Open Source

Auto Added by WPeMatico

Sakana AI Introduces KAME: A Tandem Speech-to-Speech Architecture That Injects LLM Knowledge in Real Time

The fundamental tension in conversational AI has always been a binary choice: respond fast or respond smart. Real-time speech-to-speech (S2S) models — the kind that power natural-feeling voice assistants — start talking almost instantly, but their answers tend to be shallow. Cascaded systems that route speech through a large language model (LLM) are far more

Sakana AI Introduces KAME: A Tandem Speech-to-Speech Architecture That Injects LLM Knowledge in Real Time Read More »

Mistral AI Launches Remote Agents in Vibe and Mistral Medium 3.5 with 77.6% SWE-Bench Verified Score

Mistral AI has been quietly building one of the more practical coding agent ecosystems in the open-source/weights AI space, and they are shipping its most significant infrastructure upgrade yet. Mistral team announced remote agents in Vibe, its coding agent platform, alongside the public preview of Mistral Medium 3.5 — a new 128B dense model that

Mistral AI Launches Remote Agents in Vibe and Mistral Medium 3.5 with 77.6% SWE-Bench Verified Score Read More »

Qwen AI Releases Qwen-Scope: An Open-Source Sparse AutoEncoders (SAE) Suite That Turns LLM Internal Features into Practical Development Tools

Large language models are remarkably capable, yet frustratingly opaque. When a model misbehaves — generating responses in the wrong language, repeating itself endlessly, or refusing safe requests — AI devs have very few tools to diagnose why it happened at the level of internal computations. That’s the problem Qwen-Scope is built to solve. Qwen Team

Qwen AI Releases Qwen-Scope: An Open-Source Sparse AutoEncoders (SAE) Suite That Turns LLM Internal Features into Practical Development Tools Read More »

Moonshot AI Open-Sources FlashKDA: CUTLASS Kernels for Kimi Delta Attention with Variable-Length Batching and H20 Benchmarks

The team behind Kimi.ai (Moonshot AI) just made a significant contribution to the open-source AI infrastructure space. The research team has made a significant contribution to the open-source AI infrastructure space. They released FlashKDA (Flash Kimi Delta Attention), a high-performance CUTLASS-based kernel implementation of the Kimi Delta Attention (KDA) mechanism. The FlashKDA library is available

Moonshot AI Open-Sources FlashKDA: CUTLASS Kernels for Kimi Delta Attention with Variable-Length Batching and H20 Benchmarks Read More »

Microsoft Research’s World-R1 Uses Flow-GRPO and 3D-Aware Rewards to Inject Geometric Consistency Into Wan 2.1 Without Architectural Changes

Video foundation models can paint a beautiful frame. They are still notoriously bad at remembering it. Push the camera through a corridor in Wan 2.1 or CogVideoX and walls warp, objects morph, and details vanish — the giveaway that these models are fitting 2D pixel correlations rather than simulating a coherent 3D scene. A team

Microsoft Research’s World-R1 Uses Flow-GRPO and 3D-Aware Rewards to Inject Geometric Consistency Into Wan 2.1 Without Architectural Changes Read More »

IBM Releases Two Granite Speech 4.1 2B Models: Autoregressive ASR with Translation and Non-Autoregressive Editing for Fast Inference

IBM released two new open speech recognition models— Granite Speech 4.1 2B and Granite Speech 4.1 2B-NAR — and they make a compelling case for what a ~2B-parameter speech model can do. Both are available on Hugging Face under the Apache 2.0 license. The pair targets a specific problem that enterprise AI teams know well:

IBM Releases Two Granite Speech 4.1 2B Models: Autoregressive ASR with Translation and Non-Autoregressive Editing for Fast Inference Read More »

Qwen Team Releases FlashQLA: a High-Performance Linear Attention Kernel Library That Achieves Up to 3× Speedup on NVIDIA Hopper GPUs

The race to make large language models faster and cheaper to run has largely been fought at two levels: the model architecture and the hardware. But there is a third, often underappreciated frontier — the GPU kernel. A kernel is the low-level computational routine that actually executes a mathematical operation on the GPU. Writing a

Qwen Team Releases FlashQLA: a High-Performance Linear Attention Kernel Library That Achieves Up to 3× Speedup on NVIDIA Hopper GPUs Read More »

Meta FAIR Releases NeuralSet: A Python Package for Neuro-AI That Supports fMRI, M/EEG, Spikes, and HuggingFace Embeddings

Researchers at Meta’s FAIR lab have released NeuralSet, a Python framework designed to eliminate one of the most persistent bottlenecks in Neuro-AI research: the painful, fragmented process of getting brain data into a deep learning pipeline. https://kingjr.github.io/files/neuralset.pdf The Problem: Neuroscience Data Is Stuck in the Pre-Deep-Learning Era Neuroscience already has excellent, battle-tested software. Tools like

Meta FAIR Releases NeuralSet: A Python Package for Neuro-AI That Supports fMRI, M/EEG, Spikes, and HuggingFace Embeddings Read More »

smol-audio: A Colab-Friendly Notebook Collection for Fine-Tuning Whisper, Parakeet, Voxtral, Granite Speech, and Audio Flamingo 3

Audio AI has had a breakout year. Automatic speech recognition has gotten dramatically better with models like OpenAI’s Whisper variants, NVIDIA’s Parakeet, and Mistral’s Voxtral. Audio understanding stepped forward with models like NVIDIA’s Audio Flamingo 3. Dialogue-grade text-to-speech arrived via Nari Labs’ Dia-1.6B. And Meta shipped the Perception Encoder Audiovisual (PE-AV), a multimodal encoder capable

smol-audio: A Colab-Friendly Notebook Collection for Fine-Tuning Whisper, Parakeet, Voxtral, Granite Speech, and Audio Flamingo 3 Read More »