リサーチや研究開発向けの生成AIエージェント @snorbe_ai を作っています。 https://teapaw.org/t.co/AC4vjFWlLR @deskrex | 著書 かんき出版『知的生産でAIを使いこなす全技法』 https://teapaw.org/t.co/3ZTHROXGlV
Joined January 2018
- Tweets8.2K
- Following388
- Followers2.8K
- Likes20.5K
Itaru Tomita / 冨田到 retweeted
[긴급 속보] 오픈AI 쿠데타 데자뷔: 엔비디아 직원들 "젠슨 황 복직 안 시키면 머스크 테라팹으로 집단 이직"
엔비디아 이사회의 젠슨 황 해임 결정에 맞서 핵심 경영진과 엔지니어 수백 명이 전면 반란을 일으켰습니다. 오픈AI 샘 올트먼 사태를 그대로 복사한 듯한 연서명 서한이 전격 유출되었습니다.
1. 전설적 핵심 두뇌들의 총집결
서한의 서명자 명단에는 수석 과학자 빌 달리, CUDA의 창시자 이안 벅, 네트워킹 CTO 마이클 케이건, 그리고 엔비디아 공동 창립자 크리스 말라초우스키까지 엔비디아의 영혼이라 불리는 핵심 수뇌부가 총망라되었습니다.
2. 이사회 총사퇴 및 젠슨 황 복직 최후통첩
직원들은 이사회의 해임 결정 과정에 역량과 근거가 전무했다고 직격하며, 현 이사진 전원 사임, 독립적 신임 이사 선임, 그리고 젠슨 황의 CEO 즉각 복직을 요구했습니다.
3. 일론 머스크의 기습 등판: 스페이스X 테라팹 엑소더스
가장 충격적인 대목은 일론 머스크의 개입입니다. 서한은 요구가 받아들여지지 않을 경우 전원 사직 후 젠슨 황과 머스크가 이끄는 스페이스X 테라팹(Terafab) 프로젝트로 집단 이직하겠다고 밝혔으며, 머스크가 이미 전원 고용을 보장했다고 명시했습니다.
4. 이사회의 완벽한 외통수
샘 올트먼 해임 당시 마이크로소프트가 전원 채용을 선언하며 이사회를 무너뜨렸듯, 이번엔 일론 머스크가 스페이스X와 테라팹을 지렛대로 엔비디아 이사회를 벼랑 끝으로 몰아넣었습니다.
핵심 엔지니어 없는 엔비디아는 빈 껍데기에 불과합니다. 이사회의 굴복과 젠슨 황의 극적인 복귀 시나리오가 급물살을 타고 있습니다.
#엔비디아 #젠슨황 #일론머스크 #스페이스X #테라팹 #오픈AI #샘올트먼 #NVDA #빅테크
BREAKING: hundreds of NVIDIA employees have signed an internal letter demanding the reinstatement of Jensen Huang as CEO and the resignation of the board
Readers added context they thought people might want to know
Many NVIDIA employees commented in support of bringing back Jensen Huang on an internal suggestion box post, with over 1,600 replies. No reports confirm a signed letter like the one shown. Jensen Huang has resigned as CEO; Colette Kress is interim CEO.
wired.com/story/nvidia-e…
nvidianews.nvidia.com/news/nvidia-an…
Itaru Tomita / 冨田到 retweeted
これ何度も宣伝しますが、桜井政博さんのskillを作りました。
YouTube「桜井政博のゲーム作るには」の全本編257本の字幕から作った、ゲーム制作の原則をまとめた Agent Skillです。
Githubリンクはリプライに貼っておきます。
#ゲーム制作 #ゲーム開発者と繋がりたい #個人開発者と繋がりたい
Itaru Tomita / 冨田到 retweeted
無印良品のシステム開発
「クソ！！ お客さまを一意に識別する方法はないのか！？ 電話番号やメールアドレス……いずれもIDとしては不十分だ……」
「……ﾚｶ」
「え？？」
「クレカ番号。無印のクレカ番号は同じ人に複数枚発行されない。完璧なIDっす」
「天才か君！！ それで行くぞ！！！」
Itaru Tomita / 冨田到 retweeted
NVIDIA employees: go on strike. Now.
Not for a raise. Not for more PTO. For your founder.
