rian-dolphin

def extract_dict_from_json_llm_response(text):
    text = "{"+"{".join([xi.strip() for xi in text.split("{")[1:]])
    text = "}".join([xi.strip() for xi in text.split("}")[:-1]])+"}"
    return eval(text)

def extract_python_dict_from_response(text):
    """
    LLM response formatted as python dicts.
    
    Args:

rian-dolphin

#-- Find index of max sharpe portfolio
max_sharpe_index = np.argmax(mean_variance_pairs[:,0]/mean_variance_pairs[:,1]**0.5)
#-- Add max sharpe portfolio as a seperate point to the plot
fig.add_trace(go.Scatter(x=mean_variance_pairs[max_sharpe_index,1]**0.5, y=mean_variance_pairs[max_sharpe_index,0],
                         marker=dict(color='red', size=14),
                         mode='markers'))

rian-dolphin

# -- Create the figure in Plotly
fig = px.scatter(
    filtered_df,
    x="gdpPercap",
    y="lifeExp",
    size="pop",
    color="continent",
    hover_name="country",
    log_x=log_x_choice,
    size_max=60,

rian-dolphin

# -- Read in the data
df = px.data.gapminder()
# -- Apply the year filter given by the user
filtered_df = df[(df.year == year_choice)]
# -- Apply the continent filter
if continent_choice != "All":
    filtered_df = filtered_df[filtered_df.continent == continent_choice]

rian-dolphin

# -- Get the user input
year_col, continent_col, log_x_col = st.columns([5, 5, 5])
with year_col:
    year_choice = st.slider(
        "What year would you like to examine?",
        min_value=1952,
        max_value=2007,
        step=5,
        value=2007,
    )

rian-dolphin

# -- Create three columns
col1, col2, col3 = st.columns([5, 5, 20])
# -- Put the image in the middle column
# - Commented out here so that the file will run without having the image downloaded
# with col2:
# st.image("streamlit.png", width=200)
# -- Put the title in the last column
with col3:
    st.title("Streamlit Demo")
# -- We use the first column here as a dummy to add a space to the left

rian-dolphin

import streamlit as st
import plotly.express as px

rian-dolphin

import streamlit as st
import plotly.express as px

st.set_page_config(layout="wide")

# -- Create three columns
col1, col2, col3 = st.columns([5, 5, 20])
# -- Put the image in the middle column
# - Commented out here so that the file will run without having the image downloaded
# with col2:

rian-dolphin

x = [dt.datetime.date(d) for d in df.index]
fig = plt.figure(figsize=(10,5))
plt.title('Walmart Quarterly Earnings')
plt.ylabel('Earnings (Billions)')
plt.grid(True)
plt.plot(x[:-len(predictions)],
         df.Earnings[:-len(predictions)],
         "b-")
plt.plot(x[-len(predictions):],
         df.Earnings[-len(predictions):],

rian-dolphin

model.eval()
with torch.no_grad():
    predictions, _ = model(train_scaled[-train_periods:], None)
#-- Apply inverse transform to undo scaling
predictions = scaler.inverse_transform(np.array(predictions.reshape(-1,1)))