Spaces:
Sleeping
Sleeping
File size: 2,430 Bytes
64c6e7a a85678b a2142e5 a85678b 64c6e7a a85678b 4c93e49 a85678b 4c93e49 0aa2d35 4c93e49 a85678b 4c93e49 a85678b bdedfa4 4c93e49 64c6e7a 4c93e49 |
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 streamlit as st
from transformers import GPT2Tokenizer, GPT2LMHeadModel
# Function to generate a response
def generate_response(input_text):
# Adjusted input to include the [Bot] marker
#adjusted_input = f"{input_text} [Bot]"
# Encode the adjusted input
inputs = tokenizer(input_text, return_tensors="pt")
# Generate a sequence of text with a slightly increased max_length to account for the prompt length
output_sequences = model.generate(
input_ids=inputs['input_ids'],
attention_mask=inputs['attention_mask'],
max_length=100, # Adjusted max_length
temperature=0.7,
top_k=50,
top_p=0.95,
no_repeat_ngram_size=2,
pad_token_id=tokenizer.eos_token_id,
#early_stopping=True,
do_sample=True
)
# Decode the generated sequence
full_generated_text = tokenizer.decode(output_sequences[0], skip_special_tokens=True)
# Extract the generated response after the [Bot] marker
bot_response_start = full_generated_text.find('[Bot]') + len('[Bot]')
bot_response = full_generated_text[bot_response_start:]
# Trim the response to end at the last period within the specified max_length
last_period_index = bot_response.rfind('.')
if last_period_index != -1:
bot_response = bot_response[:last_period_index + 1]
return bot_response.strip()
# Load pre-trained model tokenizer (vocabulary) and model
model_name = 'KhantKyaw/Chat_GPT-2'
tokenizer = GPT2Tokenizer.from_pretrained(model_name)
model = GPT2LMHeadModel.from_pretrained(model_name)
st.title("Chat with GPT-2")
# Initialize chat history
if "messages" not in st.session_state:
st.session_state.messages = []
# Display chat messages from history on app rerun
for message in st.session_state.messages:
with st.container():
st.markdown(f"**{message['role'].capitalize()}**: {message['content']}")
# React to user input
prompt = st.text_input("What is up?", key="chat_input")
if prompt:
with st.container():
st.markdown(f"**User**: {prompt}")
st.session_state.messages.append({"role": "user", "content": prompt})
# Decode the generated tokens and remove the eos token
response = generate_response(prompt)
with st.container():
st.markdown(f"**GPT-2**: {response}")
st.session_state.messages.append({"role": "assistant", "content": response})
|