asdwd

app.py

@@ -1,51 +1,45 @@
-from fastapi import FastAPI, Request
+from fastapi import FastAPI
 from pydantic import BaseModel
 from llama_cpp import Llama
-from huggingface_hub import hf_hub_download
 import os
+import requests
 
 app = FastAPI()
 
-# === Configuration ===
-REPO_ID = "google/gemma-2b-it-GGUF"
-FILENAME = "gemma-2b-it.gguf"
-HF_TOKEN = os.environ.get("HF_TOKEN")
+# === Constants ===
+MODEL_REPO = "google/gemma-2b-it-GGUF"
+MODEL_FILE = "gemma-2b-it.gguf"
+MODEL_URL = f"https://huggingface.co/{MODEL_REPO}/resolve/main/{MODEL_FILE}"
 MODEL_DIR = "./models"
-CACHE_DIR = "./models/.hf_cache"
-MODEL_PATH = os.path.join(MODEL_DIR, FILENAME)
+MODEL_PATH = os.path.join(MODEL_DIR, MODEL_FILE)
 
-# === Ensure directories exist ===
+HF_TOKEN = os.getenv("HF_TOKEN")
+
+# === Create model directory ===
 os.makedirs(MODEL_DIR, exist_ok=True)
-os.makedirs(CACHE_DIR, exist_ok=True)
 
-# === Download model if not exists ===
+# === Manual download of GGUF ===
 if not os.path.exists(MODEL_PATH):
-    try:
-        print("📦 Downloading model from Hugging Face Hub...")
-        hf_hub_download(
-            repo_id=REPO_ID,
-            filename=FILENAME,
-            token=HF_TOKEN,
-            cache_dir=CACHE_DIR,
-            local_dir=MODEL_DIR,
-            local_dir_use_symlinks=False,
-        )
-        print(f"✅ Model downloaded to {MODEL_PATH}")
-    except Exception as e:
-        print(f"❌ Download failed: {e}")
-        raise
+    print("📦 Downloading GGUF model manually from Hugging Face...")
+    headers = {"Authorization": f"Bearer {HF_TOKEN}"}
+    response = requests.get(MODEL_URL, headers=headers, stream=True)
+    if response.status_code != 200:
+        raise RuntimeError(f"❌ Failed to download model. Status: {response.status_code}")
+    with open(MODEL_PATH, "wb") as f:
+        for chunk in response.iter_content(chunk_size=8192):
+            f.write(chunk)
+    print(f"✅ Model downloaded to {MODEL_PATH}")
 
 # === Load model ===
 print("🔧 Loading GGUF model...")
 llm = Llama(model_path=MODEL_PATH, n_ctx=512)
 
-# === Request schema ===
+# === Inference ===
 class PromptRequest(BaseModel):
     prompt: str
     max_tokens: int = 256
     temperature: float = 0.7
 
-# === Inference endpoint ===
 @app.post("/prompt")
 def generate_prompt(req: PromptRequest):
     output = llm(