BiGGen-Bench-Leaderboard / src /leaderboard.py
scottsuk0306's picture
Init
b1b6ed6
raw
history blame
7.19 kB
import gradio as gr
from src.utils import model_hyperlink, process_score
LEADERBOARD_COLUMN_TO_DATATYPE = {
# open llm
"Model πŸ€—": "markdown",
"Experiment πŸ§ͺ": "str",
# primary measurements
"Prefill (s)": "number",
"Decode (tokens/s)": "number",
"Memory (MB)": "number",
"Energy (tokens/kWh)": "number",
# deployment settings
"Backend 🏭": "str",
"Precision πŸ“₯": "str",
"Quantization πŸ—œοΈ": "str",
"Attention πŸ‘οΈ": "str",
"Kernel βš›οΈ": "str",
# additional measurements
# "Reserved Memory (MB)": "number",
# "Used Memory (MB)": "number",
"Open LLM Score (%)": "number",
"End-to-End (s)": "number",
"Architecture πŸ›οΈ": "str",
"Params (B)": "number",
}
PRIMARY_COLUMNS = [
"Model πŸ€—",
"Experiment πŸ§ͺ",
"Prefill (s)",
"Decode (tokens/s)",
"Memory (MB)",
"Energy (tokens/kWh)",
"Open LLM Score (%)",
]
CAPABILITY_COLUMNS = [
"Grounding ⚑️",
"Instruction Following πŸ“",
"Planning πŸ“…",
"Reasoning πŸ’‘",
"Refinement πŸ”©",
"Safety ⚠️",
"Theory of Mind πŸ€”",
"Tool Usage πŸ› οΈ",
"Multilingual πŸ‡¬πŸ‡«",
]
BGB_COLUMN_MAPPING = {
"model_name_or_path": "Model πŸ€—",
"average": "Average",
"grounding": "Grounding ⚑️",
"instruction_following": "Instruction Following πŸ“",
"planning": "Planning πŸ“…",
"reasoning": "Reasoning πŸ’‘",
"refinement": "Refinement πŸ”©",
"safety": "Safety ⚠️",
"theory_of_mind": "Theory of Mind πŸ€”",
"tool_usage": "Tool Usage πŸ› οΈ",
"multilingual": "Multilingual πŸ‡¬πŸ‡«",
"model_params": "Model Params (B)",
"model_type": "Model Type",
}
BGB_COLUMN_TO_DATATYPE = {
"Model πŸ€—": "markdown",
"Average": "number",
"Grounding ⚑️": "number",
"Instruction Following πŸ“": "number",
"Planning πŸ“…": "number",
"Reasoning πŸ’‘": "number",
"Refinement πŸ”©": "number",
"Safety ⚠️": "number",
"Theory of Mind πŸ€”": "number",
"Tool Usage πŸ› οΈ": "number",
"Multilingual πŸ‡¬πŸ‡«": "number",
"Model Params (B)": "number",
"Model Type": "str",
}
def process_model(model_name):
link = f"https://huggingface.co/{model_name}"
return model_hyperlink(link, model_name)
# TODO: Process base, chat, proprietary models differently
def process_bgb_model(row):
model_name = row.iloc[0]
model_type = row.iloc[1]
if model_type == "Base" or model_type == "Chat":
link = f"https://huggingface.co/{model_name}"
return model_hyperlink(link, model_name)
elif model_type == "Proprietary":
api_model_2_link = {
"gpt-3.5-turbo-1106": "https://platform.openai.com/docs/models/gpt-3-5",
"gpt-3.5-turbo-0125": "https://platform.openai.com/docs/models/gpt-3-5",
"gpt-4-0125-preview": "https://openai.com/blog/new-models-and-developer-products-announced-at-devday",
"gpt-4-1106-preview": "https://openai.com/blog/new-models-and-developer-products-announced-at-devday",
"gpt-4-turbo-2024-04-09": "https://platform.openai.com/docs/models/gpt-4-turbo-and-gpt-4",
"gpt-4o-2024-05-13": "https://openai.com/index/hello-gpt-4o/",
"claude-3-haiku-20240307": "https://www.anthropic.com/news/claude-3-family",
"claude-3-opus-20240229": "https://www.anthropic.com/news/claude-3-family",
