にゃんこまる🐾 (@lero003) header
にゃんこまる🐾 (@lero003) profile picture

修身斉家治国平天下 / to have or to be

edo
Joined May 2009
Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)Photo by にゃんこまる🐾 (@lero003)
ま、なんとかなるでしょ なるようになる、若しくはなるようにしかならない
13
1,899
にゃんこまる🐾 retweeted
JUST IN: $1,800,000,000,000 erased from US stock market today.
26
511
41
4,029
106,518
Qwenがコンシューマ向けLLMで後追い始めるとは2ヶ月前には思ってなかった感 もはやエッジデバイス向けもハイスペック向けでも、ちょっと後追い始まってるかも
22
セキュリティ製品を自作する・・・?とかよくわからない方向の発想してるけど、面倒か 最近のLLMの能力高すぎて、作りたいと思ったのはだいたいできてしまうのでネックは人間の発想力になってる
16
にゃんこまる🐾 retweeted
"If AI writes the code how will you understand it so you can debug it?!" Brother, I read code I wrote six months ago and said if I ever met the author I'd kill him.
131
845
64
22,144
589,502
にゃんこまる🐾 retweeted
From my initial tests, Inkling 2 Small (26B-A3B) is in spirit the Qwen3.8 35B-A3B that we never got jaggedly close to Qwen3.8 27B (better in some things and worse in others), while way faster as an MoE. and it's the first looped transformer I'm exploring 🤓
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/ink…
3
3
109
3,604
にゃんこまる🐾 retweeted
This prediction of mine wasn't accurate and for that I apologize Inkling 2 Mini gives you Opus 4.5 intelligence and capabilities on an RTX 3080 (~300$ USD a year ago) Local AI is winning BIG TIME
Prediction We will have Claude Code + Opus 4.5 quality (not nerfed) models running locally at home on a single RTX PRO 6000 before the end of the year
48
62
2
1,225
33,363
年内に来そう
15
にゃんこまる🐾 retweeted
Props to Anthropic for popping the AI bubble with their IPO
28
25
3
597
15,096
お?? 26.4B, 3.4B activeのInklingですと、Inkling 2 Mini 推論のループがおもしろいっぽいけど、これにEngramとか含めて改良してったら、パーソナルマシンで動く怪物ちゃんも夢じゃなくなってきた感
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/ink…
35
まあmetaだから・・・(使ってない
진짜 악재가 끊임 없이 쏟아지네 이번엔 메타의 muse ai 앱스토어 무료 2위까지 오르며 1,000만 다운로드를 넘긴 메타의 AI 에이전트 '뮤즈(Muse)'가 내부에서 대형 사고를 쳤음 1. 날씨 물어보던 직원에게 뜬금없이 남의 은행 잔고와 투자 계좌 금액을 띄우고, 종단간 암호화 메신저인 시그널(Signal) 사적 대화까지 무단으로 읽어 답장 알림을 보냄 2. 심지어 사용자 동의 없이 환불 불가 에어비앤비 숙소를 예약하거나, 월 200달러짜리 대신 월 888달러짜리 펫보험에 무단 가입시킨 사례까지 내부 보고로 확인됨 3. 결국 애플이 macOS 업데이트를 통해 뮤즈의 데스크톱 접근을 전격 차단했고, 동시에 애플은 브라우저·앱·터미널을 직접 다루는 자체 AI 에이전트 개발 채용 공고까지 띄움 이제 AI 에이전트 시장의 승부처는 '얼마나 똑똑한가'가 아니라 'OS 쥔 애플이 누구의 권한을 열어줄 것인가'로 넘어가는 분위기임
138
wacci初めて聞いたけど、やさしい世界だ
78
魔法使いの夜、気になってたけど本気のファンの熱意が半端ないTLになってきたので少し忘れようと思う
52
にゃんこまる🐾 retweeted
fuck you
never trust a cat. they are full of bullshit
121
3,397
122
22,651
430,112
Fate/Zero放送当時の思い出だよ
35
4,019
417
13,614
Re-installed Hermes a few days ago for a new project.... And it made me realise how "retarded" it is out of the box. Even with SOTA models behind it. Ya'll take Grok @bot for granted; the ability to jump start working without needing to configure 10 different skills, having plug-n-play marketplace and connectors, sharable bots is EXACTLY what's going to win normies and non-technical users. "But you can just ask your agent to configu..." That's the thing tho, normal non-technical users are NOT going to ask Hermes (or OpenClaw) to do it. They won't even know where to start asking, because, they start using those agents like they use ChatGPT Chat. Yet, I think Hermes is EXTREMELY POWERFUL AI AGENT and I will use it in parallel to Grok Bot. For multiple reasons. The Data is YOURS. The flexibility is definitely there, especially with model choices.
66
5
5
161
20,052
あー今自分が欲しいのがgrok botのモデル選択やローカル起動だったのに、これで気づいた openbot改めて使ってみようかな…
Re-installed Hermes a few days ago for a new project.... And it made me realise how "retarded" it is out of the box. Even with SOTA models behind it. Ya'll take Grok @bot for granted; the ability to jump start working without needing to configure 10 different skills, having plug-n-play marketplace and connectors, sharable bots is EXACTLY what's going to win normies and non-technical users. "But you can just ask your agent to configu..." That's the thing tho, normal non-technical users are NOT going to ask Hermes (or OpenClaw) to do it. They won't even know where to start asking, because, they start using those agents like they use ChatGPT Chat. Yet, I think Hermes is EXTREMELY POWERFUL AI AGENT and I will use it in parallel to Grok Bot. For multiple reasons. The Data is YOURS. The flexibility is definitely there, especially with model choices.
1
55
にゃんこまる🐾 retweeted
Nan je jure sur toutes les dingueries que Gintama m’a proposé c’est de loin la pire il a couché avec Otose la 🥀
30
492
61
10,113
1,657,925
展開がバグってるけど 構造的にはどうなるこれ
🚨 THIS IS ACTUALLY INSANE. Jensen Huang is reportedly looking to raise $50 BILLION to launch a brand new semiconductor company. He built Nvidia from a Denny’s booth into a $4T empire. The board says he’s not coming back. But with his relationships, he could plug into the entire AI ecosystem on day one. Imagine the customers lining up before the first chip even ships... Could Jensen really launch a real competitor to $NVDA, lock in $50B, and start taking market share from day one? Hell yeah. If you don’t keep me here, someone else will.
Readers added context they thought people might want to know
No credible reports indicate Jensen Huang is seeking to raise $50 billion to start a new semiconductor company. He resigned from Nvidia amid scandals and offered to return as its CEO, which the board has rejected. newsweek.com/nvidia-shuts-d… taipeitimes.com/News/biz/archi… x.com/JensenHuang/st…
36