Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
Microsoft has released the Phi-3 series of language models under the MIT license, including Phi-3 small (7B) and medium (14B) models. These models are said to outperform competitors like Llama 3, GPT-3.5, and Command R+. Additionally, a new multimodal Phi-3-vision model has been introduced.
Microsoft has released the remaining Phi-3 models! This includes the 7B and 14B models This also include a previously unannounced multimodal phi model! Phi-3-small (7B): https://t.co/RBdUsOufFA Phi-3-medium (14B): https://t.co/yf1EDeTWrg Phi-3-vision: https://t.co/jz4ZR6re8N https://t.co/LQA2hGPDQj
A new series of Phi-3 mini, small & medium models (MIT license (3.8B - 7B - 14B) Long-context Phi-3-mini-128k-instruct: https://t.co/ZrgdzOIBuV Phi-3 small 128k: https://t.co/I2C2cLEAAf Phi-3 medium 128k: https://t.co/Z627FRj6Mx Multimodal: Phi-3-vision-128k-instruct:… https://t.co/MqESI1oiz0
✨ Announced at #MSBuild, the latest @microsoft Phi-3 family of SLMs are GPU-optimized with NVIDIA TensorRT-LLM and available as NVIDIA NIM inference microservices that can be deployed anywhere. ➡️ https://t.co/0ZEBq5JjNL https://t.co/9XqQQxIfgp
