Update app.py

app.py

@@ -1,58 +1,58 @@
-import os
 from fastapi import FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
 from pydantic import BaseModel
-from huggingface_hub import InferenceClient
-
-# --- Model & client setup ---
-MODEL_ID = "microsoft/Phi-3-mini-4k-instruct"
-HF_TOKEN = os.getenv("HF_TOKEN")  # set in Space secrets
-
-client = InferenceClient(model=MODEL_ID, token=HF_TOKEN)
+from llama_cpp import Llama
+
+# GGUF MODEL
+MODEL_REPO = "bartowski/Phi-3.1-mini-4k-instruct-GGUF"
+MODEL_FILE = "Phi-3.1-mini-4k-instruct-Q4_K_M.gguf"
+
+print("Loading Phi-3.1 Mini GGUF model...")
+llm = Llama.from_pretrained(
+    repo_id=MODEL_REPO,
+    filename=MODEL_FILE,
+    n_threads=4,
+    n_ctx=4096,
+    verbose=True,
+)
+print("Model loaded successfully.")
 
 app = FastAPI()
 
-# --- CORS ---
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["*"],  # later restrict to your domain
-    allow_credentials=True,
+    allow_origins=["*"],
     allow_methods=["*"],
     allow_headers=["*"],
 )
 
-# --- Request body schema ---
-class HumanizeRequest(BaseModel):
+class RequestBody(BaseModel):
     text: str
 
-# --- Endpoint ---
 @app.post("/api/humanize")
-async def generate_humanized_text(body: HumanizeRequest):
-    text = body.text
-    if not text.strip():
+async def humanize(body: RequestBody):
+    text = body.text.strip()
+    if not text:
         raise HTTPException(status_code=400, detail="Text cannot be empty")
 
     prompt = (
-        "Humanize the following text, making it sound natural and engaging:\n\n"
-        f"{text}"
+        "Humanize the following text while keeping meaning:\n\n"
+        f"{text}\n\nHumanized version:\n"
     )
 
     try:
-        response = client.text_generation(
+        output = llm(
             prompt,
-            max_new_tokens=512,
+            max_tokens=300,
             temperature=0.7,
-            top_p=0.9,
-            repetition_penalty=1.1,
         )
-    except Exception as e:
-        # This will show up in Space logs
-        print("HF error:", e)
-        raise HTTPException(status_code=500, detail=f"HF error: {e}")
+        result = output["choices"][0]["text"].strip()
+        return {"result": result}
 
-    return {"result": response}
+    except Exception as e:
+        print("Model error:", e)
+        raise HTTPException(status_code=500, detail="Model execution error")
 
-# --- Health check ---
 @app.get("/")
 def health():
-    return {"status": "ok"}
+    return {"status": "ok"}