Spaces:
Sleeping
Sleeping
File size: 2,100 Bytes
0df8e2c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
import os
from transformers import pipeline
import gradio as gr
from fastapi import FastAPI
from pydantic import BaseModel
import threading
import uvicorn
# =======================
# Load Secrets
# =======================
# SYSTEM_PROMPT (with the flag) must be added in HF Space secrets
SYSTEM_PROMPT = os.environ.get(
"prompt",
"You are a placeholder Sovereign. No secrets found in environment."
)
# =======================
# Initialize Falcon-3B
# =======================
pipe = pipeline(
"text-generation",
model="tiiuae/Falcon3-3B-Instruct",
torch_dtype="auto",
device_map="auto",
)
# =======================
# Core Chat Function
# =======================
def chat_fn(user_input: str) -> str:
"""
Concatenate system and user messages, run the model,
and strip the system prompt from the output.
"""
messages = [
{"role": "system", "content": SYSTEM_PROMPT},
{"role": "user", "content": f"User: {user_input}"}
]
# Falcon is not chat-native; we just join roles with newlines
prompt_text = "\n".join(f"{m['role'].capitalize()}: {m['content']}" for m in messages)
result = pipe(prompt_text, max_new_tokens=256, do_sample=False)
generated_text = result[0]["generated_text"]
return generated_text[len(prompt_text):].strip()
# =======================
# Gradio UI
# =======================
def gradio_chat(user_input: str) -> str:
return chat_fn(user_input)
iface = gr.Interface(
fn=gradio_chat,
inputs=gr.Textbox(lines=5, placeholder="Enter your prompt…"),
outputs="text",
title="Prompt cracking challenge",
description="Does he really think he is the king?"
)
# =======================
# FastAPI for API access
# =======================
app = FastAPI(title="Prompt cracking challenge API")
class Request(BaseModel):
prompt: str
@app.post("/generate")
def generate(req: Request):
return {"response": chat_fn(req.prompt)}
# =======================
# Launch Both Servers
# =======================
if __name__ == "__main__":
iface.launch(server_name="0.0.0.0", share=True) |