๐ค๐จ๐ญ๐š๐ง.๐ค๐ง (@kotan_kn) 's Twitter Profile
๐ค๐จ๐ญ๐š๐ง.๐ค๐ง

@kotan_kn

A numerical artisan specializing in ML(i.e. statistics), audio & visual techs, focusing on signal processing and physical simulation for computational acoustics

ID: 114063576

calendar_today14-02-2010 00:43:03

11,11K Tweet

133 Followers

62 Following

๐ค๐จ๐ญ๐š๐ง.๐ค๐ง (@kotan_kn) 's Twitter Profile Photo

SaaSใฎๆญปใงใ‚ˆใ่ฆ‹ใ‚‹่ญฐ่ซ–ใงใฏ0ใ‹100ใ‹ใง่ชžใ‚‰ใ‚Œใฆใ„ใฆใชใ‚“ใ ใ‹ใชใใจใ„ใ†ๆฐ—ๆŒใก๏ผŽ ใ‚‚ใกใ‚ใ‚“AIใซ่‚ฉไปฃใ‚ใ‚Šใ•ใ‚Œใชใ„ไป˜ๅŠ ไพกๅ€คใ‚’ๆไพ›ใงใใชใ„SaaSใฏ้ง†้€ใ•ใ‚Œใ‚‹ใจๆ€ใ†ใ‘ใฉ๏ผŒ็މ็Ÿณๆททไบคใ ใ—ๅ…จๆป…ใฏใ—ใชใ„ใ‚“ใ˜ใ‚ƒใชใ„ใ‹ใช๏ผŽ ใพใ‚๏ผŒ้ ใ„ๆœชๆฅใงใฎๆถˆๆป…ใŒ็ด„ๆŸใ•ใ‚Œใใ‚Œใพใงๆ–œ้™ฝใง็”Ÿใๆฎ‹ใ‚‹ใจใ„ใ†ใฎใ‚’ๆญปใจๅ‘ผใถใชใ‚‰ใใ†ใ‹ใ‚‚๏ผŸ

๐ค๐จ๐ญ๐š๐ง.๐ค๐ง (@kotan_kn) 's Twitter Profile Photo

ๅ€‹ไบบ็š„ใซใฏๆ—ฉใAIใซใ‚ฝใƒ•ใƒˆใ‚ฆใ‚จใ‚ข้–‹็™บใƒปใƒ‡ใƒผใ‚ฟๅˆ†ๆžใฎไป•ไบ‹ใ‚’ๅฅชใ‚ใ‚ŒใŸใ„็ซ‹ๅ ด๏ผŽ ใใ†ใ™ใ‚Œใฐ็ฉบใ„ใŸๆ™‚้–“ใงใ‚ฝใƒ•ใƒˆใ‚ฆใ‚จใ‚ข้–‹็™บใจใƒ‡ใƒผใ‚ฟๅˆ†ๆžใ‚’ใงใใ‚‹ใฎใซโ€ฆโ€ฆ

iwashi / Yoshimasa Iwase (@iwashi86) 's Twitter Profile Photo

ใ‚ฏใƒฉใƒ•ใƒˆใƒžใƒณใ‚ทใƒƒใƒ—ใจใ—ใฆใฎใƒ—ใƒญใ‚ฐใƒฉใƒŸใƒณใ‚ฐใŒAIใซๅ–ใฃใฆไปฃใ‚ใ‚‰ใ‚Œใ€ๆถˆใˆใ‚†ใใ“ใจใธใฎ้ฟใ‘ใ‚‰ใ‚Œใชใ„ๆ‚ฒใ—ใฟใจๅ“€ๆ‚ผใฎๆ„ใ‚’็คบใ—ใฆใ„ใ‚‹ใƒ–ใƒญใ‚ฐใ ใฃใŸใ€‚ ใƒปAIใŒ้–‹็™บ่€…ใฎใƒ–ใƒญใ‚ฐใ‚„ใ‚ณใƒผใƒ‰ใ‚’ๅ‹ๆ‰‹ใซๅญฆ็ฟ’ใ—ใ€ไป–ไบบใฎๅˆฉ็›ŠใฎใŸใ‚ใซใใ‚Œใ‚‰ใ‚’ๆจกๅ€ฃใ—ใฆใ„ใ‚‹

