ãƒ–ãƒã‚°
* ã“ã®æŠ•ç¨¿ã¯ç±³å›½æ™‚é–“ 5 æœˆ 12 æ—¥ã«æŠ•ç¨¿ã•ã‚ŒãŸã‚‚ã®ï¼ˆæŠ•ç¨¿ã¯ã“ã¡ã‚‰ï¼‰ã®æŠ„è¨³ã§ã™ã€‚ Posted by ä½è—¤ä¸€æ†², Staff Developer Advocate, Google Cloud Cliff Young, Software Engineer, Google Brain David Patterson, Distinguished Engineer, Google Brain Google æ¤œç´¢ã€ã‚¹ãƒˆãƒªãƒ¼ãƒˆãƒ“ãƒ¥ãƒ¼ã€Google ãƒ•ã‚©ãƒˆã€ãã—ã¦Google ç¿»è¨³ã€‚ã“ã‚Œã‚‰ã®ã‚µãƒ¼ãƒ“ã‚¹ã«å…±é€šã™ã‚‹ã®ã¯ã€ã„ãšã‚Œã‚‚ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ï¼ˆNNï¼‰ã®è¨ˆç®—å‡¦ç†ã®é«˜é€ŸåŒ–ã®ãŸã‚ã« Google ã®ç¬¬ä¸€ä¸–ä»£ã® Tensor Processing Unit (TPU) ãŒç”¨ã„ã‚‰ã‚Œã¦ã„ã‚‹ç‚¹ã§ã™ã€‚ Google ã® Tensor Processing Unit (TPU) ãŒæè¼‰ã•ã‚ŒãŸå›žè·¯åŸºæ¿ï¼ˆå·¦ï¼‰ã¨ã€ G
fm315 2017/05/13
Google

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
1