# -*- coding: utf-8 -*-
# @Date : 2025/2/5 16:26
# @Author : q275343119
# @File : data_page.py
# @Description:
import io
from st_aggrid import AgGrid, JsCode, ColumnsAutoSizeMode
import streamlit as st
from app.backend.app_init_func import LEADERBOARD_MAP
from app.backend.constant import LEADERBOARD_ICON_MAP
COLUMNS = ['model_name',
'embd_dtype', 'embd_dim', 'num_params', 'max_tokens', 'similarity',
'query_instruct', 'corpus_instruct', 'reference'
]
HEADER_STYLE = {'fontSize': '18px'}
CELL_STYLE = {'fontSize': '18px'}
def is_section(group_name):
for k, v in LEADERBOARD_MAP.items():
leaderboard_name = v[0][0]
if group_name == leaderboard_name:
return True
return False
def get_closed_dataset():
data_engine = st.session_state["data_engine"]
closed_list = []
results = data_engine.results
for result in results:
if result.get("is_closed"):
closed_list.append(result.get("dataset_name"))
return closed_list
def convert_df_to_csv(df):
output = io.StringIO()
df.to_csv(output, index=False)
return output.getvalue()
def render_page(group_name):
# Add theme color and grid styles
st.title("Retrieval Embedding Benchmark (RTEB)")
st.markdown("""
""", unsafe_allow_html=True)
# logo
# st.markdown('
', unsafe_allow_html=True)
title = f'{LEADERBOARD_ICON_MAP.get(group_name.capitalize(), "")} {group_name.capitalize()}
'
if is_section(group_name):
title = f'{LEADERBOARD_ICON_MAP.get(group_name.capitalize() + " Leaderboard", "")} {group_name.capitalize() + " Leaderboard"}
'
# title
st.markdown(title, unsafe_allow_html=True)
data_engine = st.session_state["data_engine"]
df = data_engine.jsons_to_df().copy()
csv = convert_df_to_csv(df)
file_name = f"{group_name.capitalize()} Leaderboard" if is_section(group_name) else group_name.capitalize()
st.download_button(
label="Download CSV",
data=csv,
file_name=f"{file_name}.csv",
mime="text/csv",
icon=":material/download:",
)
# get columns
column_list = []
avg_column = None
if is_section(group_name):
avg_columns = []
for column in df.columns:
if column.startswith("Average"):
avg_columns.insert(0, column)
continue
if "Average" in column:
avg_columns.append(column)
continue
avg_column = avg_columns[0]
column_list.extend(avg_columns)
else:
for column in df.columns:
if column.startswith(group_name.capitalize() + " "):
avg_column = column
column_list.append(avg_column)
dataset_list = []
for dataset_dict in data_engine.datasets:
if dataset_dict["name"] == group_name:
dataset_list = dataset_dict["datasets"]
if not is_section(group_name):
column_list.extend(dataset_list)
closed_list = get_closed_dataset()
close_avg_list = list(set(dataset_list) & set(closed_list))
df["Closed average"] = df[close_avg_list].mean(axis=1).round(2)
column_list.append("Closed average")
open_avg_list = list(set(dataset_list) - set(closed_list))
df["Open average"] = df[open_avg_list].mean(axis=1).round(2)
column_list.append("Open average")
df = df[COLUMNS + column_list].sort_values(by=avg_column, ascending=False)
# rename avg column name
if not is_section(group_name):
new_column = avg_column.replace(group_name.capitalize(), "").strip()
df.rename(columns={avg_column: new_column}, inplace=True)
column_list.remove(avg_column)
avg_column = new_column
# setting column config
grid_options = {
'columnDefs': [
{
'headerName': 'Model Name',
'field': 'model_name',
'pinned': 'left',
'sortable': False,
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
"tooltipValueGetter": JsCode(
"""function(p) {return p.value}"""
),
"width": 250,
'cellRenderer': JsCode("""class CustomHTML {
init(params) {
const link = params.data.reference;
this.eGui = document.createElement('div');
this.eGui.innerHTML = link ?
`${params.value} ` :
params.value;
}
getGui() {
return this.eGui;
}
}"""),
'suppressSizeToFit': True
},
{'headerName': "Overall Score",
'field': avg_column,
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
# 'suppressSizeToFit': True
},
# Add Open average column definition
{'headerName': 'Open Average',
'field': 'Open average',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
# 'suppressSizeToFit': True
},
{'headerName': 'Closed Average',
'field': 'Closed average',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
# 'suppressSizeToFit': True
},
{
'headerName': 'Embd Dtype',
'field': 'embd_dtype',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
# 'suppressSizeToFit': True,
},
{
'headerName': 'Embd Dim',
'field': 'embd_dim',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
# 'suppressSizeToFit': True,
},
{
'headerName': 'Number of Parameters',
'field': 'num_params',
'cellDataType': 'number',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
'valueFormatter': JsCode(
"""function(params) {
const num = params.value;
if (num >= 1e9) return (num / 1e9).toFixed(2) + "B";
if (num >= 1e6) return (num / 1e6).toFixed(2) + "M";
if (num >= 1e3) return (num / 1e3).toFixed(2) + "K";
return num;
}"""
),
# 'suppressSizeToFit': True,
},
{
'headerName': 'Context Length',
'field': 'max_tokens',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
# 'suppressSizeToFit': True,
},
*[{'headerName': column if "Average" not in column else column.replace("Average", "").strip().capitalize(),
'field': column,
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
# 'suppressSizeToFit': True
} for column in column_list if
column not in (avg_column, "Closed average", "Open average")]
],
'defaultColDef': {
'filter': True,
'sortable': True,
'resizable': True,
'headerClass': "multi-line-header",
'autoHeaderHeight': True
},
"autoSizeStrategy": {
"type": 'fitGridWidth',
"defaultMinWidth": 105,
},
"tooltipShowDelay": 500,
}
custom_css = {
".a-cell": {
"display": "inline-block",
"white-space": "nowrap",
"overflow": "hidden",
"text-overflow": "ellipsis",
"width": "100%",
"min-width": "0"
},
".multi-line-header": {
"text-overflow": "clip",
"overflow": "visible",
"white-space": "normal",
"height": "auto",
}
}
AgGrid(
df,
enable_enterprise_modules=False,
gridOptions=grid_options,
allow_unsafe_jscode=True,
columns_auto_size_mode=ColumnsAutoSizeMode.FIT_CONTENTS,
theme="streamlit",
custom_css=custom_css
)