もか
@mt03kkk18wu
物流系スタートアップでChief Everything Officerをやっていて、エーアイのお世話やさんを自称しているカワウソです。
ベクトル検索と画像認識が比較的得意です。
ダーツとウイスキーと野球とアイドルと動物が好きです。
KagglerでしたがGMにはなれませんでした。
1K Following 1.5K Followers
JUST IN: Anthropic CEO reportedly increasingly concerned that new hires are working there for a paycheck, as opposed to the company’s mission.
Spotted Codex making friends at the OpenAI office
Rolling Shutter Relative Pose Estimation Made Practical
@majti89
tl;dr: affine correspondences for rolling shutter, because you need 20 PC otherwise. You also can get them from dense matcher like RoMa.
P.S. (or we can ignore RS first?)
Show more
Self-Managed OAuth is now available to all developers on Cloudflare. Here's how we executed a zero-downtime migration of our core OAuth engine to make it happen.
We've really normalized the fact that Ronaldo jumped so high his knees were level with his opponent's head.
We’ve designed and built our first AI chip: Jalapeño.
Designed from the ground up by OpenAI and brought to production with @Broadcom, Jalapeño is purpose-built for the LLM workloads powering ChatGPT, Codex, the API, and future agentic products.
Chips are foundational to the AI economy. Building our own expands our full-stack platform from products to models to infrastructure, and will help us scale intelligence, serve more people, and expand access to AI.
Show more
OCR but with a working memory...?!
This paper, Unlimited OCR, replaces decoder full self-attention with Reference Sliding Window Attention, giving the model a working memory where each token attends to the fixed visual and prompt references plus only the most recent output tokens, making decode KV cache constant instead of linear in generation length.
Combined with DeepEncoder’s 16x visual compression, this working-memory-style attention enables one-shot multi-page OCR with stable memory and latency, while improving OmniDocBench performance over DeepSeek OCR.
Show more
Mistral OCR 4 reading a handwritten Henri Poincaré letter from 1905.
Historical manuscripts usually break OCR models. This one held up well.
Solid release from @MistralAI.
Introducing Mistral OCR 4. It creates structure with bounding boxes, block classification, and inline confidence scores in 170 languages. 🧵👇
For the first time, robots now outnumber humans at Figure
