File size: 9,961 Bytes
b4966ee
 
 
 
78db81b
 
 
b4966ee
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ec6b925
 
 
 
0d4db15
 
ec6b925
0d4db15
78db81b
 
 
 
 
 
0d4db15
 
 
 
 
 
 
 
 
 
 
 
 
 
 
78db81b
 
0d4db15
78db81b
 
0d4db15
 
 
 
 
 
 
78db81b
b4966ee
 
 
a479746
b4966ee
0d4db15
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4966ee
0d4db15
 
 
 
 
 
b4966ee
 
0d4db15
 
 
 
bc83dc3
0d4db15
 
 
 
 
 
 
 
 
 
 
 
 
 
78db81b
0d4db15
 
 
 
 
bc83dc3
 
0d4db15
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
82ad940
 
78db81b
82ad940
 
 
b4966ee
0d4db15
 
 
 
 
 
 
 
569183a
b4966ee
 
 
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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
import gradio as gr
import requests
import pandas as pd
from huggingface_hub.hf_api import SpaceInfo
from huggingface_hub import HfApi, hf_hub_download
from huggingface_hub.repocard import metadata_load

path = f"https://huggingface.co/api/spaces"

def get_blocks_party_spaces():
    r = requests.get(path)
    d = r.json()
    spaces = [SpaceInfo(**x) for x in d]
    blocks_spaces = {}
    for i in range(0,len(spaces)):
        if spaces[i].id.split('/')[0] == 'Gradio-Blocks' and hasattr(spaces[i], 'likes') and spaces[i].id != 'Gradio-Blocks/Leaderboard' and spaces[i].id != 'Gradio-Blocks/README':
            blocks_spaces[spaces[i].id]=spaces[i].likes
    df = pd.DataFrame(
    [{"Spaces_Name": Spaces, "likes": likes} for Spaces,likes in blocks_spaces.items()])
    df = df.sort_values(by=['likes'],ascending=False)
    return df

def make_clickable_model(model_name):
    # remove user from model name
    model_name_show = ' '.join(model_name.split('/')[1:])
    link = "https://huggingface.co/" + model_name
    return f'<a target="_blank" style="text-decoration: underline" href="{link}">{model_name_show}</a>'


def get_mteb_data(task="Clustering", metric="v_measure", lang=None):
    api = HfApi()
    models = api.list_models(filter="mteb")
    df_list = []
    for model in models:
        readme_path = hf_hub_download(model.modelId, filename="README.md")
        meta = metadata_load(readme_path)
        # Use "get" instead of dict indexing to ignore incompat metadata instead of erroring out
        if lang is None:
            out = list(
                map(
                    lambda x: {x["dataset"]["name"].replace("MTEB ", ""): round(list(filter(lambda x: x["type"] == metric, x["metrics"]))[0]["value"], 2)}, 
                    filter(lambda x: x.get("task", {}).get("type", "") == task, meta["model-index"][0]["results"])
                )
            )
        else:
            # Multilingual
            out = list(
                map(
                    lambda x: {x["dataset"]["name"].replace("MTEB ", ""): round(list(filter(lambda x: x["type"] == metric, x["metrics"]))[0]["value"], 2)}, 
                    filter(lambda x: (x.get("task", {}).get("type", "") == task) and (x.get("dataset", {}).get("config", "") in ("default", *lang)), meta["model-index"][0]["results"])
                )
            )
        out = {k: v for d in out for k, v in d.items()}
        out["Model"] = make_clickable_model(model.modelId)
        df_list.append(out)
    df = pd.DataFrame(df_list)
    # Put 'Model' column first
    cols = sorted(list(df.columns))
    cols.insert(0, cols.pop(cols.index("Model")))
    df = df[cols]

    df.fillna('', inplace=True)
    return df.astype(str) # Cast to str as Gradio does not accept floats

block = gr.Blocks()

