Y Hacker News new | ask | show | jobs

user: victormustar created: 2013-02-22 karma: 924 https://twitter.com/victormustar submissions: Designing the HF CLI for both humans and agents 2 points | 0 comments Nemotron 3 Ultra: Open Moe Hybrid Mamba-Transformer for Agentic Reasoning [pdf] 23 points | 2 comments Ideogram 4.0: A 9.3B open-weight image model 3 points | 0 comments China's lab-grown diamonds emerging as a surprising beneficiary of the AI boom 3 points | 0 comments Nvidia Dynamo Snapshot: Fast Startup for Inference Workloads on Kubernetes 2 points | 0 comments Carbon: open-source DNA foundation models with training code and data pipeline 2 points | 0 comments Physics-intern: an autonomous agentic framework for physics research 3 points | 0 comments Bonsai 1.7B in the browser: a 290MB 1-bit LLM on WebGPU 8 points | 0 comments Hugging Face Storage Buckets: Mutable, non-versioned object storage at $12/TB 18 points | 0 comments 0 points | 0 comments Zuna: A 380M-parameter foundation model for EEG signals 2 points | 1 comments 0 points | 0 comments Show HN: Nanbeige 4.1-3B running in the browser via WebGPU 6 points | 1 comments FinePDFs: Liberating 3T of the finest tokens from PDFs 2 points | 0 comments 0 points | 0 comments Aligning to What? Rethinking Agent Generalization in MiniMax M2 2 points | 0 comments Nvidia Publishes Complete Evaluation Recipe for Nemotron 3 Nano 2 points | 0 comments Using Claude Code to Fine-Tune Open Source LLMs 1 points | 0 comments Eiffel Llama: Open-Source Replication of Claude's Golden Gate Experiment 1 points | 0 comments Porting nanochat to Transformers: an AI modeling history lesson 1 points | 0 comments Kimi-K2-Thinking: open weights LLM with frontier performance 7 points | 0 comments Llama.cpp launches official WebUI for local LLMs 4 points | 0 comments Granite-4.0-Micro: a 3.4B parameter LLM that runs in the browser 4 points | 0 comments Drag-and-Drop LLMs: Zero-Shot Prompt-to-Weights 1 points | 0 comments How I Use LLMs to Write 20 points | 10 comments Tiny Agents in Python: an MCP-powered agent in ~70 lines of code 2 points | 0 comments An MCP-powered agent in 50 lines of code 5 points | 0 comments The 4 Things the Qwen-3's Chat Template Teaches Us 2 points | 0 comments Hugging Face acquires open source robot startup Pollen Robotics 2 points | 0 comments Gemini Co-Drawing: Doodle on a Canvas with Google Gemini 2.0 2 points | 0 comments