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
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
|
import unittest
import pytest
import yaml
from huggingface_hub import SpaceCardData
from huggingface_hub.repocard_data import (
CardData,
DatasetCardData,
EvalResult,
ModelCardData,
eval_results_to_model_index,
model_index_to_eval_results,
)
OPEN_LLM_LEADERBOARD_URL = "https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard"
DUMMY_METADATA_WITH_MODEL_INDEX = """
language: en
license: mit
library_name: timm
tags:
- pytorch
- image-classification
datasets:
- beans
metrics:
- acc
model-index:
- name: my-cool-model
results:
- task:
type: image-classification
dataset:
type: beans
name: Beans
metrics:
- type: acc
value: 0.9
source:
name: Open LLM Leaderboard
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard
"""
class BaseCardDataTest(unittest.TestCase):
def test_metadata_behave_as_dict(self):
metadata = CardData(foo="bar")
# .get and __getitem__
self.assertEqual(metadata.get("foo"), "bar")
self.assertEqual(metadata.get("FOO"), None) # case-sensitive
self.assertEqual(metadata["foo"], "bar")
with self.assertRaises(KeyError): # case-sensitive
_ = metadata["FOO"]
# __setitem__
metadata["foo"] = "BAR"
self.assertEqual(metadata.get("foo"), "BAR")
self.assertEqual(metadata["foo"], "BAR")
# __contains__
self.assertTrue("foo" in metadata)
self.assertFalse("FOO" in metadata)
# default value
# Should return default when key is not in metadata
self.assertEqual(metadata.get("FOO", "default"), "default")
# Should return default when key is in metadata but value is None
metadata.FOO = None
self.assertEqual(metadata.get("FOO", "default"), "default")
# export
self.assertEqual(str(metadata), "foo: BAR")
# .pop
self.assertEqual(metadata.pop("foo"), "BAR")
class ModelCardDataTest(unittest.TestCase):
def test_eval_results_to_model_index(self):
expected_results = yaml.safe_load(DUMMY_METADATA_WITH_MODEL_INDEX)
eval_results = [
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
source_name="Open LLM Leaderboard",
source_url=OPEN_LLM_LEADERBOARD_URL,
),
]
model_index = eval_results_to_model_index("my-cool-model", eval_results)
self.assertEqual(model_index, expected_results["model-index"])
def test_model_index_to_eval_results(self):
model_index = [
{
"name": "my-cool-model",
"results": [
{
"task": {
"type": "image-classification",
},
"dataset": {
"type": "cats_vs_dogs",
"name": "Cats vs. Dogs",
},
"metrics": [
{
"type": "acc",
"value": 0.85,
},
{
"type": "f1",
"value": 0.9,
},
],
},
{
"task": {
"type": "image-classification",
},
"dataset": {
"type": "beans",
"name": "Beans",
},
"metrics": [
{
"type": "acc",
"value": 0.9,
"verified": True,
"verifyToken": 1234,
}
],
"source": {
"name": "Open LLM Leaderboard",
"url": OPEN_LLM_LEADERBOARD_URL,
},
},
],
}
]
model_name, eval_results = model_index_to_eval_results(model_index)
self.assertEqual(len(eval_results), 3)
self.assertEqual(model_name, "my-cool-model")
self.assertEqual(eval_results[0].dataset_type, "cats_vs_dogs")
self.assertIsNone(eval_results[0].source_name)
self.assertIsNone(eval_results[0].source_url)
self.assertEqual(eval_results[1].metric_type, "f1")
self.assertEqual(eval_results[1].metric_value, 0.9)
self.assertIsNone(eval_results[1].source_name)
self.assertIsNone(eval_results[1].source_url)
self.assertEqual(eval_results[2].task_type, "image-classification")
self.assertEqual(eval_results[2].dataset_type, "beans")
self.assertEqual(eval_results[2].verified, True)
self.assertEqual(eval_results[2].verify_token, 1234)
self.assertEqual(eval_results[2].source_name, "Open LLM Leaderboard")
self.assertEqual(eval_results[2].source_url, OPEN_LLM_LEADERBOARD_URL)
def test_card_data_requires_model_name_for_eval_results(self):
with pytest.raises(ValueError, match="`eval_results` requires `model_name` to be set."):
ModelCardData(
eval_results=[
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
),
],
)
data = ModelCardData(
model_name="my-cool-model",
eval_results=[
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
),
],
)
model_index = eval_results_to_model_index(data.model_name, data.eval_results)
self.assertEqual(model_index[0]["name"], "my-cool-model")
