Tag: hugging face

IBM updates Granite Switch adapters, leaving base model untouched

Written by Felipe Santos on November 2, 2026. Posted in AI.

IBM has released Granite Switch 4.1.1, a patch release for its modular Switch architecture, alongside an updated speech recognition model and an updated Decision adapter. The notable part is how quickly the update shipped: apart from training the new components, nothing was recomputed, and the base model is untouched. Granite Switch 4.1, released just over a week earlier, showed that the Switch architecture lets users configure a language model with specialized, cache-efficient adapters, adding new expertise on a single endpoint. Version 4.1.1 is a patch in the ordinary software sense. It...

Grumpy garage door talks back using NVIDIA Nemotron

Written by Felipe Santos on October 31, 2026. Posted in AI.

The author built MonsterHouse as an interactive Halloween projection. He listens to trick-or-treaters, answers in the voice of a grumpy old pirate ship, and softens when visitors remember their manners. The project runs on the NVIDIA Nemotron Voice Agent with Nemotron Speech ASR, Nemotron 3.5 Bolt E4B 4GB, Magpie TTS Zeroshot and Audio2Face-3D, all on a single workstation. The author wanted the kids visiting the house at Halloween to interact in a fun and unique way. The house was going to be in charge and giving out the treats, but more grumpy than scary. Somehow, the solution was to give...

NVIDIA Packs 1 Petaflop AI Into Thin Laptops

Written by Felipe Santos on October 28, 2026. Posted in Hardware.

NVIDIA's RTX Spark laptops, built on a single Blackwell GPU and Grace CPU superchip with up to 128GB of unified memory, are now sold by ASUS, Dell, HP, Lenovo, Microsoft and MSI. The systems deliver up to 1 petaflop of FP4 AI performance in premium Windows laptops, bringing the CUDA platform, RTX graphics and what NVIDIA calls a personal AI supercomputer into machines designed to travel. The launch comes with a set of software updates for local AI, gaming and creative work, including a new open speech model, a new Liquid AI model family, a new resource management control and a launch-day...

NVIDIA’s open model transcribes 40 languages, labels speakers

Written by Felipe Santos on October 28, 2026. Posted in AI.

NVIDIA has released Nemotron 3.5 Transcribe, an open-weight, promptable speech model that turns recorded audio into text and can also attribute speakers, translate, and answer questions about the recording. Most transcription tools produce a word-for-word record and stop there. A transcript can capture every spoken word and still leave people wondering who said what, how to translate it, or how to spell domain jargon. Multilingual calls, fast-paced meetings, and medical dictation each need different output. Transcribe puts these tasks in one checkpoint and lets developers choose the output...

IBM details training of Granite 4.2 reasoning models

Written by Felipe Santos on October 24, 2026. Posted in AI.

Authors: Granite Team, IBM Overview Granite 4.2 is IBM's first family of dense, decoder-only reasoning LLMs, released in three sizes: 3B, 8B, and 30B. These models are post-trained from Granite-4.1 base models. The Granite-4.1 base models were pre-trained from scratch on roughly 15T tokens with a five-phase strategy that extends the context window to 512K tokens. They were then supervised fine-tuned on chain-of-thought, reasoning, and agentic-trajectory data, and post-trained with a multi-stage reinforcement learning pipeline. That pipeline includes agentic RL, where the 8B and 30B models...

Database-checked rewards lifted 27B agent’s dependable tasks 2.8x

Written by Felipe Santos on October 24, 2026. Posted in AI.

Microsoft and Hugging Face's ThinkingBox-Training, now on GitHub, uses the same executable database checks that grade agents as training rewards. In one run, Qwen3.8-27B's average success rose from 51.70% to 60.78% after 50 GRPO steps, and tasks it completed correctly on all twenty attempts rose from 38 to 107. This is a follow-up to the team's first post on ThinkingBox, a benchmark that tests whether an agent's work actually changed the database it was supposed to change. In that earlier post, a model closed a customer's ticket as resolved while the carrier exception behind it was still...

PyTorch 2.15 speeds Mac MoE decoding up to 6.5×

Written by Felipe Santos on October 23, 2026. Posted in AI.

PyTorch 2.15 brings large speedups and new capabilities to Apple silicon Macs. The release is from the PyTorch team at Hugging Face, which works on the MPS backend to make the Mac a first-class place to run and train PyTorch models. At the start of 2026, running even small models on MPS often meant working around unsupported operations, performance cliffs and unexpected memory growth. The team runs real models from the Hugging Face Hub across language, image and video, finds where MPS falls short, and fixes it upstream in PyTorch, increasingly by replacing MPSGraph with dedicated Metal...

Meta’s Muse Spark Open runs 42B active parameters

Written by Felipe Santos on October 23, 2026. Posted in AI.

Muse Spark Open is an open-weight model from Meta. It is a mixture-of-experts model that accepts text, images, audio, and video, with approximately 816 billion parameters, around 42 billion active per token, and a context window of up to one million tokens. It generates text, giving developers a common interface for asking questions about different kinds of input. Following Muse Glimmer, Muse Spark Open introduces a mixture-of-experts architecture at a much larger scale, and it is the most capable US open-weights model to date. Meta trained Muse Spark Open for agentic coding: reading a...

NVIDIA Cosmos delivers up to 2x faster inference

Written by Felipe Santos on October 22, 2026. Posted in AI.

NVIDIA has expanded its Cosmos platform with new training tools, simulation checkpoints and Cosmos Data Factory blueprints, aimed at developers building world models, which help physical AI systems understand and simulate the real world. NVIDIA says the platform ranks in the top five on six different open model leaderboards. It pairs the frontier open models with data, training and optimization tools and recipes so builders can scale training data, adapt models, test behavior and optimize performance before deployment. .abn-galeria .uk-slidenav, .abn-galeria...

Tag: hugging face

Privacy Settings

Functional

Preferences