add quantization_config.ignore=1 (downstream audit fix)

Add files using upload-large-folder tool

Replace INT4 vision with BF16 originals (model-00001-of-00001.safetensors)

Replace INT4 vision with BF16 originals (config.json)

Replace INT4 vision with BF16 originals (model.safetensors.index.json)

Replace INT4 vision with BF16 originals (model-vision.safetensors)

Update vision status: untestable due to server crash

Add known limitations (vision status)

Add model card for Gemma 4 26B MoE AWQ 4-bit

Gemma 4 26B MoE AWQ: forced-routing GPTQ calibration for all 128 experts, 30 tok/s on RDNA4

initial commit