AI updates
Gemma 3n: on-device multimodal models with "effective" sizes
Gemma 3n arrived across open-source tools; its E2B and E4B models have 5B and 8B real parameters but need only 2B- and 4B-class memory, and accept image, audio, video and text.
- "E2B" means effective 2B: 5B real parameters, but a 2B-class memory footprint, running in about 2GB of GPU RAM.
- It takes image, text, audio and video, with a MobileNet-V5 vision encoder and an audio encoder.