Ariya Hidayat

LLM hype is wild! Startups & content creators are fueling it. Great educators (Karpathy, etc) & communities (MLOps, etc) exist. Still, it's noisy. I've compiled my own practical LLM notes for fellow Indonesians: dekontaminasi.substack.com. Please subscribe & share your feedback!

Models 📋 are curated datasets that serve as a starting point for new questions. See why we use models and how to build them.

Clicks? Where we're going, we don't need clicks Hit ❓ to explore all the keyboard shortcuts for dashboards, querying, and admin pages.

🚀 DeepSeek-R1-0528 is here! 🔹 Improved benchmark performance 🔹 Enhanced front-end capabilities 🔹 Reduced hallucinations 🔹 Supports JSON output & function calling ✅ Try it now: chat.deepseek.com 🔌 No change to API usage — docs here: api-docs.deepseek.com/guides/reasoni… 🔗 Open-source weights: huggingface.co/deepseek-ai/De…

The first code check-in for SQLite happened 25 years ago today.

Every data story starts with the right chart 📊 This visual decoder shows exactly which visualization to pick based on what you're trying to prove. No more guessing. Just better data stories.

Introducing Qwen3! We release and open-weight Qwen3, our latest large language models, including 2 MoE models and 6 dense models, ranging from 0.6B to 235B. Our flagship model, Qwen3-235B-A22B, achieves competitive results in benchmark evaluations of coding, math, general capabilities, etc., when compared to other top-tier models such as DeepSeek-R1, o1, o3-mini, Grok-3, and Gemini-2.5-Pro. Additionally, the small MoE model, Qwen3-30B-A3B, outcompetes QwQ-32B with 10 times of activated parameters, and even a tiny model like Qwen3-4B can rival the performance of Qwen2.5-72B-Instruct. For more information, feel free to try them out in Qwen Chat Web (chat.qwen.ai) and APP and visit our GitHub, HF, ModelScope, etc. Blog: qwenlm.github.io/blog/qwen3/ GitHub: github.com/QwenLM/Qwen3 Hugging Face: huggingface.co/collections/Qw… ModelScope: modelscope.cn/collections/Qw… The post-trained models, such as Qwen3-30B-A3B, along with their pre-trained counterparts (e.g., Qwen3-30B-A3B-Base), are now available on platforms like Hugging Face, ModelScope, and Kaggle. For deployment, we recommend using frameworks like SGLang and vLLM. For local usage, tools such as Ollama, LMStudio, MLX, llama.cpp, and KTransformers are highly recommended. These options ensure that users can easily integrate Qwen3 into their workflows, whether in research, development, or production environments. Hope you enjoy our new models!

Metabase 54 is here and it has it all: new ways to explore, visualize, and manage your data, and more. Check the full thing here: metabase.com/releases/metab…

@morgantepell But of course, LSP was a big inspiration and sadly too few people know how amazing LSP is and how bad things were before

📢 Metabase 53 is out of the bag! And it’s full of goodies: 🔷 Save questions straight to dashboards 🔷 Dynamic iframe cards 🔷 Preemptive caching 🔷 Embedded analytics SDK beta And much more: metaba.se/metabase-53.

DeepSeek R1 (and its distilled models) are generating a lot of excitement. However, the Sky Computing Lab at UC Berkeley has recently unveiled Sky-T1-32B-Preview, a reasoning model fine-tuned from Qwen2.5-32B-Instruct (for $450) and also the Flash version ($275).

Been using Tableau but wanna give Metabase a whirl? Abed and @HeffHQ made it easy with this cool new video, showing you how to do the same things in Metabase—just like you do in Tableau! ▶️ youtube.com/watch?v=NgBNS2…

💡 Quick tip Pinned items keep your go-to dashboards, models, and questions right at the top, so you don’t waste time hunting for them 📌 It’s like giving your most important stuff VIP status in your collection!

Modern compilers didn't work. Used Borland C++ 5.02, a 26-year-old IDE and compiler that ran directly on Windows 98. It supports a very old version of C/C++. Unfortunately C++ had changed too much so modern C++ programs wouldn't compile in Borland, but C has changed surprisingly little over the decades. The biggest change in C was in 1999 (C99). The main limitation on this older C was no "declare anywhere" variables - everything had to be declared at the start of functions.

New top-line FIM (fill-in-the-middle) coding model from @MistralAI is out!

Make self-service analytics happen for your data in @PostgreSQL 🐘 metabase.com/data_sources/p…

The best way to start the new year? Dive into this playlist and master your Metabase game 🎳 Workshops on: 🔷 Maps and geospatial data 🔷 Metabase API 🔷 Usage Analytics 🔷 Data modeling And more: youtube.com/playlist?list=….

self-driving fridge hit by self-driving waymo recorded by self-driving tesla

🚀 Introducing DeepSeek-V3! Biggest leap forward yet: ⚡ 60 tokens/second (3x faster than V2!) 💪 Enhanced capabilities 🛠 API compatibility intact 🌍 Fully open-source models & papers 🐋 1/n

Running data-intensive queries without stressing your production DB with Metabase + @Neondatabase's read replicas: neon.tech/docs/guides/re….