# -*- coding: utf-8 -*-
# @Date : 2025/2/5 16:26
# @Author : q275343119
# @File : data_page.py
# @Description:
from st_aggrid import AgGrid, JsCode, ColumnsAutoSizeMode
import streamlit as st
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 render_page(group_name):
# Add theme color and grid styles
st.markdown("""
""", unsafe_allow_html=True)
# logo
# st.markdown('
', unsafe_allow_html=True)
# title
st.markdown('Embedding Benchmark For Retrieval
', unsafe_allow_html=True)
data_engine = st.session_state["data_engine"]
df = data_engine.jsons_to_df()[:]
# get columns
column_list = []
avg_column = None
if group_name == "text":
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
new_column = avg_column.replace(group_name.capitalize(), "").strip()
df.rename(columns={avg_column: new_column}, inplace=True)
column_list.append(new_column)
avg_column = new_column
dataset_list = []
if group_name != "text":
for dataset_dict in data_engine.datasets:
if dataset_dict["name"] == group_name:
dataset_list = dataset_dict["datasets"]
column_list.extend(dataset_list)
df = df[COLUMNS + column_list].sort_values(by=avg_column, ascending=False)
# setting column config
grid_options = {
'columnDefs': [
{
'headerName': 'Model Name',
'field': 'model_name',
'pinned': 'left',
'sortable': False,
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
'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;
}
}"""),
},
{'headerName': avg_column,
'field': avg_column,
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
'suppressSizeToFit': True},
{
'headerName': 'Data Type',
'field': 'embd_dtype',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
},
{
'headerName': 'Embd Dim',
'field': 'embd_dim',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
},
{
'headerName': 'Model Size (# of Parameters)',
'field': 'num_params',
'cellDataType': 'number',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
},
{
'headerName': 'Context Length',
'field': 'max_tokens',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
},
{
'headerName': 'Query Instruction',
'field': 'query_instruct',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
'suppressSizeToFit': True,
},
{
'headerName': 'Corpus Instruction',
'field': 'corpus_instruct',
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
'suppressSizeToFit': True,
},
*[{'headerName': column,
'field': column,
'headerStyle': HEADER_STYLE,
'cellStyle': CELL_STYLE,
'suppressSizeToFit': True} for column in column_list if column != avg_column]
],
'defaultColDef': {
'filter': True,
'sortable': True,
'resizable': True
},
'autoSizeStrategy': {
'type': 'fitCellContents'
}
}
AgGrid(
df,
enable_enterprise_modules=False,
gridOptions=grid_options,
allow_unsafe_jscode=True,
columns_auto_size_mode=ColumnsAutoSizeMode.FIT_CONTENTS,
theme="streamlit",
)