File size: 4,256 Bytes
0c2bd43 |
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 |
import pandas as pd
import streamlit as st
from util.evaluator import evaluator, write_evaluation_commentary
import os
# Predefined examples
examples = {
'good': {
'question': "What causes rainbows to appear in the sky?",
'explanation': "Rainbows appear when sunlight is refracted, dispersed, and reflected inside water droplets in the atmosphere, resulting in a spectrum of light appearing in the sky."
},
'bad': {
'question': "What causes rainbows to appear in the sky?",
'explanation': "Rainbows happen because light in the sky gets mixed up and sometimes shows colors when it's raining or when there is water around."
}
}
# Function to check password
def check_password():
with st.sidebar:
password_input = st.text_input("Enter Password:", type="password")
submit_button = st.button("Submit")
if submit_button:
if password_input == os.getenv('PASSWORD'):
st.session_state['password_verified'] = True
st.experimental_rerun()
else:
st.error("Incorrect Password, please try again.")
# Function to evaluate batch data
def evaluate_batch(uploaded_file):
df = pd.read_csv(uploaded_file)
eval_instance = evaluator(model_name=st.session_state['model_name'])
results = []
for _, row in df.iterrows():
scores = eval_instance(row['question'], row['explanation'])
commentary = write_evaluation_commentary(scores)
result = {
'Question': row['question'],
'Explanation': row['explanation'],
**{c['Principle']: c['Score'] for c in commentary}
}
results.append(result)
return pd.DataFrame(results)
# Main app logic
def main():
st.title('Natural Language Explanation Demo')
model_name = st.selectbox('Select a model:', ['gpt4-1106', 'gpt35-1106'])
st.session_state['model_name'] = model_name # Save model name to session state for use in batch processing
input_type = st.radio("Choose input type:", ('Use predefined example', 'Enter your own', 'Upload CSV for batch evaluation'))
if input_type == 'Use predefined example':
example_type = st.radio("Select an example type:", ('good', 'bad'))
question = examples[example_type]['question']
explanation = examples[example_type]['explanation']
elif input_type == 'Enter your own':
question = st.text_input('Enter your question:', '')
explanation = st.text_input('Enter your explanation:', '')
else:
uploaded_file = st.file_uploader("Upload a CSV file", type='csv')
if uploaded_file and st.button('Evaluate Batch'):
result_df = evaluate_batch(uploaded_file)
st.write('### Evaluated Results')
st.dataframe(result_df)
csv = result_df.to_csv(index=False)
st.download_button(
label="Download evaluated results as CSV",
data=csv,
file_name='batch_evaluation_results.csv',
mime='text/csv'
)
return
if st.button('Evaluate Explanation'):
if question and explanation:
eval_instance = evaluator(model_name)
scores = eval_instance(question, explanation)
st.write('### Scores')
details = write_evaluation_commentary(scores)
df = pd.DataFrame(details)
st.write(df)
data = {
'Question': question,
'Explanation': explanation,
**{detail['Principle']: detail['Score'] for detail in details}
}
df = pd.DataFrame([data])
# Convert DataFrame to CSV for download
csv = df.to_csv(index=False)
st.download_button(
label="Download evaluation as CSV",
data=csv,
file_name='evaluation.csv',
mime='text/csv',
)
else:
st.error('Please enter both a question and an explanation to evaluate.')
if __name__ == '__main__':
if 'password_verified' not in st.session_state or not st.session_state['password_verified']:
check_password()
else:
main()
|