MaAI updates its English VAP model
On September 6, 2026, MaAI said its English VAP model was trained on a larger scale of data. The project predicts conversational cues and head movements in real time.
Read articleKnowledge you can apply
Radar
Daily bookmarks: relevant news, facts and discoveries.
OmniVoice: speech synthesis and cloning in over 600 languages
A September 6, 2026 publication describes OmniVoice, a Python text-to-speech project with voice cloning, zero-shot synthesis, and generation controls.
Read articleripwire builds structural repository context for coding agents
Published on September 6, 2026, ripwire is a dependency-free C++23 CLI and MCP server that analyzes 21 languages and returns relevant symbols, relationships, and tests within a token limit.
Read articlePhoenix-VAD: semantic end-of-speech detection in streaming
Published on September 4, 2026, the article presents an LLM-based model for semantically detecting the end of speech in full-duplex voice interactions, using sliding-window training.
Read articleFRMs apply recurrent flows to structured reasoning
A post dated September 3, 2026 describes flow-based reasoning models that refine outputs on discrete tasks such as Sudoku.
Read articleMicrosoft introduces the MAI-Transcribe-2 speech model
Announced on September 3, 2026, MAI-Transcribe-2 is a transcription model reporting 2.0% AA-WER, about 411× real-time speed, and support for 60 languages.
Read articleLanguage models and control of their own attention
A paper explores whether language models can control their own attention. The available material introduces the topic but does not report methods or results.
Read articleLaunch announces streaming ASR with diarization
A September 3, 2026 publication announces a streaming speech recognition model combining transcription and diarization. The text provides no performance metrics or details.
Read article163 procedures for research agents across 16 fields
Published on September 2, 2026, the article presents an open library of research procedures and highlights domain-specific choices and caveats.
Read articleDepth-wise batching for recurrent models
A September 2, 2026 post introduces depth-wise batching: overlapping requests at different recurrent depths to potentially improve utilization with adaptive compute.
Read articleNoRA changes LoRA initialization
A September 2, 2026 post describes NoRA, which normalizes the columns of LoRA matrix A at initialization, and reports higher average results than several adaptation methods in SFT and RLVR.
Read articlePhoton 2.1 expands streaming speech recognition options
The update adds automatic speech recognition for three models and support for the NVIDIA B200 GPU. The company says it outperformed each model’s reference engine in comparable tests.
Read articleAstra and the challenges of monitoring recurrent reasoning
A September 2, 2026 publication says Astra uses “recurrent depth,” an approach that researchers reportedly fear could hinder monitoring and affect costs and performance.
Read articleZvec team launches local search tool zg
On September 2, 2026, Alibaba's Zvec team said it released zg as open source, a local search tool for developers and AI agents.
Read articleChecks reduce errors in AI-generated papers
Published on September 1, 2026, this Radar item summarizes an arXiv paper: checking manuscript claims against execution logs reduced the reported rate of serious hallucinations to 4%.
Read articleHy4 Preview reduced to 214 GB with 1.25-bit quantization
A Tencent publication says Sherry quantization reduces Hy4 Preview from 1.5 TB to 214 GB at 1.25 bits per weight, and describes running the model on GPUs distributed across multiple machines.
Read articleCERN slides introduce LLM training
A post dated August 31, 2026, says slides from a CERN talk on LLM training are available as a PDF and that an online recording is available.
Read articleArticle reviews policy gradients and variants
An article on reinforcement learning foundations derives the policy gradient algorithm and discusses variants, providing review material for engineers.
Read articleKV Cache: Queries Are Not Kept
A post dated August 30, 2026 distinguishes the one-time use of queries from the retention of keys and values in the KV cache during inference.
Read articleHy4: calibration-guided low-bit compression
A post reports reducing the Hy4 preview from 1.5 TB to 200 GB with per-layer precision settings, and compares results on two benchmarks.
Read articleRota Nacional
Bring privacy into your workflow.
30 days, no card, with a starting quota. After that, Pix credit from R$ 10,00.