ababio's picture
Create app.y
357410f verified
raw
history blame
6.55 kB
from typing import Dict, List, Union
from google.cloud import aiplatform
from google.protobuf import json_format
from google.protobuf.struct_pb2 import Value
import os
import re
import pandas as pd
import plotly.express as px
import plotly.graph_objects as go
import streamlit as st
import nltk
import json
import tempfile
# process of getting credentials
def get_credentials():
creds_json_str = os.getenv("JSONSTR") # get json credentials stored as a string
if creds_json_str is None:
raise ValueError("GOOGLE_APPLICATION_CREDENTIALS_JSON not found in environment")
# create a temporary file
with tempfile.NamedTemporaryFile(mode="w+", delete=False, suffix=".json") as temp:
temp.write(creds_json_str) # write in json format
temp_filename = temp.name
return temp_filename
# pass
os.environ["GOOGLE_APPLICATION_CREDENTIALS"]= get_credentials()
max_seq_length = 2048
dtype = None
load_in_4bit = True
# Check if 'punkt' is already downloaded, otherwise download it
try:
nltk.data.find('tokenizers/punkt')
except LookupError:
nltk.download('punkt')
text_split_tokenizer = nltk.data.load('tokenizers/punkt/english.pickle')
def predict_custom_trained_model_sample(
project: str,
endpoint_id: str,
instances: Union[Dict, List[Dict]],
location: str = "us-east4",
api_endpoint: str = "us-east4-aiplatform.googleapis.com",
) -> List[str]:
"""
`instances` can be either single instance of type dict or a list
of instances.
"""
client_options = {"api_endpoint": api_endpoint}
client = aiplatform.gapic.PredictionServiceClient(client_options=client_options)
instances = instances if isinstance(instances, list) else [instances]
instances = [
json_format.ParseDict(instance_dict, Value()) for instance_dict in instances
]
parameters_dict = {}
parameters = json_format.ParseDict(parameters_dict, Value())
endpoint = client.endpoint_path(
project=project, location=location, endpoint=endpoint_id
)
response = client.predict(
endpoint=endpoint, instances=instances, parameters=parameters
)
predictions_list = []
predictions = response.predictions
for prediction in predictions:
if isinstance(prediction, str):
clean_prediction = re.sub(r'(\n|Origin|###|Optimization|Response:)', '', prediction)
split_predictions = clean_prediction.split()
predictions_list.extend(split_predictions)
else:
print(" prediction (unknown type, skipping):", prediction)
return [emotion for emotion in predictions_list if emotion in d_emotion.values()]
d_emotion = {0: 'admiration', 1: 'amusement', 2: 'anger', 3: 'annoyance', 4: 'approval', 5: 'caring', 6: 'confusion',
7: 'curiosity', 8: 'desire', 9: 'disappointment', 10: 'disapproval', 11: 'disgust', 12: 'embarrassment',
13: 'excitement', 14: 'fear', 15: 'gratitude', 16: 'grief', 17: 'joy', 18: 'love', 19: 'nervousness',
20: 'optimism', 21: 'pride', 22: 'realization', 23: 'relief', 24: 'remorse', 25: 'sadness', 26: 'surprise',
27: 'neutral'}
st.write("Write or paste any number of document texts to analyse the emotion percentage with your document")
user_input = st.text_area('Enter Text to Analyze')
button = st.button("Analyze")
if user_input and button:
alpaca_prompt = """Below is a conversation between a human and an AI agent. write a response based on the input.
### Instruction:
predict the emotion word or words
### Input:
{}
### Response:
"""
instances = []
input_array = text_split_tokenizer.tokenize(user_input)
for sentence in input_array:
formatted_input = alpaca_prompt.format(sentence.strip())
instance = {
"inputs": formatted_input,
"parameters": {
"max_new_tokens": 4,
"temperature": 0.00001,
"top_p": 0.9,
"top_k": 10
}
}
instances.append(instance)
predictions = predict_custom_trained_model_sample(
project=os.environ["project"],
endpoint_id=os.environ["endpoint_id"],
location=os.environ["location"],
instances=instances
)
emotion_counts = pd.Series(predictions).value_counts(normalize=True).reset_index()
emotion_counts.columns = ['Emotion', 'Percentage']
emotion_counts['Percentage'] *= 100 # Convert to percentage
fig_pie = px.pie(emotion_counts, values='Percentage', names='Emotion', title='Percentage of Emotions in Given Text')
fig_pie.update_traces(textposition='inside', textinfo='percent+label')
@st.cache_data
def get_emotion_chart(predictions):
emotion_counts = pd.Series(predictions).value_counts().reset_index()
emotion_counts.columns = ['Emotion', 'Count']
fig_bar = go.Figure()
fig_bar.add_trace(go.Bar(
x=emotion_counts['Emotion'],
y=emotion_counts['Count'],
marker_color='indianred'
))
fig_bar.update_layout(title='Count of Each Emotion in Given Text', xaxis_title='Emotion', yaxis_title='Count')
return fig_bar
fig_bar = get_emotion_chart(predictions)
@st.cache_data
def get_emotion_heatmap(predictions):
# Create a matrix for heatmap
# Count occurrences of each emotion
emotion_counts = pd.Series(predictions).value_counts().reset_index()
emotion_counts.columns = ['Emotion', 'Count']
heatmap_matrix = pd.DataFrame(0, index=d_emotion.values(), columns=d_emotion.values())
for index, row in emotion_counts.iterrows():
heatmap_matrix.at[row['Emotion'], row['Emotion']] = row['Count']
fig = go.Figure(data=go.Heatmap(
z=heatmap_matrix.values,
x=heatmap_matrix.columns.tolist(),
y=heatmap_matrix.index.tolist(),
text=heatmap_matrix.values,
hovertemplate="Count: %{text}",
colorscale='Viridis'
))
fig.update_layout(title='Emotion Heatmap', xaxis_title='Predicted Emotion', yaxis_title='Predicted Emotion')
return fig
fig_dist = get_emotion_heatmap(predictions)
tab1, tab2, tab3 = st.tabs(["Emotion Analysis", "Emotion Counts Distribution", "Heatmap"])
with tab1:
st.plotly_chart(fig_pie)
with tab2:
st.plotly_chart(fig_bar)
with tab3:
st.plotly_chart(fig_dist)