Hugging Face's logo

Spaces:
allenai
/
URIAL-Bench
Running

App Files Community
3
URIAL-Bench
File size: 4,971 Bytes 
810d25e
 
a415f27
 
 
 
810d25e
 
a415f27
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e6fa278
 
 
a415f27
810d25e
 
 
 
 
 
 
a415f27
 
 
 
 
 
1596e83
 
 
 
 
 
 
 
 
 
 
 
a415f27
a085dc1
 
a415f27
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e6fa278
 
a415f27
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
 
"""A gradio app that renders a static leaderboard. This is used for Hugging Face Space.""" 
import argparse  
import gradio as gr
import numpy as np
import pandas as pd
import gradio as gr
import pandas as pd 
from constants import BANNER, INTRODUCTION_TEXT, CITATION_TEXT, METRICS_TAB_TEXT, DIR_OUTPUT_REQUESTS  

LAST_UPDATED = "Feb 27th 2024"

css = """
.markdown-text{font-size: 16pt}
th {
  text-align: center;
}
td {
  font-size: 16px; /* Adjust the font size as needed */
  text-align: center;
}

"""

column_names = {
    "model": "Model",
    "Overall": "All 🎯",    
    "Turn 1": "Turn 1️⃣", 
    "Turn 2": "Turn 2️⃣", 
    }

model_info = {
    "gpt-4": {"hf_name": "https://platform.openai.com/", "pretty_name": "gpt-4"}, 
    "gpt-3.5-turbo": {"hf_name": "https://platform.openai.com/", "pretty_name": "gpt-3.5-turbo"}, 
    "Llama-2-70b-hf": {"hf_name": "meta-llama/Llama-2-70b-hf", "pretty_name": "Llama-2-70B"}, 
    "Llama-2-13b-hf": {"hf_name": "meta-llama/Llama-2-13b-hf", "pretty_name": "Llama-2-13B"}, 
    "Llama-2-7b-hf": {"hf_name": "meta-llama/Llama-2-7b-hf", "pretty_name": "Llama-2-7B"}, 
    "Mixtral-8x7B-v0.1": {"hf_name": "mistralai/Mixtral-8x7B-v0.1", "pretty_name": "Mixtral-8x7B"}, 
    "Mistral-7b-v0.1": {"hf_name": "mistralai/Mistral-7B-v0.1", "pretty_name": "Mistral-7B"}, 
    "Yi-34B": {"hf_name": "01-ai/Yi-34B", "pretty_name": "Yi-34B"}, 
    "Yi-6B": {"hf_name": "01-ai/Yi-6B", "pretty_name": "Yi-6B"}, 
    "gemma-7b": {"hf_name": "google/gemma-7b", "pretty_name": "Gemma-7B"}, 
    "gemma-2b": {"hf_name": "google/gemma-2b", "pretty_name": "Gemma-2B"}, 
    "phi-2": {"hf_name": "microsoft/phi-2", "pretty_name": "Phi-2 (2.7B)"}, 
    "olmo": {"hf_name": "allenai/OLMo-7B", "pretty_name": "OLMo-7B"},  
} 


def formatter(x):
    x = round(x, 2)
    return x

def make_clickable_model(model_name, model_info):
    if model_info[model_name]['hf_name'].startswith("http"):
        link = model_info[model_name]['hf_name']
    else:
        link = f"https://huggingface.co/{model_info[model_name]['hf_name']}"
    return f'<a target="_blank" href="{link}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{model_info[model_name]["pretty_name"]}</a>'


def build_demo(original_df, TYPES):
    with gr.Blocks(theme=gr.themes.Soft(), css=css) as demo:
        # gr.HTML(BANNER, elem_id="banner")
        gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")

        # with gr.Tabs(elem_classes="tab-buttons") as tabs:
        #     with gr.TabItem("🏅 Leaderboard", elem_id="od-benchmark-tab-table", id=0):
        leaderboard_table = gr.components.Dataframe(
            value=original_df,
            datatype=TYPES,
            height=1000,
            wrap=False,
            elem_id="leaderboard-table",
            interactive=False,
            visible=True,
            min_width=60,
            )

            # with gr.TabItem("📈 Metrics", elem_id="od-benchmark-tab-table", id=1):
            #     gr.Markdown(METRICS_TAB_TEXT, elem_classes="markdown-text")
 

        gr.Markdown(f"Last updated on **{LAST_UPDATED}**", elem_classes="markdown-text")
        
        with gr.Row():
            with gr.Accordion("📙 Citation", open=False):
                gr.Textbox(
                    value=CITATION_TEXT, lines=7,
                    label="Copy the BibTeX to cite URIAL and MT-Bench",
                    elem_id="citation-button",
                    show_copy_button=True)
                # ).style(show_copy_button=True)

    return demo

if __name__ == "__main__": 

    parser = argparse.ArgumentParser()
    parser.add_argument("--share", action="store_true")
    parser.add_argument("--result_file", help="Path to results table", default="leaderboard_data.jsonl")
    args = parser.parse_args()

    bench_results = args.result_file
 
    original_df = pd.read_json(bench_results, lines=True)

    print(original_df.columns)

    for col in original_df.columns:
        if col == "model":
            original_df[col] = original_df[col].apply(lambda x: x.replace(x, make_clickable_model(x, model_info)))
        else: 
            original_df[col] = original_df[col].apply(formatter) # For numerical values
    
    # Define the first column explicitly, add 'Overall' as the second column, and then append the rest excluding 'Overall'
    new_order = [original_df.columns[0], 'Overall'] + [col for col in original_df.columns if col not in [original_df.columns[0], 'Overall']]

    # Reorder the DataFrame columns using the new order
    reordered_df = original_df[new_order]

    reordered_df.sort_values(by='Overall', inplace=True, ascending=False)

    reordered_df.rename(columns=column_names, inplace=True)

    # COLS = [c.name for c in fields(AutoEvalColumn)]
    # TYPES = [c.type for c in fields(AutoEvalColumn)]

    TYPES = ["markdown", "number"]
    demo = build_demo(reordered_df, TYPES)
    demo.launch(share=args.share)