igpu

from llama_cpp import Llama

llm = Llama(
model_path="C:\\Users\\ArabTech\\Desktop\\4\\phi-3.5-mini-instruct-q4_k_m.gguf",
n_gpu_layers=-1,
verbose=True,
)
output = llm(
"Q: Who is Napoleon Bonaparte A: ",
max_tokens=1024,
stop=["\n"] # Add a stop sequence to end generation at a newline
)
print(output)

n_gpu_layers=-1
n_gpu_layers=32

not work on igpu intel

how ofload model on igpu intel?