Liquid AI's new DSpark drafter brings speculative decoding to LFM2.5-VL-3B, delivering up to 3.13x faster decoding on Apple Silicon with lossless output.
A new study pits TypeSafe's typed classifier against flash-tier LLM judges on rubric grading, finding huge speed and cost wins but correlated errors that break cascades.
An open-source desktop app funnels Slack, Gmail, GitHub, Jira and Notion pings through local LLMs and turns each into an approve-or-dismiss Action Card.
A new benchmark shows long-context LLMs handle retrieval fine but crumble on tasks like finding contradictions, breaking common architecture assumptions.
Meta's new Muse Realtime Avatar streams expressive video of any character in live conversation, distilling a 120-step diffusion teacher into a 2-step causal student.
Tencent Hunyuan researchers show that retuning the learning rate lets you scale RL batch sizes for 29% faster training and 2.29x generation throughput.
A Tsinghua team gives the Unitree G1 humanoid pressure-sensing insoles and phase-aware rewards, cutting landing impact by 48.8% and impact noise by 30.1 dB.
Featherless AI's open-source Simple Jev turns any Hugging Face model into a typed classifier by reading next-token logits, skipping JSON generation entirely.
Ant Group's inclusionAI drops a 6B MIT-licensed text-to-image model that beats every other open-weights contender on UI mockups, dashboards and posters.
A community fine-tune of the Qwen-Image-2.1 VAE decoder wipes out checkerboard artifacts and boosts perceptual quality with just 7.5M trainable parameters.