|
import plotly.graph_objects as go |
|
import plotly.express as px |
|
import numpy as np |
|
|
|
|
|
def plot_elo_mle(df): |
|
fig = px.scatter(df, x="model", y="rating", error_y="error_y", |
|
error_y_minus="error_y_minus", |
|
|
|
) |
|
fig.update_layout(xaxis_title="Model", |
|
yaxis_title="Rating", |
|
autosize=True, |
|
|
|
|
|
) |
|
return fig |
|
|
|
|
|
def plot_solve_rate(df, task, rows=30, cols=38): |
|
keys = df["task_id"] |
|
values = df["solve_rate"] |
|
|
|
values = np.array(values) |
|
|
|
n = len(values) |
|
values = np.pad(values, (0, rows * cols - n), 'constant', constant_values=np.nan).reshape((rows, cols)) |
|
keys = np.pad(keys, (0, rows * cols - n), 'constant', constant_values='').reshape((rows, cols)) |
|
|
|
hover_text = np.empty_like(values, dtype=object) |
|
for i in range(rows): |
|
for j in range(cols): |
|
if not np.isnan(values[i, j]): |
|
hover_text[i, j] = f"{keys[i, j]}<br>Solve Rate: {values[i, j]:.2f}" |
|
else: |
|
hover_text[i, j] = "NaN" |
|
|
|
upper_solve_rate = round(np.count_nonzero(values)/n*100, 2) |
|
fig = go.Figure(data=go.Heatmap( |
|
z=values, |
|
text=hover_text, |
|
hoverinfo='text', |
|
colorscale='teal', |
|
zmin=0, |
|
zmax=100 |
|
)) |
|
|
|
fig.update_layout( |
|
title=f'BigCodeBench-{task}<br><i>Lowest Upper Limit: {upper_solve_rate}%</i>', |
|
xaxis_nticks=cols, |
|
yaxis_nticks=rows, |
|
xaxis=dict(showticklabels=False), |
|
yaxis=dict(showticklabels=False), |
|
autosize=True, |
|
|
|
|
|
) |
|
|
|
return fig |