Hugging Face's logo

Spaces:
AIR-Bench
/
leaderboard
Restarting on CPU Upgrade

App Files Community
28
leaderboard
File size: 6,797 Bytes 
3478401
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
 
from pathlib import Path

import pandas as pd
import pytest

from src.columns import COL_NAME_RERANKING_MODEL, COL_NAME_RETRIEVAL_MODEL
from src.models import TaskType, model_hyperlink
from src.utils import (
    _update_df_elem,
    calculate_mean,
    filter_models,
    filter_queries,
    get_default_cols,
    get_leaderboard_df,
    get_selected_cols,
    remove_html,
    select_columns,
)

cur_fp = Path(__file__)

NUM_QA_BENCHMARKS_24_05 = 53
NUM_DOC_BENCHMARKS_24_05 = 11
NUM_QA_BENCHMARKS_24_04 = 13
NUM_DOC_BENCHMARKS_24_04 = 15


@pytest.fixture
def toy_df():
    return pd.DataFrame(
        {
            "Retrieval Method": ["bge-m3", "bge-m3", "jina-embeddings-v2-base", "jina-embeddings-v2-base"],
            "Reranking Model": ["bge-reranker-v2-m3", "NoReranker", "bge-reranker-v2-m3", "NoReranker"],
            "Rank 🏆": [1, 2, 3, 4],
            "Revision": ["123", "234", "345", "456"],
            "Submission Date": ["", "", "", ""],
            "Average ⬆️": [0.6, 0.4, 0.3, 0.2],
            "wiki_en": [0.8, 0.7, 0.2, 0.1],
            "wiki_zh": [0.4, 0.1, 0.4, 0.3],
            "news_en": [0.8, 0.7, 0.2, 0.1],
            "news_zh": [0.4, 0.1, 0.2, 0.3],
            "Anonymous Submission": [False, False, False, True],
        }
    )


def test_remove_html():
    model_name = "jina-embeddings-v3"
    html_str = model_hyperlink("https://jina.ai", model_name)
    output_str = remove_html(html_str)
    assert output_str == model_name


def test_calculate_mean():
    valid_row = [1, 3]
    invalid_row = [2, pd.NA]
    df = pd.DataFrame([valid_row, invalid_row], columns=["a", "b"])
    result = list(df.apply(calculate_mean, axis=1))
    assert result[0] == sum(valid_row) / 2
    assert result[1] == -1


@pytest.mark.parametrize(
    "models, expected",
    [
        (["model1", "model3"], 2),
        (["model1", "model_missing"], 1),
        (["model1", "model2", "model3"], 3),
        (
            [
                "model1",
            ],
            1,
        ),
        ([], 3),
    ],
)
def test_filter_models(models, expected):
    df = pd.DataFrame(
        {
            COL_NAME_RERANKING_MODEL: [
                "model1",
                "model2",
                "model3",
            ],
            "col2": [1, 2, 3],
        }
    )
    output_df = filter_models(df, models)
    assert len(output_df) == expected


@pytest.mark.parametrize(
    "query, expected",
    [
        ("model1;model3", 2),
        ("model1;model4", 1),
        ("model1;model2;model3", 3),
        ("model1", 1),
        ("", 3),
    ],
)
def test_filter_queries(query, expected):
    df = pd.DataFrame(
        {
            COL_NAME_RETRIEVAL_MODEL: [
                "model1",
                "model2",
                "model3",
            ],
            COL_NAME_RERANKING_MODEL: [
                "model4",
                "model5",
                "model6",
            ],
        }
    )
    output_df = filter_queries(query, df)
    assert len(output_df) == expected


