Usage:
from llama_cpp import Llama
GGUF_PATH = "./phi_4_mini_instruct_Q4_K_M.gguf"
print("⏳ Initializing long-context GGUF engine directly into memory layout...")
llm = Llama(
model_path=GGUF_PATH,
n_ctx=16384,
flash_attn=True,
n_threads=4,
verbose=False
)
