Gemma 4 QAT models in NVFP4 with FP8-calibrated KV cache using multilingual and tool-use data. 31B, 26B and 12B use imatrix_mse.
Yasuhiro Ohno
yasu-oh
AI & ML interests
None yet
Recent Activity
updated a model about 10 hours ago
yasu-oh/gemma-4-12B-it-qat-NVFP4 updated a model about 10 hours ago
yasu-oh/gemma-4-26B-A4B-it-qat-NVFP4 updated a model about 10 hours ago
yasu-oh/gemma-4-31B-it-qat-NVFP4Organizations
None yet