with block:    
    gr.Markdown("""Leaderboard for XX most popular Blocks Event Spaces. To learn more and join, see <a href="https://huggingface.co/Gradio-Blocks" target="_blank" style="text-decoration: underline">Blocks Party Event</a>""")
    with gr.Tabs():
        with gr.TabItem("Classification"):
            with gr.TabItem("English"):
                with gr.Row():
                    gr.Markdown("""Leaderboard for Classification""")
                with gr.Row():
                    data_classification_en = gr.components.Dataframe(
                        datatype=["markdown"] * 500,
                        type="pandas",
                        col_count=(13, "fixed"),
                    )
                with gr.Row():
                    data_run = gr.Button("Refresh")
                    task_classification_en = gr.Variable(value="Classification")
                    metric_classification_en = gr.Variable(value="accuracy")
                    lang_classification_en = gr.Variable(value=["en"])
                    data_run.click(get_mteb_data, inputs=[task_classification_en, metric_classification_en, lang_classification_en], outputs=data_classification_en)
            with gr.TabItem("Multilingual"):
                with gr.Row():
                    gr.Markdown("""Multilingual Classification""")
                with gr.Row():
                    data_classification = gr.components.Dataframe(
                        datatype=["markdown"] * 500,
                        type="pandas",
                    )
                with gr.Row():
                    data_run = gr.Button("Refresh")
                    task_classification = gr.Variable(value="Classification")
                    metric_classification = gr.Variable(value="accuracy")
                    data_run.click(get_mteb_data, inputs=[task_classification, metric_classification], outputs=data_classification)
        with gr.TabItem("Clustering"):
            with gr.Row():
                gr.Markdown("""Leaderboard for Clustering""")
            with gr.Row():
                data_clustering = gr.components.Dataframe(
                    datatype=["markdown"] * 500,
                    type="pandas",
                )
            with gr.Row():
                data_run = gr.Button("Refresh")
                task_clustering = gr.Variable(value="Clustering")
                metric_clustering = gr.Variable(value="v_measure")
                data_run.click(get_mteb_data, inputs=[task_clustering, metric_clustering], outputs=data_clustering)
        with gr.TabItem("Retrieval"):
            with gr.Row():
                gr.Markdown("""Leaderboard for Retrieval""")
            with gr.Row():
                data_retrieval = gr.components.Dataframe(
                    datatype=["markdown"] * 500,
                    type="pandas",
                )
            with gr.Row():
                data_run = gr.Button("Refresh")
                task_retrieval = gr.Variable(value="Retrieval")
                metric_retrieval = gr.Variable(value="ndcg_at_10")
                data_run.click(get_mteb_data, inputs=[task_retrieval, metric_retrieval], outputs=data_retrieval)
        with gr.TabItem("Reranking"):
            with gr.Row():
                gr.Markdown("""Leaderboard for Reranking""")
            with gr.Row():
                data_reranking = gr.components.Dataframe(
                    datatype=["markdown"] * 500,
                    type="pandas",
                    #col_count=(12, "fixed"),
                )
            with gr.Row():
                data_run = gr.Button("Refresh")
                task_reranking = gr.Variable(value="Reranking")
                metric_reranking = gr.Variable(value="map")
                data_run.click(get_mteb_data, inputs=[task_reranking, metric_reranking], outputs=data_reranking)                
        with gr.TabItem("STS"):
            with gr.TabItem("English"):
                with gr.Row():
                    gr.Markdown("""Leaderboard for STS""")
                with gr.Row():
                    data_sts_en = gr.components.Dataframe(
                        datatype=["markdown"] * 500,
                        type="pandas",
                    )
                with gr.Row():
                    data_run_en = gr.Button("Refresh")
                    task_sts_en = gr.Variable(value="STS")
                    metric_sts_en = gr.Variable(value="cos_sim_spearman")
                    lang_sts_en = gr.Variable(value=["en", "en-en"])
                    data_run.click(get_mteb_data, inputs=[task_sts_en, metric_sts_en, lang_sts_en], outputs=data_sts_en)
            with gr.TabItem("Multilingual"):
                with gr.Row():
                    gr.Markdown("""Leaderboard for STS""")
                with gr.Row():
                    data_sts = gr.components.Dataframe(
                        datatype=["markdown"] * 500,
                        type="pandas",
                    )
                with gr.Row():
                    data_run = gr.Button("Refresh")
                    task_sts = gr.Variable(value="STS")
                    metric_sts = gr.Variable(value="cos_sim_spearman")
                    data_run.click(get_mteb_data, inputs=[task_sts, metric_sts], outputs=data_sts)
        with gr.TabItem("Summarization"):
            with gr.Row():
                gr.Markdown("""Leaderboard for Summarization""")
            with gr.Row():
                data_summarization = gr.components.Dataframe(
                    datatype=["markdown"] * 500,
                    type="pandas",
                )
            with gr.Row():
                data_run = gr.Button("Refresh")
                task_summarization = gr.Variable(value="Summarization")
                metric_summarization = gr.Variable(value="cos_sim_spearman")
                data_run.click(get_mteb_data, inputs=[task_summarization, metric_summarization], outputs=data_summarization)                   
        with gr.TabItem("Blocks Party Leaderboard2"):
            with gr.Row():
                data = gr.components.Dataframe(type="pandas")
            with gr.Row():
                data_run = gr.Button("Refresh")
                data_run.click(get_blocks_party_spaces, inputs=None, outputs=data)
    # running the function on page load in addition to when the button is clicked
    block.load(get_mteb_data, inputs=[task_classification_en, metric_classification_en], outputs=data_classification_en)
    block.load(get_mteb_data, inputs=[task_classification, metric_classification], outputs=data_classification)
    block.load(get_mteb_data, inputs=[task_clustering, metric_clustering], outputs=data_clustering)
    block.load(get_mteb_data, inputs=[task_retrieval, metric_retrieval], outputs=data_retrieval)
    block.load(get_mteb_data, inputs=[task_reranking, metric_reranking], outputs=data_reranking)
    block.load(get_mteb_data, inputs=[task_sts, metric_sts], outputs=data_sts)    
    block.load(get_mteb_data, inputs=[task_summarization, metric_summarization], outputs=data_summarization)        

    block.load(get_blocks_party_spaces, inputs=None, outputs=data)  

block.launch()