* Start adding llama2.c. * Model loading. * Add the llama-v2 model. * Start converting the weights. * Rotary embedding tweaks. * Get the model to generate some tokens.