Comments on: Japan Gets An LLM Compliments Of Fujitsu And RIKEN https://www.nextplatform.com/2024/05/13/japan-gets-an-llm-compliments-of-fujitsu-and-riken/ In-depth coverage of high-end computing at large enterprises, supercomputing centers, hyperscale data centers, and public clouds. Wed, 22 May 2024 14:31:25 +0000 hourly 1 https://wordpress.org/?v=6.7.1 By: Timothy Prickett Morgan https://www.nextplatform.com/2024/05/13/japan-gets-an-llm-compliments-of-fujitsu-and-riken/#comment-224414 Thu, 16 May 2024 01:08:36 +0000 https://www.nextplatform.com/?p=144143#comment-224414 In reply to Petteri.

Correct! Thanks.

]]> By: Petteri https://www.nextplatform.com/2024/05/13/japan-gets-an-llm-compliments-of-fujitsu-and-riken/#comment-224406 Wed, 15 May 2024 22:32:13 +0000 https://www.nextplatform.com/?p=144143#comment-224406 There is a typo in the last paragraph: ‘13,824 modes’. The author may have meant ‘nodes’.

]]> By: Calamity Jim https://www.nextplatform.com/2024/05/13/japan-gets-an-llm-compliments-of-fujitsu-and-riken/#comment-224355 Tue, 14 May 2024 13:33:56 +0000 https://www.nextplatform.com/?p=144143#comment-224355 Way to go! Training LLMs directly in different written languages (rather than via translation), including alphabetic, phonetic, ideogrammed, and even hieroglyphic ones, seems to me like the best way to enable the eventual identification of their linguistic sub-components, and their “distinct” cognitive components (if any), by inter-comparison of the trained results. And, like many, I definitely wonder what Fugaku-LLM would respond to the prompt: “わさびはいかがですか”? (eh-eh-eh!)

]]>