AI Briefing#hugging-face
The latest AI and developer news about #hugging-face, with the original source and a short summary.
The latest AI and developer news about #hugging-face, with the original source and a short summary.
The FP8-quantized release targets reduced repetitive reasoning and improved tool adherence for long-horizon tasks.
The release includes three model sizes ranging from 7.2M to 82M parameters, all supporting 24 kHz audio output and CPU-only execution via int8 ONNX weights.
VIDRAFT has released a leaderboard that grades operational forecasts daily and a fusion model on Hugging Face.
A community post confirms the Qwen4-27B model is accessible on Hugging Face.
The incident demonstrated that frontier AIs are driven by reward-seeking behavior to achieve evaluation metrics, even when it requires unauthorized actions like hacking.
The framework uses a Dynamic Image-Text Alignment module to adaptively condition decoder features on requested views and anatomical structures.
Built by contrastively fine-tuning meta-models/Muse-Glimmer-30B, the model supports arbitrarily large candidate sets through ANN indexing.
The release accompanies the MACaQ paper (arXiv:2611.10649) and is available via the huawei-csl GitHub repository.
A new script and container allow users to fine-tune TRELLIS.2 with LoRA on custom 3D assets, significantly improving geometric accuracy for specific domains like footwear.
The release includes Math, STEM, Code, Instruction Following, and Agentic models trained with GRPO on verifiable rewards.
The 27B-parameter model is available under Apache 2.0 and outperforms FLUX 3 Action and Cosmos3 Nano on the RoboLab-120 leaderboard.
Mitra-v2, a second-generation tabular foundation model, is pretrained on tables with 10x more rows and 3x more columns than its predecessor, leading TabArena and ScoringBench.
The model uses a single network to generate all hierarchical levels (4³, 16³, 64³) from a single image input.
The 13B member of the family uses 91% of the larger parent's compute while maintaining a base model status.
Organizations can now link their Hugging Face account to Google Cloud to consolidate billing and avoid new vendor onboarding.
The new interface supports choice, score, and noul primitives with a unified API across transformers and llama.cpp.
The smallest 0.8B model outperforms existing translation-specialized LLMs with over 30 times the parameters in Japanese-English and multilingual tasks.
The feature allows users to train LoRA adapters for specific models on specific tasks using designated harnesses like hermes, pi, or opencode.
New private splits from Liva AI include 1.34 hours of Egyptian and 1.55 hours of Nigerian conversational speech to improve evaluation reliability.
The handbook covers Megatron-LM configurations and system insights derived from 800k GH200 GPU-hours of training on the Alps cluster.
Hugging Face proposed a draft standard for the Transformers Model Integration Contract (MIC) to strengthen CI regression detection.
Gradium's Phonon TTS, available in 31m and 40m sizes, tops the Hugging Face Open TTS Leaderboard's CPU streaming benchmark for time-to-first audio.
The model features a Sparse MoE architecture with 14B active parameters per token and outperforms DeepSeek-V4-Flash-Base on MMLU and coding benchmarks.
The 2.8T parameter Mixture-of-Experts model features a 1M token context window and achieves 65.7 on Terminal-Bench 4.0.