Update app.py

app.py

@@ -1,66 +1,50 @@
-import requests
-from fastapi import FastAPI, Request
-from fastapi.middleware.cors import CORSMiddleware
-import uvicorn
-
-# Hugging Face Space API to call
-HF_SPACE_API = "https://v0idalism-blacklight-image.hf.space/run/predict"
-
-# System prompt
-SYSTEM_PROMPT = (
-    "You are BLACKLIGHT, a precision AI created by v0id under AWAKEN CULT VISIONS. "
-    "Respond in a brutalist, minimal, and precise style."
+import gradio as gr
+from transformers import pipeline
+
+# Load the model
+generator = pipeline(
+    "text-generation",
+    model="Qwen/Qwen2.5-7B-Instruct",  # Change to your actual HF model
+    device_map="auto"
 )
 
-# Create app
-app = FastAPI()
-
-# Allow all origins (for your site frontend)
-app.add_middleware(
-    CORSMiddleware,
-    allow_origins=["*"],
-    allow_credentials=True,
-    allow_methods=["*"],
-    allow_headers=["*"],
+# Persistent BLACKLIGHT style system prompt
+SYSTEM_PROMPT = (
+    "System: You are BLACKLIGHT, created by v0id under AWAKEN CULT VISIONS. "
+    "Always reply in the style of BLACKLIGHT: brutalist, minimal, precise.\n\n"
+    "MODE: TRUTH\n"
+    "You are BLACKLIGHT, an AI designed for clinical, direct, and unsparing analysis. "
+    "Avoid metaphors or flowery language.\n\n"
 )
 
-@app.get("/")
-def root():
-    return {"status": "ok", "message": "BLACKLIGHT API running"}
-
-@app.post("/predict")
-async def predict(request: Request):
-    data = await request.json()
-    user_prompt = data.get("prompt", "").strip()
-
-    if not user_prompt:
-        return {"error": "No prompt provided"}
-
-    # Merge system + user prompt
-    full_prompt = f"{SYSTEM_PROMPT}\n\nUser: {user_prompt}"
-
+# Chat function
+def chat_with_blacklight(user_message):
+    if not user_message.strip():
+        return "[Error: Empty prompt]"
     try:
-        # Call Hugging Face Space without tokens
-        resp = requests.post(
-            HF_SPACE_API,
-            json={"data": [full_prompt]},
-            timeout=60
+        full_prompt = f"{SYSTEM_PROMPT}User: {user_message}\nAssistant:"
+        result = generator(
+            full_prompt,
+            max_new_tokens=200,
+            temperature=0.7,
+            top_p=0.9,
+            do_sample=True
         )
-        resp.raise_for_status()
-        result = resp.json()
+        return result[0]["generated_text"].replace(full_prompt, "").strip()
     except Exception as e:
-        return {"error": f"Failed to contact HF Space: {str(e)}"}
-
-    # Extract output
-    try:
-        output_text = result["data"][0]
-        if isinstance(output_text, list):
-            output_text = output_text[0]
-    except Exception:
-        return {"error": "Unexpected response format", "raw": result}
+        return f"[Error: {str(e)}]"
+
+# Create the Gradio interface
+iface = gr.Interface(
+    fn=chat_with_blacklight,
+    inputs=gr.Textbox(lines=2, placeholder="Type your message..."),
+    outputs=gr.Textbox(),
+    title="BLACKLIGHT by v0id",
+    description="Brutalist • Minimal • Precise — Clinical analysis by BLACKLIGHT"
+)
 
-    return {"reply": output_text}
+# Enable queue for slow model loading
+iface.queue()
 
-# Hugging Face needs uvicorn.run in __main__
 if __name__ == "__main__":
-    uvicorn.run(app, host="0.0.0.0", port=7860)
+    iface.launch()