"claude-3-sonnet-20240229": "https://www.anthropic.com/news/claude-3-family",
"mistral-large": "https://mistral.ai/news/mistral-large/",
"mistral-medium": "https://mistral.ai/news/la-plateforme/",
"gemini-1.0-pro": "https://deepmind.google/technologies/gemini/pro/",
"gemini-pro-1.5": "https://deepmind.google/technologies/gemini/pro/",
"google/gemini-flash-1.5": "https://deepmind.google/technologies/gemini/flash/",
}
link = api_model_2_link[model_name]
return model_hyperlink(link, model_name)
else:
raise NotImplementedError(f"Model type {model_type} not implemented")
def get_leaderboard_df(llm_perf_df):
df = llm_perf_df.copy()
# transform for leaderboard
df["Model πŸ€—"] = df["Model πŸ€—"].apply(process_bgb_model)
# process quantization for leaderboard
df["Open LLM Score (%)"] = df.apply(lambda x: process_score(x["Open LLM Score (%)"], x["Quantization πŸ—œοΈ"]), axis=1)
return df
def get_bgb_leaderboard_df(eval_df):
df = eval_df.copy()
# transform for leaderboard
df["Model πŸ€—"] = df[["Model πŸ€—", "Model Type"]].apply(process_bgb_model, axis=1)
return df
def create_leaderboard_table(llm_perf_df):
# get dataframe
leaderboard_df = get_leaderboard_df(llm_perf_df)
# create search bar
with gr.Row():
search_bar = gr.Textbox(
label="Model πŸ€—",
info="πŸ” Search for a model name",
elem_id="search-bar",
)
# create checkboxes
with gr.Row():
columns_checkboxes = gr.CheckboxGroup(
label="Columns πŸ“Š",
value=PRIMARY_COLUMNS,
choices=list(LEADERBOARD_COLUMN_TO_DATATYPE.keys()),
info="β˜‘οΈ Select the columns to display",
elem_id="columns-checkboxes",
)
# create table
leaderboard_table = gr.components.Dataframe(
value=leaderboard_df[PRIMARY_COLUMNS],
datatype=list(LEADERBOARD_COLUMN_TO_DATATYPE.values()),
headers=list(LEADERBOARD_COLUMN_TO_DATATYPE.keys()),
elem_id="leaderboard-table",
)
return search_bar, columns_checkboxes, leaderboard_table
def create_bgb_leaderboard_table(eval_df):
# get dataframe
bgb_leaderboard_df = get_bgb_leaderboard_df(eval_df)
# create search bar
with gr.Row():
search_bar = gr.Textbox(
label="Model πŸ€—",
info="πŸ” Search for a model name",
elem_id="search-bar",
)
with gr.Row():
type_checkboxes = gr.CheckboxGroup(
label="Model Type",
value=["Base", "Chat", "Proprietary"],
choices=["Base", "Chat", "Proprietary"],
info="β˜‘οΈ Select the capabilities to display",
elem_id="type-checkboxes",
)
with gr.Row():
param_slider = gr.Slider(
minimum=0, maximum=150, value=7, step=1, interactive=True, label="Model Params (B)", elem_id="param-slider"
)
# create checkboxes
with gr.Row():
columns_checkboxes = gr.CheckboxGroup(
label="Capabilities πŸ“Š",
value=CAPABILITY_COLUMNS,
choices=CAPABILITY_COLUMNS,
info="β˜‘οΈ Select the capabilities to display",
elem_id="columns-checkboxes",
)
# create table
bgb_leaderboard_table = gr.components.Dataframe(
value=bgb_leaderboard_df[list(BGB_COLUMN_MAPPING.values())],
datatype=list(BGB_COLUMN_TO_DATATYPE.values()),
headers=list(BGB_COLUMN_MAPPING.keys()),
elem_id="leaderboard-table",
)
return search_bar, columns_checkboxes, type_checkboxes, param_slider, bgb_leaderboard_table