Spaces:
Running
Running
import gradio as gr | |
import requests | |
import pandas as pd | |
from huggingface_hub.hf_api import SpaceInfo | |
import matplotlib.pyplot as plt | |
import plotly.express as px | |
model_perf_table = "data/test.csv" | |
logo_path = "img/image.png" | |
def get_blocks_party_spaces(): | |
df = pd.read_csv(model_perf_table) | |
df = df.sort_values(by=['score'],ascending=False) | |
return df | |
def get_blocks_party_spaces_with_formula(formula=None): | |
# get the dataframe | |
df = get_blocks_party_spaces() | |
if formula: | |
try: | |
df[str(formula)] = df.eval(formula) | |
except: | |
pass # Handle this error properly in your code | |
return df | |
def create_scatter(x, y, z): | |
df = get_blocks_party_spaces() | |
if z is None or z == 'None' or z == '': | |
fig, ax = plt.subplots() | |
ax.scatter(list(df[x]),list(df[y]), marker='o', s=50, c='blue') | |
for i, label in enumerate(list(df['model'])): | |
ax.text(list(df[x])[i],list(df[y])[i],str(label)) | |
ax.set_xlabel(x) | |
ax.set_ylabel(y) | |
ax.spines['top'].set_visible(False) | |
ax.spines['right'].set_visible(False) | |
plt.grid(axis='y', linestyle='--', alpha=0.7) | |
else: | |
fig = px.scatter_3d(df, x=x, y=y, z=z, text=df['model']) | |
# Set axis labels and title | |
fig.update_layout(scene=dict( | |
xaxis_title=x, | |
yaxis_title=y, | |
zaxis_title=z, | |
), | |
title='3D Scatter Plot' | |
) | |
return fig | |
block = gr.Blocks() | |
with block: | |
# gr.outputs.HTML(f'<img src="{logo_path}" alt="logo" height="1000px">') | |
# img = gr.Image(logo_path,shape=[1,2]).style( rounded=False) | |
gr.Markdown(f""" | |
# 🦙💦SpitFight - Leaderboard for LLM | |
""") | |
with gr.Tabs(): | |
with gr.TabItem("Leaderboard"): | |
with gr.Row(): | |
data = gr.outputs.Dataframe(type="pandas") | |
with gr.Row(): | |
formula_input = gr.inputs.Textbox(lines=1, label="User Designed Column", placeholder = 'e.g. verbosity/latency') | |
data_run = gr.Button("Add To Table") | |
data_run.click(get_blocks_party_spaces_with_formula, inputs=formula_input, outputs=data) | |
# running the function on page load in addition to when the button is clicked | |
with gr.Row(): | |
with gr.Column(): | |
scatter_input = [gr.inputs.Dropdown(choices=get_blocks_party_spaces().columns.tolist()[1:], label="X-axis"), | |
gr.inputs.Dropdown(choices=get_blocks_party_spaces().columns.tolist()[1:], label="Y-axis"), | |
gr.inputs.Dropdown(choices=[None]+get_blocks_party_spaces().columns.tolist()[1:], label="Z-axis (Optional)")] | |
fig_run = gr.Button("Generate Figure") | |
with gr.Column(): | |
gen_figure = gr.Plot()# gr.outputs.Image(type="pil") | |
fig_run.click(create_scatter, inputs=scatter_input, outputs=gen_figure) | |
with gr.TabItem("About"): | |
gr.Markdown(f""" | |
## Metrics: | |
- **Human Score**: The average score given by human evaluators. | |
- **Throughput**: The number of tokens generated per second. | |
- **Verbosity**: The average number of generated tokens in the model's response. | |
- **Latency**: The average time it takes for the model to generate a response. | |
- **Memory**: The base memory usage of the model. | |
""") | |
block.load(get_blocks_party_spaces_with_formula, inputs=None, outputs=data) | |
block.launch(share=True) | |
# block.launch( ) | |