__init__.py 16.3 KB
Newer Older
1
import collections
2
3
import logging
import os
4
from functools import partial
5
from typing import Dict, List, Mapping, Optional, Union
&'s avatar
& committed
6

7
from lm_eval import utils
8
9
from lm_eval.api.task import ConfigurableGroup, ConfigurableTask, GroupConfig, Task

10
11

GROUP_ONLY_KEYS = list(GroupConfig().to_dict().keys())
lintangsutawika's avatar
lintangsutawika committed
12

13

14
15
16
class TaskManager:
    """TaskManager indexes all tasks from the default `lm_eval/tasks/`
    and an optional directory if provided.
17

18
19
    """

20
    def __init__(self, verbosity="INFO", include_path: Optional[str] = None) -> None:
21
22
23
24
25
        self.verbosity = verbosity
        self.include_path = include_path
        self.logger = utils.eval_logger
        self.logger.setLevel(getattr(logging, f"{verbosity}"))

26
        self._task_index = self.initialize_tasks(include_path=include_path)
27
        self._all_tasks = sorted(list(self._task_index.keys()))
28

29
        self.task_group_map = collections.defaultdict(list)
30

31
32
    def initialize_tasks(self, include_path: Optional[str] = None):
        """Creates a dictionary of tasks index.
33
34
35
36
37
38
39
40
41
42
43
44

        :param include_path: str = None
            An additional path to be searched for tasks

        :return
            Dictionary of task names as key and task metadata
        """
        all_paths = [os.path.dirname(os.path.abspath(__file__)) + "/"]
        if include_path is not None:
            if isinstance(include_path, str):
                include_path = [include_path]
            all_paths.extend(include_path)
45

46
47
48
49
        task_index = {}
        for task_dir in all_paths:
            tasks = self._get_task_and_group(task_dir)
            task_index = {**tasks, **task_index}
lintangsutawika's avatar
format  
lintangsutawika committed
50

51
52
53
54
55
56
57
58
59
60
61
        return task_index

    @property
    def all_tasks(self):
        return self._all_tasks

    @property
    def task_index(self):
        return self._task_index

    def match_tasks(self, task_list):
62
        return utils.pattern_match(task_list, self.all_tasks)
63

64
    def _name_is_registered(self, name) -> bool:
65
66
67
68
        if name in self.all_tasks:
            return True
        return False

69
    def _name_is_task(self, name) -> bool:
70
71
72
73
        if self._name_is_registered(name) and ("task" in self.task_index[name]["type"]):
            return True
        return False

lintangsutawika's avatar
lintangsutawika committed
74
75
76
77
78
79
80
    def _name_is_tag(self, name) -> bool:
        if self._name_is_registered(name) and (
            self.task_index[name]["type"] == "tag"
        ):
            return True
        return False

81
    def _name_is_group(self, name) -> bool:
82
83
84
        if self._name_is_registered(name) and (
            self.task_index[name]["type"] == "group"
        ):
85
86
87
88
            return True
        return False

    def _name_is_python_task(self, name):
89
90
91
        if self._name_is_registered(name) and (
            self.task_index[name]["type"] == "python_task"
        ):
92
93
94
            return True
        return False

95
    def _config_is_task(self, config) -> bool:
96
97
98
99
        if ("task" in config) and isinstance(config["task"], str):
            return True
        return False

100
    def _config_is_group(self, config) -> bool:
101
102
103
104
        if ("task" in config) and isinstance(config["task"], list):
            return True
        return False

105
    def _config_is_python_task(self, config) -> bool:
106
107
108
109
110
        if "class" in config:
            return True
        return False

    def _get_yaml_path(self, name):
111
112
        if name not in self.task_index:
            raise ValueError
113
114
115
        return self.task_index[name]["yaml_path"]

    def _get_config(self, name):
116
117
        if name not in self.task_index:
            raise ValueError
118
119
120
121
122
123
124
        yaml_path = self._get_yaml_path(name)
        if yaml_path == -1:
            return {}
        else:
            return utils.load_yaml_config(yaml_path, mode="full")

    def _get_tasklist(self, name):
125
126
        if self._name_is_task(name):
            raise ValueError
127
128
129
130
131
132
133
134
135
136
137
138
        return self.task_index[name]["task"]

    def _process_alias(self, config, group=None):
        # If the group is not the same as the original
        # group which the group alias was intended for,
        # Set the group_alias to None instead.
        if ("group_alias" in config) and ("group" in config) and group is not None:
            if config["group"] != group:
                config["group_alias"] = None
        return config

