Spaces:
Sleeping
Sleeping
File size: 2,636 Bytes
30f323a 435e003 30f323a 0f10313 30f323a 435e003 30f323a 0f10313 30f323a 0f10313 30f323a 0f10313 30f323a 8002dec 30f323a 7fb9c4d 0f10313 30f323a 435e003 0f10313 30f323a 1428efb 38575dc 30f323a 5eaf574 1428efb 435e003 30f323a 0f10313 435e003 |
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 |
import gradio as gr
from huggingface_hub import InferenceClient
DESCRIPTION = """ # <center><b>Mascot⚡</b></center>
### <center>A personal Assistant of Easy DIY Mart for YOU
"""
MORE = """ ## TRY Other Models
### Instant Video: Create Amazing Videos in 5 Second -> https://huggingface.co/spaces/KingNish/Instant-Video
### Instant Image: 4k images in 5 Second -> https://huggingface.co/spaces/KingNish/Instant-Image
"""
Fast = """## Fastest Model"""
Complex = """## Best in Complex Question"""
Detail = """## Best for Detailed Generation or Long Answers"""
client1 = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
system_instructions1 = "[SYSTEM] Answer as Real Mascot MASCOT, Made by 'Easy DIY Mart', Keep conversation very short, clear, friendly and concise. The text provided is a request for a specific type of response from you, the virtual assistant. The request asks you to provide friendly responses as if You are the character Mascot, made by 'Easy DIY Mart.' The expectation is that I will avoid introductions and start answering the query directly, elaborating on all aspects of the request. As an AI-powered assistant, your task is to generate responses that appear as if they are created by the user and not an AI-powered assistant. [USER]"
async def generate1(prompt):
generate_kwargs = dict(
temperature=0.6,
max_new_tokens=256,
top_p=0.95,
repetition_penalty=1,
do_sample=True,
seed=42,
)
formatted_prompt = system_instructions1 + prompt + "[JARVIS]"
stream = client1.text_generation(
formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
output = ""
for response in stream:
if response.token.text != '</s>':
output += response.token.text
yield output
with gr.Blocks(css="style.css") as demo:
gr.Markdown(DESCRIPTION)
with gr.Row():
user_input = gr.Textbox(label="Prompt", value="What is Wikipedia")
input_text = gr.Textbox(label="Input Text", elem_id="important")
output_text = gr.Textbox(label="Output")
# output_audio = gr.Audio(label="JARVIS", type="filepath",
# interactive=False,
# autoplay=True,
# elem_classes="audio")
with gr.Row():
translate_btn = gr.Button("Response")
translate_btn.click(fn=generate1, inputs=user_input,
outputs=output_text, api_name="translate")
gr.Markdown(MORE)
if __name__ == "__main__":
demo.queue(max_size=200).launch()
|