[pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

src/fastserve/models/vllm.py

@@ -2,7 +2,6 @@
 import os
 from typing import Any, List, Optional
 
-from llama_cpp import Llama
 from pydantic import BaseModel
 
 from fastserve.core import FastServe
@@ -12,7 +11,6 @@
 DEFAULT_MODEL = "openhermes-2-mistral-7b.Q6_K.gguf"
 
 
-
 class PromptRequest(BaseModel):
     prompt: str = "Llamas are cute animal"
     temperature: float = 0.8
@@ -37,11 +35,11 @@ def __init__(
         *args,
         **kwargs,
     ):
-        from vllm import LLM, SamplingParams
+        from vllm import LLM
 
         if not os.path.exists(model_path):
             raise FileNotFoundError(f"{model_path} not found.")
-        
+
         self.llm = LLM(model_path)
         self.model_path = model_path
         self.args = args
@@ -56,7 +54,9 @@ def __init__(
     def __call__(self, request: PromptRequest) -> Any:
         from vllm import SamplingParams
 
-        sampling_params = SamplingParams(temperature=request.temperature, top_p=request.top_p)
+        sampling_params = SamplingParams(
+            temperature=request.temperature, top_p=request.top_p
+        )
         result = self.llm(request.prompt, sampling_params=sampling_params)
         logger.info(result)
         return result

src/fastserve/utils.py

@@ -1,5 +1,6 @@
 import os
 from typing import Any
+
 from pydantic import BaseModel
 
 
@@ -24,23 +25,21 @@ def get_ui_folder():
     return path
 
 
-def download_file(url:str, dest:str):
+def download_file(url: str, dest: str):
     import requests
     from tqdm import tqdm
-    from huggingface_hub import HfApi, ModelFilter
-
 
     if dest is None:
         dest = os.path.abspath(os.path.basename(dest))
 
     response = requests.get(url, stream=True)
     response.raise_for_status()
-    total_size = int(response.headers.get('content-length', 0))
+    total_size = int(response.headers.get("content-length", 0))
     block_size = 1024
-    with open(dest, 'wb') as file, tqdm(
+    with open(dest, "wb") as file, tqdm(
         desc=dest,
         total=total_size,
-        unit='iB',
+        unit="iB",
         unit_scale=True,
         unit_divisor=1024,
     ) as bar: