I am using this repo to create an onnx model for tiny llama which don’t have any dynamic shape. GitHub - KaihuaTang/Minimalist-TinyLLaMA-to-Onnx: Export TinyLLaMA to Onnx and Conduct LLM inference using onnxruntime
The command I used to convert the onnx model to relay module mod, params = relay.frontend.from_onnx(model=model, shape=input_dict,dtype=“float16”) input_dict have correct shape as I have tested the onnx model by executing it.
The size of model is around 2GB This conversion keeps on failing any suggestion would be greatly appreciated! Thank You