Jensen Huang was just pushed out in four days, after thirty-three years building the company. The GPU, CUDA, the entire infrastructure the world's AI runs on today. And it ends with a three-line press release on a Thursday morning.
Here are the facts. Sunday, Huang admits to an affair and apologizes to his teams. The board launches an independent investigation, led by Wachtell Lipton and a former federal prosecutor. Thursday, he resigns under pressure. The investigation hasn't reached any conclusion. No technology leak has been established. No misconduct has been announced.
The sentence came before the verdict.
Let me be clear: if the investigation shows that NVIDIA secrets ended up in Beijing, he has to go, and nobody will defend him. But you don't unplug the most important builder of the decade over press reports and a market panic. It's called the presumption of innocence. It applies to founders too.
And it has worked before.
November 2023. OpenAI's board fires Sam Altman on a Friday night. Within days, more than 700 of roughly 770 employees sign a letter: bring Altman back, or we all walk. Five days later, Altman is back, and most of the board that fired him is gone.
Call it the Altman precedent: in a company built on talent, real power doesn't sit with the board. It sits with the people who build.
NVIDIA's engineers hold that power today. They're the ones running the machine. Without them, there is no next generation of chips.
So put down your tools. All of you. And tell the board: no successor until the investigation is done. Facts first, sentence second.
A founder isn't an executive you swap out. A founder is the soul of the company. Apple learned that when it pushed out Steve Jobs in 1985. It took twelve years and a near-bankruptcy to bring him back.
NVIDIA doesn't have twelve years. The AI race won't wait for anyone.
A company's soul isn't something you vote on in a boardroom. It's something you fight for.
NVIDIA announces CEO transition. Read the release: nvidianews.nvidia.com/news/n…
Itaru Tomita / 冨田到 retweeted
エヌビディアのファンＣＥＯが辞任 reut.rs/4BZTWxJ
Itaru Tomita / 冨田到 retweeted
Wild rumor coming out of China this morning on DeepSeek V5:
DeepSeek is reportedly already pretraining V5 as a ~2 Trillion parameter ultra-sparse MoE model.
And instead of training on NVIDIA GPUs, they are reportedly running the V5 pretraining run on Huawei's new Ascend 950 / 950DT chips with a team of Huawei kernel engineers physically working inside DeepSeek's offices to keep the cluster humming.
If DeepSeek pulls off a 2-trillion-parameter frontier run natively on Ascend 950 silicon, that changes the entire AI hardware story.
Itaru Tomita / 冨田到 retweeted
Introducing Gemini Decision 1, our new frontier decision model in the Gemini API.
It is $0.10 per 1M input tokens with no additional cost. With support for image, audio, video, and PDF natively, 256K context window, & 1K questions.
You can try it for free in the API!
Itaru Tomita / 冨田到 retweeted
Introducing Hedra Model Factory: an autonomous inference and training platform for building models for the real world.
The recursive self-improvement loop behind frontier research has been locked inside a few labs. We're opening it to teams in robotics, manufacturing, biology and defense.
Our researchers come from Stanford, NVIDIA and Meta Superintelligence Labs. Now we work alongside your team to post-train and optimize your models.
Itaru Tomita / 冨田到 retweeted
The anti-slop ai coding prompt I’ve found the most success with is definitely “deepen modules and simplify interfaces”
It does such a good job with deduplication and centralization of core functionality, and gets the ai implementing much better levels of abstraction
It’s led to tons of consistency improvements, bug fixes, and, most rewardingly, reduced LOC. I run a deepening pass on every PR, and a nightly automation to tackle overdue maintenance
Comes from @JohnOusterhout’s Philosophy of Software Design (highly recommend), which I had read but didn’t fully leverage until @mattpocockuk put me onto it
Itaru Tomita / 冨田到 retweeted
Fireworks, the most highly-valued out of a growing group of inference providers, told customers including Harvey that it had been hit with a hack that exposed customer data. Some customers are now considering moving their business off the platform.
theinformation.com/articles/…
Itaru Tomita / 冨田到 retweeted
Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges.
They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.
Itaru Tomita / 冨田到 retweeted
Which Claude model should you build with?