@pytest.mark.parametrize(
    "task_type, slug, add_fix_cols, expected",
    [
        (TaskType.qa, "air_bench_2404", True, NUM_QA_BENCHMARKS_24_04),
        (TaskType.long_doc, "air_bench_2404", True, NUM_DOC_BENCHMARKS_24_04),
        (TaskType.qa, "air_bench_2405", False, NUM_QA_BENCHMARKS_24_05),
        (TaskType.long_doc, "air_bench_2405", False, NUM_DOC_BENCHMARKS_24_05),
    ],
)
def test_get_default_cols(task_type, slug, add_fix_cols, expected):
    attr_cols = ["Rank 🏆", "Retrieval Method", "Reranking Model", "Revision", "Submission Date", "Average ⬆️"]
    cols, types = get_default_cols(task_type, slug)
    cols_set = frozenset(cols)
    attrs_set = frozenset(attr_cols)
    if add_fix_cols:
        assert attrs_set.issubset(cols_set)
    benchmark_cols = list(cols_set.difference(attrs_set))
    assert len(benchmark_cols) == expected


@pytest.mark.parametrize(
    "task_type, domains, languages, expected",
    [
        (
            TaskType.qa,
            ["wiki", "news"],
            [
                "zh",
            ],
            ["wiki_zh", "news_zh"],
        ),
        (
            TaskType.qa,
            [
                "law",
            ],
            ["zh", "en"],
            ["law_en"],
        ),
        (
            TaskType.long_doc,
            ["healthcare"],
            ["zh", "en"],
            [
                "healthcare_en_pubmed_100k_200k_1",
                "healthcare_en_pubmed_100k_200k_2",
                "healthcare_en_pubmed_100k_200k_3",
                "healthcare_en_pubmed_40k_50k_5_merged",
                "healthcare_en_pubmed_30k_40k_10_merged",
            ],
        ),
    ],
)
def test_get_selected_cols(task_type, domains, languages, expected):
    slug = "air_bench_2404"
    cols = get_selected_cols(task_type, slug, domains, languages)
    assert sorted(cols) == sorted(expected)


@pytest.mark.parametrize("reset_rank", [False])
def test_select_columns(toy_df, reset_rank):
    expected = [
        "Rank 🏆",
        "Retrieval Method",
        "Reranking Model",
        "Revision",
        "Submission Date",
        "Average ⬆️",
        "news_zh",
    ]
    df_result = select_columns(toy_df, ["news"], ["zh"], version_slug="air_bench_2404", reset_ranking=reset_rank)
    assert len(df_result.columns) == len(expected)
    if reset_rank:
        assert df_result["Average ⬆️"].equals(df_result["news_zh"])
    else:
        assert df_result["Average ⬆️"].equals(toy_df["Average ⬆️"])


@pytest.mark.parametrize(
    "reset_rank, show_anony",
    [
        (False, True),
        (True, True),
        (True, False),
    ],
)
def test__update_df_elem(toy_df, reset_rank, show_anony):
    df = _update_df_elem(TaskType.qa, "AIR-Bench_24.04", toy_df, ["news"], ["zh"], [], "", show_anony, reset_rank)
    if show_anony:
        assert df.shape[0] == 4
    else:
        assert df.shape[0] == 3
    if show_anony:
        if reset_rank:
            assert df["Average ⬆️"].equals(df["news_zh"])
        else:
            assert df["Average ⬆️"].equals(toy_df["Average ⬆️"])


@pytest.mark.parametrize(
    "version, task_type",
    [
        ("AIR-Bench_24.04", TaskType.qa),
        ("AIR-Bench_24.04", TaskType.long_doc),
        ("AIR-Bench_24.05", TaskType.qa),
        ("AIR-Bench_24.05", TaskType.long_doc),
    ],
)
def test_get_leaderboard_df(version, task_type):
    from src.loaders import load_raw_eval_results
    from src.models import LeaderboardDataStore, get_safe_name

    raw_data = load_raw_eval_results(cur_fp.parents[1] / f"toydata/eval_results/{version}")
    ds = LeaderboardDataStore(version, get_safe_name(version), raw_data=raw_data)
    df = get_leaderboard_df(ds, task_type, "ndcg_at_10")
    assert df.shape[0] == 1