File size: 2,005 Bytes
21b8ce0 b5c263a d8a3c53 606c0ce d8a3c53 b0f00f9 ec06a49 98758c3 d8a3c53 3b38821 d8a3c53 4d0808e d8a3c53 606c0ce ec06a49 606c0ce 6c9b3f7 d8a3c53 606c0ce 827a15f 606c0ce d8a3c53 606c0ce d8a3c53 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 |
import spaces
import subprocess
import gradio as gr
from huggingface_hub import hf_hub_download
from llama_cpp import Llama
from llama_cpp_agent import LlamaCppAgent
from llama_cpp_agent import MessagesFormatterType
from llama_cpp_agent.providers import LlamaCppPythonProvider
subprocess.run('pip install llama-cpp-python==0.2.75 --no-build-isolation --no-cache-dir --upgrade --only-binary=:all: --extra-index-url=https://abetlen.github.io/llama-cpp-python/whl/cu124', env={'CMAKE_ARGS': "-DLLAMA_CUDA=on"}, shell=True)
hf_hub_download(repo_id="TheBloke/Mistral-7B-Instruct-v0.2-GGUF", filename="mistral-7b-instruct-v0.2.Q6_K.gguf", local_dir = "./models")
@spaces.GPU(duration=120)
def respond(
message,
history: list[tuple[str, str]],
system_message,
max_tokens,
temperature,
top_p,
):
llama_model = Llama(r"models/mistral-7b-instruct-v0.2.Q6_K.gguf", n_batch=1024, n_threads=0, n_gpu_layers=33, n_ctx=8192, verbose=False)
provider = LlamaCppPythonProvider(llama_model)
agent = LlamaCppAgent(
provider,
system_prompt=f"{system_message}",
predefined_messages_formatter_type=MessagesFormatterType.MISTRAL,
debug_output=True
)
settings = provider.get_provider_default_settings()
settings.stream = True
settings.max_tokens = max_tokens
settings.temperature = temperature
settings.top_p = top_p
yield agent.get_chat_response(message, llm_sampling_settings=settings)
demo = gr.ChatInterface(
respond,
additional_inputs=[
gr.Textbox(value="You are a helpful assistant.", label="System message"),
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
gr.Slider(
minimum=0.1,
maximum=1.0,
value=0.95,
step=0.05,
label="Top-p (nucleus sampling)",
),
],
)
if __name__ == "__main__":
demo.launch() |