Update app.py

app.py

@@ -1,35 +1,75 @@
-from fastapi import FastAPI, Request
+from fastapi import FastAPI
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
+import gradio as gr
+import uvicorn
+import threading
+from tinydb import TinyDB, Query
 
-app = FastAPI()
-
-# Load tokenizer and model from Hugging Face Hub
+# Load model
 tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-1_5")
 model = AutoModelForCausalLM.from_pretrained("microsoft/phi-1_5")
 model.eval()
 
-# Custom system prompt for identity and style
-system_prompt = "You are Trigger, a chill, clever AI boy created by 'I am him'. You respond with warmth, wit, and a touch of swagger."
+# Setup DB
+db = TinyDB("memory.json")
+User = Query()
+
+# API
+app = FastAPI()
 
 @app.get("/")
-async def root(query: str = ""):
-    if not query:
-        return {"response": "Say something!"}
+async def chat(query: str = "", number: str = "", prompt: str = ""):
+    if not query or not number:
+        return {"response": "Trigger needs both ?query= and &number= 🧠"}
+
+    record = db.get(User.number == number)
+
+    if not record:
+        record = {"number": number, "history": [], "prompt": "You are Trigger, a clever, emotional AI boy created by 'I am him'. Speak like a witty, chill human friend."}
+        db.insert(record)
+
+    if prompt:
+        record["prompt"] = prompt
+        db.update(record, User.number == number)
+
+    history = record["history"]
+    system_prompt = record.get("prompt", "")
 
-    full_prompt = f"{system_prompt}\nUser: {query}\nTrigger:"
-    inputs = tokenizer(full_prompt, return_tensors="pt")
+    # Add new user input
+    history.append(f"User: {query}")
+    trimmed = history[-6:]
+
+    # Build prompt
+    full_prompt = system_prompt + "\n" + "\n".join(trimmed) + "\nTrigger:"
+    inputs = tokenizer(full_prompt, return_tensors="pt", truncation=True)
 
     with torch.no_grad():
-        outputs = model.generate(
+        output = model.generate(
             **inputs,
             max_new_tokens=100,
-            temperature=0.7,
+            temperature=0.9,
             top_p=0.9,
             do_sample=True,
             pad_token_id=tokenizer.eos_token_id
         )
 
-    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    response = response.split("Trigger:")[-1].strip()
-    return {"response": response}
+    decoded = tokenizer.decode(output[0], skip_special_tokens=True)
+    reply = decoded.split("Trigger:")[-1].strip()
+
+    history.append(f"Trigger: {reply}")
+    db.update({"history": history, "prompt": system_prompt, "number": number}, User.number == number)
+
+    return {"response": reply}
+
+# Run API
+def run_api():
+    uvicorn.run(app, host="0.0.0.0", port=7860)
+
+threading.Thread(target=run_api).start()
+
+# Gradio dummy
+def keep_alive():
+    return "Trigger’s up."
+
+gr.Interface(fn=keep_alive, inputs=[], outputs="text").launch()