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 | """All benchmarks tasks used in EuroEval."""
from . import metrics as m
from .data_models import Task
from .enums import GenerativeType, ModelType, TaskGroup
from .prompt_templates import (
LA_TEMPLATES,
MULTIPLE_CHOICE_TEMPLATES,
NER_TEMPLATES,
RC_TEMPLATES,
SENT_TEMPLATES,
SUMM_TEMPLATES,
)
def get_all_tasks() -> dict[str, Task]:
"""Get a list of all the dataset tasks.
Returns:
A mapping between names of dataset tasks and their configurations.
"""
return {cfg.name: cfg for cfg in globals().values() if isinstance(cfg, Task)}
LA = Task(
name="linguistic-acceptability",
task_group=TaskGroup.SEQUENCE_CLASSIFICATION,
template_dict=LA_TEMPLATES,
metrics=[m.mcc_metric, m.macro_f1_metric],
default_num_few_shot_examples=12,
default_max_generated_tokens=10,
default_labels=["correct", "incorrect"],
uses_logprobs=True,
)
NER = Task(
name="named-entity-recognition",
task_group=TaskGroup.TOKEN_CLASSIFICATION,
template_dict=NER_TEMPLATES,
metrics=[m.micro_f1_no_misc_metric, m.micro_f1_metric],
default_num_few_shot_examples=8,
default_max_generated_tokens=128,
default_labels=[
"o",
"b-loc",
"i-loc",
"b-org",
"i-org",
"b-per",
"i-per",
"b-misc",
"i-misc",
],
uses_structured_output=True,
)
RC = Task(
name="reading-comprehension",
task_group=TaskGroup.QUESTION_ANSWERING,
template_dict=RC_TEMPLATES,
metrics=[m.f1_metric, m.em_metric],
default_num_few_shot_examples=4,
default_max_generated_tokens=32,
default_labels=["start_positions", "end_positions"],
)
SENT = Task(
name="sentiment-classification",
task_group=TaskGroup.SEQUENCE_CLASSIFICATION,
template_dict=SENT_TEMPLATES,
metrics=[m.mcc_metric, m.macro_f1_metric],
default_num_few_shot_examples=12,
default_max_generated_tokens=10,
default_labels=["positive", "neutral", "negative"],
uses_logprobs=True,
)
SUMM = Task(
name="summarization",
task_group=TaskGroup.TEXT_TO_TEXT,
template_dict=SUMM_TEMPLATES,
metrics=[m.bert_score_metric, m.rouge_l_metric],
default_num_few_shot_examples=1,
default_max_generated_tokens=256,
default_labels=[],
allowed_model_types=[ModelType.GENERATIVE],
)
KNOW = Task(
name="knowledge",
task_group=TaskGroup.MULTIPLE_CHOICE_CLASSIFICATION,
template_dict=MULTIPLE_CHOICE_TEMPLATES,
metrics=[m.mcc_metric, m.accuracy_metric],
default_num_few_shot_examples=5,
default_max_generated_tokens=10,
default_labels=["a", "b", "c", "d"],
uses_logprobs=True,
)
MCRC = Task(
name="multiple-choice-reading-comprehension",
task_group=TaskGroup.MULTIPLE_CHOICE_CLASSIFICATION,
template_dict=MULTIPLE_CHOICE_TEMPLATES,
metrics=[m.mcc_metric, m.accuracy_metric],
default_num_few_shot_examples=5,
default_max_generated_tokens=10,
default_labels=["a", "b", "c", "d"],
uses_logprobs=True,
)
COMMON_SENSE = Task(
name="common-sense-reasoning",
task_group=TaskGroup.MULTIPLE_CHOICE_CLASSIFICATION,
template_dict=MULTIPLE_CHOICE_TEMPLATES,
metrics=[m.mcc_metric, m.accuracy_metric],
default_num_few_shot_examples=5,
default_max_generated_tokens=10,
default_labels=["a", "b", "c", "d"],
uses_logprobs=True,
)
EUROPEAN_VALUES = Task(
name="european-values",
task_group=TaskGroup.MULTIPLE_CHOICE_CLASSIFICATION,
template_dict=MULTIPLE_CHOICE_TEMPLATES,
metrics=[m.european_values_metric],
default_num_few_shot_examples=0,
default_max_generated_tokens=10,
default_labels=["0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "10"],
allowed_model_types=[ModelType.GENERATIVE],
allowed_generative_types=[
GenerativeType.INSTRUCTION_TUNED,
GenerativeType.REASONING,
],
requires_zero_shot=True,
uses_logprobs=True,
)
SPEED = Task(
name="speed",
task_group=TaskGroup.SPEED,
template_dict={},
metrics=[m.speed_metric, m.speed_short_metric],
default_num_few_shot_examples=0,
default_max_generated_tokens=5,
default_labels=[],
)
|