convert and ios
#7
by
l0d0v1c
- opened
Very impressive work! Two questions:
1-I tried to convert fintuned phi-3 models with optimum but I only get a 15Go onnx from quantized model. Is there a script to convert models?
2-On Ipad it is possible to enable the webgpu settings. It's only load the first part of the model but not the second one. Maybe a question of RAM?