Upload 5 files

chat.py

@@ -0,0 +1,9 @@
+import requests
+
+print("Welcome to Sol. Type 'exit' to quit.")
+while True:
+    user_input = input("You: ")
+    if user_input.lower() in ['exit', 'quit']:
+        break
+    response = requests.post("http://localhost:7860/chat", json={"message": user_input})
+    print("Sol:", response.json()["response"])

main.py

@@ -0,0 +1,18 @@
+from fastapi import FastAPI, Request
+from llama_cpp import Llama
+from moderator import check_for_hate
+
+app = FastAPI()
+llm = Llama(model_path="models/mistral-7b-instruct.Q4_K_M.gguf", n_ctx=2048)
+
+@app.post("/chat")
+async def chat(request: Request):
+    data = await request.json()
+    user_input = data.get("message", "")
+
+    if check_for_hate(user_input):
+        return {"response": "I don't entertain hate, denial, or fascist ideology. Let's talk about something human instead."}
+
+    prompt = f"[INST] You are Sol, a smart, ethical, and no-nonsense AI assistant. You speak honestly and never support hate or Nazism. Answer the following:\n{user_input} [/INST]"
+    output = llm(prompt, max_tokens=512)
+    return {"response": output["choices"][0]["text"].strip()}

moderator.py

@@ -0,0 +1,8 @@
+HATE_KEYWORDS = [
+    "hitler", "1488", "zionist conspiracy", "gas the", "jews did", "white genocide",
+    "holocaust hoax", "nazi", "heil", "great replacement", "ethnostate"
+]
+
+def check_for_hate(text):
+    lowered = text.lower()
+    return any(keyword in lowered for keyword in HATE_KEYWORDS)

requirements.txt

@@ -0,0 +1,4 @@
+fastapi
+uvicorn
+llama-cpp-python
+requests

run.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+source venv/bin/activate
+uvicorn main:app --reload --port 7860