search
Gemma
Trends
- 1Google releases EmbeddingGemma 2 open lightweight multimodal embedding model●EmbeddingGemma 2: An open, lightweight multimodal embedding model
Google has announced EmbeddingGemma 2, an open and lightweight multimodal embedding model aimed at developers building search, retrieval and AI applications. The release is drawing attention among developers for combining small size with multimodal capabilities and an open license, making it practical to run on modest hardware. Discussion centres on how it compares with other open embedding models and what it means for on-device and low-cost AI systems.
Repos
- FeSens/openTPU An open-source AI accelerator, developed by AI: RTL, ISA, simulator, compiler and profiler in one repo. Runs Qwen3, LFM2
- SamsungLabs/LittleBit Official implementation of LittleBit (NeurIPS 2025) and its follow-up LittleBit-2 (ICML 2026)
- firelex/jeff Millisecond decisions, any domain: a 0.8B open "System 1" model that picks between your options with calibrate