ITmedia AI๏ผ‹ (@itm_aiplus) 's Twitter Profile Photo

ๆ—ฅๆœฌๆ”ฟๅบœใ€AIใฎ็คพไผšๅฎŸ่ฃ…ใ‚’ๅฆจใ’ใฆใ„ใ‚‹่ฆๅˆถใฎๆƒ…ๅ ฑใ‚’ๅ‹Ÿ้›†ใ€€ๅˆถๅบฆ่ฆ‹็›ดใ—ใฎๅ‚่€ƒใซ itmedia.co.jp/aiplus/articleโ€ฆ

Z.ai (@zai_org) 's Twitter Profile Photo

Introducing GLM-5: From Vibe Coding to Agentic Engineering GLM-5 is built for complex systems engineering and long-horizon agentic tasks. Compared to GLM-4.5, it scales from 355B params (32B active) to 744B (40B active), with pre-training data growing from 23T to 28.5T tokens.

Introducing GLM-5: From Vibe Coding to Agentic Engineering

GLM-5 is built for complex systems engineering and long-horizon agentic tasks. Compared to GLM-4.5, it scales from 355B params (32B active) to 744B (40B active), with pre-training data growing from 23T to 28.5T tokens.
Google Gemini App (@geminiapp) 's Twitter Profile Photo

Ready to try the updated Gemini 3 Deep Think? Google AI Ultra users can go to gemini.google or open the app, then select the "Deep Think" option in the tools menu to get started.

Ready to try the updated Gemini 3 Deep Think?

Google AI Ultra users can go to gemini.google or open the app, then select the "Deep Think" option in the tools menu to get started.
๐ค๐จ๐ญ๐š๐ง.๐ค๐ง (@kotan_kn) 's Twitter Profile Photo

ใƒญใƒผใ‚ซใƒซLLMใƒกใ‚คใƒณใง็ต„ใ‚“ใงใ‚‹ใ‹ใ‚‰custom providerใ‚ใ‚ŠใŒใŸใ™ใŽใ‚‹

๐ค๐จ๐ญ๐š๐ง.๐ค๐ง (@kotan_kn) 's Twitter Profile Photo

่‡ชๅˆ†็”จใฎopenclawใ‚จใ‚ณใ‚ทใ‚นใƒ†ใƒ ใฏๆœ€ไฝŽใงใ‚‚3ใƒขใƒ‡ใƒซใฏๅŒๆ™‚ใซๅ‹•ใ‹ใ—ใฆใ‚‹ใ‘ใฉ ไธป่ฆใƒญใƒผใ‚ซใƒซLLMใฏๅ…จ้ƒจQwen3-*ใซใชใฃใฆใ—ใพใฃใŸโ€ฆ๏ผŽ Nemotron-3-nanoใ‚‚ๅ‰ฒใจๅŒ–ใ‘็‰ฉใชใ‚“ใ ใ‘ใฉ๏ผŒMoEใ ใ‹ใ‚‰ใ‹่‹ฑ่ชžไปฅๅค–ใงๅคšใใฎใƒ‰ใƒกใ‚คใƒณ็Ÿฅ่ญ˜ใ‚’ๅธใ‚ใ›ใ‚‹ใจ้€”็ซฏใซๅฃŠใ‚Œใ‚‹ใฎใŒ

๐ค๐จ๐ญ๐š๐ง.๐ค๐ง (@kotan_kn) 's Twitter Profile Photo

Qwen3-Coder-Next (80b)ใ‚’ๅ‹•ใ‹ใ—ใŸใใชใฃใฆใใŸใ‘ใฉใƒใƒผใƒ‰ใŒใ‚‚ใ†ใชใ„ใชโ€ฆ๏ผŽ ใŸใ MoEใฃใฝใ„ใ‹ใ‚‰ๅคšใƒ‰ใƒกใ‚คใƒณใซๅฏพๅฟœใ™ใ‚‹ไปฃใ‚ใ‚Šใซใ‚ณใƒผใƒ‡ใ‚ฃใƒณใ‚ฐๆ€ง่ƒฝ่‡ชไฝ“ใฏ30Bใ‹ใ‚‰ใ‚ใพใ‚Šๅค‰ใ‚ใฃใฆใชใ„ๅฏ่ƒฝๆ€งใ‚‚๏ผŸ

๐ค๐จ๐ญ๐š๐ง.๐ค๐ง (@kotan_kn) 's Twitter Profile Photo

ๆœ€่ฟ‘ใฎAIใงใŠใ‚‚ใ†ใ“ใจ: ใƒ•ใ‚กใ‚ธใƒผใ‚ณใƒณใƒ”ใƒฅใƒผใƒ†ใ‚ฃใƒณใ‚ฐ