Model Conversion Colab Issue

[kuaashish]

This issue pertains to the Model Conversion script, documented here and raised google-ai-edge/mediapipe#5811. When converting supported models in CPU mode, the conversion does not work as expected for all models.

Specifically, issues occur with MediaPipe Falcon 1B, StableLM 3B, Phi-2, and Gemma 2B CPU models. However, the conversion works as expected on GPU for Falcon 1B, StableLM 3B, Phi-2, Gemma 2B, and Gemma 7B.

[kuaashish]

Hi @woodyhoko, @PaulTR,

Could you please look into this issue?

Thank you!!