Kadabra Gemma4-E2B SFTJv2-SFT2

A community fine-tune of Google DeepMind's Gemma 4 E2B open-weight multimodal model, published by Kadabra on Hugging Face with GGUF packaging for efficient local inference. It is an instruction-tuned checkpoint oriented toward resource-constrained deployment rather than a hosted AI service. Source: supabase.