File size: 2,266 Bytes
fc17b4b |
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 |
import streamlit as st
from transformers import pipeline
from huggingface_hub import InferenceClient
from PIL import Image
import os
def setup_session():
if 'app_ready' not in st.session_state:
print("Powering up the Dragon Radar...")
st.session_state['app_ready'] = True
st.session_state['hf_token'] = os.getenv("HUGGINGFACE_TOKEN")
st.session_state['client'] = InferenceClient(api_key=st.session_state['hf_token'])
def main():
setup_session()
st.header("Anime & Friends Image Commentary")
st.write("Let your favorite characters react to any image!")
character = st.selectbox(
"Select your commentator",
["goku", "elmo", "kirby", "pikachu"]
)
uploaded_img = st.file_uploader("Share your image!")
if uploaded_img is not None:
image = Image.open(uploaded_img)
st.image(image)
caption_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-large")
base_caption = caption_model(image)[0]['generated_text']
character_reactions = {
"goku": f"Describe this image like you're Goku from Dragon Ball Z, mentioning power levels: {base_caption}",
"elmo": f"Describe this image like you're Elmo from Sesame Street, speaking in third person: {base_caption}",
"kirby": f"Describe this image like you're Kirby, being cute and mentioning food: {base_caption}",
"pikachu": f"Describe this image like you're Pikachu, using 'pika' frequently: {base_caption}"
}
messages = [
{
"role": "user",
"content": character_reactions[character]
}
]
# Generate character response using Llama
response_stream = st.session_state['client'].chat.completions.create(
model="meta-llama/Llama-3.2-3B-Instruct",
messages=messages,
max_tokens=500,
stream=True
)
character_response = ''
for chunk in response_stream:
character_response += chunk.choices[0].delta.content
st.write(character_response)
if __name__ == '__main__':
main() |