3 ms·
The official announcement of the QAT models happened on Friday 18th, two days ago. It looks like they uploaded them to HF in advance of that announcement: https
by simonw 2y ago
The official announcement of the QAT models happened on Friday 18th, two days ago. It looks like they uploaded them to HF in advance of that announcement: https://developers.googleblog.com/en/gemma-3-quantized-aware-trained-state-of-the-art-ai-to-consumer-gpus/ https://developers.googleblog.com/en/gemma-3-quantized-aware...
The partnership with Ollama and MLX and LM Studio and llama.cpp was revealed in that announcement, which made the models a lot easier for people to use.