Please help me understand how you quantized this.
Hello, I have recently followed the Phixtral recipe with another architecture and every step has been successful until I came up against conversion. I notice that the Phixtral and greater Phi-2 model does not have a 'tokenizer.model' file. I am attempting to convert my StableLM MOE using the 'convert-hf-to-gguf.py' script and I am faced with this error:
This file does not exist in the StableLM architecture, and as far as I can tell, it does not exist in the Phi architecture. Can you please share with me how you converted and quantized this model without that file? Thank you.
Here is how I did it https://colab.research.google.com/drive/130HcbgMiHQC7yadIDfZZrnnPz-K9Ynqc#scrollTo=QQAVuZ8kGTTq
But be aware that I don't think the model actually works. I got answers with random stuff.