self.assertEqual(model_index[0]["results"][0]["task"]["type"], "image-classification")
def test_arbitrary_incoming_card_data(self):
data = ModelCardData(
model_name="my-cool-model",
eval_results=[
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
),
],
some_arbitrary_kwarg="some_value",
)
self.assertEqual(data.some_arbitrary_kwarg, "some_value")
data_dict = data.to_dict()
self.assertEqual(data_dict["some_arbitrary_kwarg"], "some_value")
def test_eval_result_with_incomplete_source(self):
# Source url without name: ok
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
source_url=OPEN_LLM_LEADERBOARD_URL,
)
# Source name without url: not ok
with self.assertRaises(ValueError):
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
source_name="Open LLM Leaderboard",
)
def test_model_card_unique_tags(self):
data = ModelCardData(tags=["tag2", "tag1", "tag2", "tag3"])
assert data.tags == ["tag2", "tag1", "tag3"]
def test_remove_top_level_none_values(self):
as_obj = ModelCardData(tags=["tag1", None], foo={"bar": 3, "baz": None}, pipeline_tag=None)
as_dict = as_obj.to_dict()
assert as_obj.tags == ["tag1", None]
assert as_dict["tags"] == ["tag1", None] # none value inside list should be kept
assert as_obj.foo == {"bar": 3, "baz": None}
assert as_dict["foo"] == {"bar": 3, "baz": None} # none value inside dict should be kept
assert as_obj.pipeline_tag is None
assert "pipeline_tag" not in as_dict # top level none value should be removed
def test_eval_results_requires_evalresult_type(self):
with pytest.raises(ValueError, match="should be of type `EvalResult` or a list of `EvalResult`"):
ModelCardData(model_name="my-cool-model", eval_results="this is not an EvalResult")
with pytest.raises(ValueError, match="should be of type `EvalResult` or a list of `EvalResult`"):
ModelCardData(model_name="my-cool-model", eval_results=["accuracy: 0.9", "f1: 0.85"])
data = ModelCardData(
model_name="my-cool-model",
eval_results="this is not an EvalResult",
ignore_metadata_errors=True,
)
assert data.eval_results is not None and data.eval_results == "this is not an EvalResult"
def test_model_name_required_with_eval_results(self):
with pytest.raises(ValueError, match="`eval_results` requires `model_name` to be set"):
ModelCardData(
eval_results=[
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
),
],
)
eval_results = [
EvalResult(
task_type="image-classification",
dataset_type="beans",
dataset_name="Beans",
metric_type="acc",
metric_value=0.9,
),
]
data = ModelCardData(
eval_results=eval_results,
ignore_metadata_errors=True,
)
assert data.eval_results is not None and data.eval_results == eval_results
class DatasetCardDataTest(unittest.TestCase):
def test_train_eval_index_keys_updated(self):
train_eval_index = [
{
"config": "plain_text",
"task": "text-classification",
"task_id": "binary_classification",
"splits": {"train_split": "train", "eval_split": "test"},
"col_mapping": {"text": "text", "label": "target"},
"metrics": [
{
"type": "accuracy",
"name": "Accuracy",
},
{"type": "f1", "name": "F1 macro", "args": {"average": "macro"}},
],
}
]
card_data = DatasetCardData(
language="en",
license="mit",
pretty_name="My Cool Dataset",
train_eval_index=train_eval_index,
)
# The init should have popped this out of kwargs and into train_eval_index attr
self.assertEqual(card_data.train_eval_index, train_eval_index)
# Underlying train_eval_index gets converted to train-eval-index in DatasetCardData._to_dict.
# So train_eval_index should be None in the dict
self.assertTrue(card_data.to_dict().get("train_eval_index") is None)
# And train-eval-index should be in the dict
self.assertEqual(card_data.to_dict()["train-eval-index"], train_eval_index)
class SpaceCardDataTest(unittest.TestCase):
def test_space_card_data(self) -> None:
card_data = SpaceCardData(
title="Dreambooth Training",
license="mit",
sdk="gradio",
duplicated_from="multimodalart/dreambooth-training",
)
self.assertEqual(
card_data.to_dict(),
{
"title": "Dreambooth Training",
"sdk": "gradio",
"license": "mit",
"duplicated_from": "multimodalart/dreambooth-training",
},
)
self.assertIsNone(card_data.tags) # SpaceCardData has some default attributes
|
