Add fp16 ONNX (higher fidelity for transformers.js)

Panda-AI Max: distilgpt2 fine-tuned on panda facts (ONNX int8 for transformers.js)

Upload PandaAIMax_weights_for_hosting.zip

initial commit