Update app.py

app.py

@@ -1,58 +1,69 @@
-from flask import Flask, request, jsonify
-from transformers import AutoTokenizer, AutoModelForCausalLM
-import torch
-import os
-
-app = Flask(__name__)
+import gradio as gr
+from transformers import pipeline
+import json
 
+# ===========================
+# LOAD MODEL
+# ===========================
 model_id = "HuggingFaceH4/zephyr-7b-beta"
+ai = pipeline("text-generation", model=model_id, max_new_tokens=200)
+
 
-print("🚀 Loading model:", model_id)
+# ===========================
+# API ENDPOINT (VERY IMPORTANT)
+# call.php expects: { "reply": "..." }
+# ===========================
+def chat_api(request):
+    try:
+        message = request["message"]
+    except:
+        return {"reply": "INVALID_REQUEST"}
 
-tokenizer = AutoTokenizer.from_pretrained(model_id, use_fast=False)
-model = AutoModelForCausalLM.from_pretrained(model_id)
+    # Run model
+    output = ai(message)[0]["generated_text"]
 
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-model.to(device)
+    # Return exactly what call.php expects
+    return {"reply": output}
 
-print("✅ Model Loaded Successfully")
 
-@app.route('/chat', methods=['POST'])
-def chat():
-    try:
-        data = request.get_json()
-        message = data.get("message", "").strip()
+# ===========================
+# NORMAL UI CHAT (optional)
+# ===========================
+def chat_ui(msg):
+    return ai(msg)[0]["generated_text"]
 
-        if not message:
-            return jsonify({"error": "Empty message"}), 400
 
-        # Simple clean prompt
-        prompt = f"Human: {message}\nAssistant:"
+# ===========================
+# GRADIO SETUP
+# ===========================
+with gr.Blocks() as app:
 
-        inputs = tokenizer(prompt, return_tensors="pt").to(device)
+    gr.Markdown("# 🚀 Space 1 — Ghost Model")
 
-        output = model.generate(
-            inputs.input_ids,
-            max_length=300,
-            do_sample=True,
-            top_k=40,
-            top_p=0.92,
-            temperature=0.72,
-            pad_token_id=tokenizer.eos_token_id
-        )
+    # UI for testing only
+    inp = gr.Textbox(label="Your Message")
+    out = gr.Textbox(label="Model Output")
+    btn = gr.Button("Send")
+    btn.click(chat_ui, inp, out)
 
-        full = tokenizer.decode(output[0], skip_special_tokens=True)
-        reply = full.split("Assistant:")[-1].strip()
+    # Real API needed for call.php
+    gr.JSON(
+        value={"message": "send JSON POST here to use"},
+        label="API Endpoint Information"
+    )
 
-        return jsonify({"reply": reply})
+app.launch(server_name="0.0.0.0", server_port=7860)
 
-    except Exception as e:
-        return jsonify({"error": str(e)}), 500
+# IMPORTANT!
+# Add an API endpoint for HF:
+import gradio as gr
+from fastapi import FastAPI, Request
 
+fastapi_app = FastAPI()
 
-@app.route('/')
-def home():
-    return "LLM Space Active."
+@fastapi_app.post("/chat")
+async def chat_endpoint(request: Request):
+    data = await request.json()
+    return chat_api(data)
 
-if __name__ == "__main__":
-    app.run(host="0.0.0.0", port=7860)
+app = gr.mount_gradio_app(fastapi_app, app, path="/")