Loading activity…
@radixark
SHIP AI FOR ALL. From the core maintainers of SGLang @sgl_project. Building AI infrastructure for inference and training.
AI
Reflection says the release includes Mirror CLI, its open-source training harness, and collaborations with 40-plus partners.
AI
finkd says the model runs well on new Mac Studios, crediting work with Apple and John Ternus.
AI
NVIDIA says Step 5 Preview runs at nearly 8K tokens per second per GPU on GB300 using NVFP4.
AI
The release announcement says models that learn from each other, such as a model and its grader, can train together.
Technology
DeepSeek says its KV cache, which lets the model reuse earlier calculations, needs a quarter of the high-bandwidth memory and an eighth of the SSD storage required by the previous generation's cache.
AI
Company highlights post-training design supporting vision-language and diffusion models.
AI
The framework aims to simplify debugging and improve hardware efficiency when scaling RL training for LLMs and multimodal models.
AI
Mark Zuckerberg posted that Meta opened weights for a local model with another release planned soon.