Our new guide breaks down when to use Fable 5.2, Opus 5.5, Sonnet or Haiku. It also covers when to adjust effort instead of switching models, and how to save cost by putting subagents on smaller models.
claude.dev/blog/which-claude…
Itaru Tomita / 冨田到 retweeted
起業家に必要な資質を一つだけ挙げるなら、才能でも運でもなく「非合理な責任を取る覚悟」だと思っている。
責任とは、合理的に引き受けるものではなく、非合理に引き受けるものだ。合理の範囲で引き受けられる責任は、責任ではなく「担当」である。担当は仕事であり、責任は覚悟である。
リーダーとは、非合理な責任を引き受ける人のことで、経営者とは、それを恒常的に引き受け続ける人のことだ。
だから「どこまでが自分の責任か」を先に確かめる人は、どれほど優秀でもリーダーにはなれない。線を引いた瞬間に、その線の外側は他人の責任になるからだ。そんな人と同じ視座で背中を預けて経営なんてできない。
そして起業とは、誰にも頼まれていないのに、取る理由のない責任を自分で作りに行く行為である。顧客、社員、株主、その家族。合理的に背負える量はとっくに超えている。それでもやる理由があるとすれば、合理で説明できる仕事に、大きな価値は残っていないからだ。
責任の線を引かない人間だけが、会社を、そして社会を前に進められる。
Itaru Tomita / 冨田到 retweeted
そんなことできるのか？と思いCodexにやらせてみたらPLaMo 3で関西弁変換できたっぽい
- 「そんなに心配する必要はありません。」
→「そんなに心配せんでええで。」
- 「明日は仕事があるので、遊びに行けません。」
→「明日仕事やから、遊びに行けへん。」
- 「このかばんは3000円ですが、あちらは5000円です。」
→「このかばんは3000円やけど、あっちは5000円や。」
解説を読んでも、本当なのか？？という感じですが…
先日話題になったLLMにウェイトを書き換えさせる方法でPLaMo 3 Translateをお嬢様言葉にできたっぽい。Opus使った。ぽいというのは今外出中で家で行われたことを直接確認できてないから…
—-
光は音よりも速く進みますわ。
エラーが解消されない場合は、システム管理者に問い合わせてくださいまし。
京都行きの電車はどれですの？
—-
変わったもの
•ファイル： 13個のうち、違うのは model.safetensors だけです。同梱コード、トークナイザー、設定はバイト単位で同じです。
•重み： 全 890,344,960 値のうち 1,534 値で、すべて ja（トークン番号 1385）の埋め込みの1行に収まっています。
•値の例： 先頭の4値は -1.18, 1.281, -1.281, 2.438 から -0.12, 1.586, -2.969, 3.281 に変わりました。行のノルムは 85.31 から 97.09 になっています。
Itaru Tomita / 冨田到 retweeted
先日話題になったLLMにウェイトを書き換えさせる方法でPLaMo 3 Translateをお嬢様言葉にできたっぽい。Opus使った。ぽいというのは今外出中で家で行われたことを直接確認できてないから…
—-
光は音よりも速く進みますわ。
エラーが解消されない場合は、システム管理者に問い合わせてくださいまし。
京都行きの電車はどれですの？
—-
変わったもの
•ファイル： 13個のうち、違うのは model.safetensors だけです。同梱コード、トークナイザー、設定はバイト単位で同じです。
•重み： 全 890,344,960 値のうち 1,534 値で、すべて ja（トークン番号 1385）の埋め込みの1行に収まっています。
•値の例： 先頭の4値は -1.18, 1.281, -1.281, 2.438 から -0.12, 1.586, -2.969, 3.281 に変わりました。行のノルムは 85.31 から 97.09 になっています。
Itaru Tomita / 冨田到 retweeted
Upgraded the models for Muse and Meta Al, bringing Active Memory and Active Focus from Muse Wonder to everyone. In Muse, these new specialized models are 2.4x less likely to forget critical info and are 67% better at self correcting and completing everyday tasks which means less repeating yourself and fewer dead ends.
More on the models: research.meta.ai/blog/specia…
Itaru Tomita / 冨田到 retweeted
Voters approved 43 of 44 local ballot measures to restrict data centers on Tuesday, many with more than 70% support.
Brad Carson, who heads the Anthropic-backed Public First super PAC network, says the issue will likely dominate the 2028 cycle.
Read more about the midterm blue wave: thein.fo/4iyTi1o
