GGUF general.architecture baichuan general.name Baichuan-7B baichuan.tensor_data_layout Meta AI original pth baichuan.context_length baichuan.embedding_length baichuan.block_count baichuan.feed_forward_length + baichuan.rope.dimension_count baichuan.attention.head_count baichuan.attention.head_count_kv ) baichuan.attention.layer_norm_rms_epsilon tokenizer.ggml.model llama tokenizer.ggml.tokens