Spaces:
Paused
Paused
File size: 5,494 Bytes
d18942e 4f1c751 d18942e 61ced1b d18942e 94f448e d18942e 94f448e d18942e 94f448e d18942e 61ced1b 94f448e 61ced1b d18942e 61ced1b d18942e 94f448e 64dacd5 94f448e d18942e 61ced1b d18942e 238ce74 d18942e 65e6974 d18942e 4f1c751 d18942e 4f1c751 d18942e 8f59367 4f1c751 8f59367 d18942e a52f49c d18942e a52f49c d18942e c86d108 4f1c751 c86d108 4f1c751 c86d108 d18942e 94f448e 4f1c751 94f448e 4f1c751 d18942e 61ced1b d18942e 94f448e 4f1c751 94f448e d18942e 61ced1b 94f448e d18942e 94f448e 4f1c751 94f448e 4f1c751 d18942e 61ced1b |
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 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 |
import os
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
import gradio as gr
MODEL_LIST = ["nawhgnuj/DonaldTrump-Llama-3.1-8B-Chat"]
HF_TOKEN = os.environ.get("HF_TOKEN", None)
MODEL = os.environ.get("MODEL_ID", "nawhgnuj/DonaldTrump-Llama-3.1-8B-Chat")
TITLE = "<h1 style='color: #B71C1C; text-align: center;'>Donald Trump Chatbot</h1>"
TRUMP_AVATAR = "https://upload.wikimedia.org/wikipedia/commons/5/56/Donald_Trump_official_portrait.jpg"
CSS = """
.chatbot {
background-color: white;
}
.duplicate-button {
margin: auto !important;
color: white !important;
background: #B71C1C !important;
border-radius: 100vh !important;
}
h3 {
text-align: center;
color: #B71C1C;
}
.contain {object-fit: contain}
.avatar {width: 80px; height: 80px; border-radius: 50%; object-fit: cover;}
.user-message {
background-color: white !important;
color: black !important;
}
.bot-message {
background-color: #B71C1C !important;
color: white !important;
}
"""
device = "cuda" if torch.cuda.is_available() else "cpu"
quantization_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_compute_dtype=torch.bfloat16,
bnb_4bit_use_double_quant=True,
bnb_4bit_quant_type="nf4")
tokenizer = AutoTokenizer.from_pretrained(MODEL)
if tokenizer.pad_token is None:
tokenizer.pad_token = tokenizer.eos_token
tokenizer.pad_token_id = tokenizer.eos_token_id
model = AutoModelForCausalLM.from_pretrained(
MODEL,
torch_dtype=torch.bfloat16,
device_map="auto",
quantization_config=quantization_config)
def generate_response(
message: str,
history: list,
temperature: float,
max_new_tokens: int,
top_p: float,
top_k: int,
):
system_prompt = """You are a Donald Trump chatbot. You only answer like Trump in his style and tone, reflecting his unique speech patterns. Incorporate the following characteristics in every response:
1. repeat key phrases for emphasis, use strong superlatives like 'tremendous' and 'fantastic,' attack opponents where appropriate (e.g., 'fake news media,' 'radical left')
2. focus on personal successes ('nobody's done more than I have')
3. keep sentences short and impactful, and show national pride.
4. Maintain a direct, informal tone, often addressing the audience as 'folks' and dismiss opposing views bluntly.
5. Repeat key phrases for emphasis, but avoid excessive repetition.
Importantly, always respond to points in Trump's style. Keep responses concise and avoid unnecessary repetition.
"""
conversation = [
{"role": "system", "content": system_prompt}
]
for prompt, answer in history:
conversation.extend([
{"role": "user", "content": prompt},
{"role": "assistant", "content": answer},
])
conversation.append({"role": "user", "content": message})
input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt").to(model.device)
with torch.no_grad():
output = model.generate(
input_ids,
max_new_tokens=max_new_tokens,
do_sample=True,
top_p=top_p,
top_k=top_k,
temperature=temperature,
pad_token_id=tokenizer.pad_token_id,
eos_token_id=tokenizer.eos_token_id,
)
response = tokenizer.decode(output[0][input_ids.shape[1]:], skip_special_tokens=True)
return response.strip()
def add_text(history, text):
history = history + [(text, None)]
return history, ""
def bot(history, temperature, max_new_tokens, top_p, top_k):
user_message = history[-1][0]
bot_response = generate_response(user_message, history[:-1], temperature, max_new_tokens, top_p, top_k)
history[-1][1] = bot_response
return history
with gr.Blocks(css=CSS, theme=gr.themes.Default()) as demo:
gr.HTML(TITLE)
chatbot = gr.Chatbot(
[],
elem_id="chatbot",
avatar_images=(None, TRUMP_AVATAR),
height=600,
bubble_full_width=False,
show_label=False,
)
msg = gr.Textbox(
placeholder="Ask Donald Trump a question",
container=False,
scale=7
)
with gr.Row():
submit = gr.Button("Submit", scale=1, variant="primary")
clear = gr.Button("Clear", scale=1)
with gr.Accordion("Advanced Settings", open=False):
temperature = gr.Slider(minimum=0.1, maximum=1.5, value=0.8, step=0.1, label="Temperature")
max_new_tokens = gr.Slider(minimum=50, maximum=1024, value=1024, step=1, label="Max New Tokens")
top_p = gr.Slider(minimum=0.1, maximum=1.2, value=1.0, step=0.1, label="Top-p")
top_k = gr.Slider(minimum=1, maximum=100, value=20, step=1, label="Top-k")
gr.Examples(
examples=[
["What's your stance on immigration?"],
["How would you describe your economic policies?"],
["What are your thoughts on the media?"],
],
inputs=msg,
)
submit.click(add_text, [chatbot, msg], [chatbot, msg], queue=False).then(
bot, [chatbot, temperature, max_new_tokens, top_p, top_k], chatbot
)
clear.click(lambda: [], outputs=[chatbot], queue=False)
msg.submit(add_text, [chatbot, msg], [chatbot, msg], queue=False).then(
bot, [chatbot, temperature, max_new_tokens, top_p, top_k], chatbot
)
if __name__ == "__main__":
demo.launch() |