Intern-Decision-0.8B Core ML: L320_F8, L512_F8 (fp16 + int8), L1024_F16, embeddings, tokenizer

initial commit