    def _load_individual_task_or_group(
139
        self,
140
141
142
143
144
        name_or_config: Optional[Union[str, dict]] = None,
        parent_name: Optional[str] = None,
        update_config: Optional[dict] = None,
        yaml_path: Optional[str] = None,
    ) -> Mapping:
145
        def load_task(config, task):
146
            if "include" in config:
147
148
                config = {
                    **utils.load_yaml_config(
149
                        yaml_path=None,
150
151
                        yaml_config={"include": config.pop("include")},
                        mode="full",
152
153
154
                    ),
                    **config,
                }
155
156
157
158
159
160
161
162
163
164
165
166
            if self._config_is_python_task(config):
                task_object = config["class"]()
            else:
                task_object = ConfigurableTask(config=config)
            return {task: task_object}

        if isinstance(name_or_config, str):
            if update_config is not None:
                # Process name_or_config as a dict instead
                name_or_config = {"task": name_or_config, **update_config}
            elif self._name_is_task(name_or_config):
                task_config = self._get_config(name_or_config)
167
                return load_task(task_config, task=name_or_config)
168
            else:
169
170
171
172
                subtask_list = self._get_tasklist(name_or_config)
                if subtask_list == -1:
                    group_config = self._get_config(name_or_config)
                    subtask_list = group_config["task"]
lintangsutawika's avatar
lintangsutawika committed
173
174
175
                    group_name = ConfigurableGroup(config=group_config)
                else:
                    group_name = name_or_config
176

177
178
        if isinstance(name_or_config, dict):
            if update_config is not None:
179
                name_or_config = {
180
181
                    **name_or_config,
                    **update_config,
182
183
                }

184
185
186
187
188
189
            if self._config_is_task(name_or_config):
                name = name_or_config["task"]
                # If the name is registered as a group
                # if self._name_is_task(name) is False:
                if self._name_is_group(name):
                    group_name = name
190
191
192
                    update_config = {
                        k: v for k, v in name_or_config.items() if k != "task"
                    }
193
194
                    subtask_list = self._get_tasklist(name)
                    if subtask_list == -1:
195
196
197
                        group_config = self._get_config(name)
                        subtask_list = group_config["task"]
                    group_name = ConfigurableGroup(config=group_config)
198
199
200
201
202
203
204
                else:
                    if self._name_is_registered(name):
                        base_task_config = self._get_config(name)

                        # Check if this is a duplicate.
                        if parent_name is not None:
                            name_or_config["group"] = parent_name
205
206
207
208
209
210
211
212
                            num_duplicate = len(
                                list(
                                    filter(
                                        lambda x: x.startswith(name),
                                        self.task_group_map[parent_name],
                                    )
                                )
                            )
213
214
215
216
                            if num_duplicate > 0:
                                name = f"{name}-{num_duplicate}"
                            self.task_group_map[parent_name].append(name)

217
218
219
220
                        task_config = {
                            **base_task_config,
                            **name_or_config,
                        }
221
222
                    else:
                        task_config = name_or_config
223
                    return load_task(task_config, task=name)
224
225
226
            else:
                group_name = name_or_config["group"]
                subtask_list = name_or_config["task"]
227
                if set(name_or_config.keys()) > {"task", "group"}:
228
                    update_config = {
229
230
                        k: v
                        for k, v in name_or_config.items()
231
                        if k not in GROUP_ONLY_KEYS
232
                    }
233
                group_name = ConfigurableGroup(config=name_or_config)
234

235
236
237
238
239
240
        fn = partial(
            self._load_individual_task_or_group,
            parent_name=group_name,
            update_config=update_config,
            yaml_path=yaml_path,
        )
lintangsutawika's avatar
lintangsutawika committed
241
        return {group_name: dict(collections.ChainMap(*map(fn, reversed(subtask_list))))}
242

243
    def load_task_or_group(self, task_list: Optional[Union[str, list]] = None) -> dict:
244
        """Loads a dictionary of task objects from a list
245

246
247
        :param task_list: Union[str, list] = None
            Single string or list of string of task names to be loaded
248

249
250
251
252
253
        :return
            Dictionary of task objects
        """
        if isinstance(task_list, str):
            task_list = [task_list]
254

255
        all_loaded_tasks = dict(
256
            collections.ChainMap(*map(self._load_individual_task_or_group, task_list))
257
258
259
260
261
262
263
        )
        return all_loaded_tasks

    def load_config(self, config: Dict):
        return self._load_individual_task_or_group(config)

    def _get_task_and_group(self, task_dir: str):
264
        """Creates a dictionary of tasks index with the following metadata,
lintangsutawika's avatar
lintangsutawika committed
265
        - `type`, that can be either `task`, `python_task`, `group` or `tags`.
266
267
            `task` refer to regular task configs, `python_task` are special
            yaml files that only consists of `task` and `class` parameters.
lintangsutawika's avatar
lintangsutawika committed
268
269
            `group` are group configs. `tags` are labels that can be assigned
            to tasks to assist in sorting and calling tasks of certain themes.
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
        - `yaml_path`, path to the yaml file. If the entry is a `group` that
            was configured through a task config, the yaml_path will be -1
            and all subtasks will be listed in `task` (see below)
        - `task`, reserved for entries with `type` as `group`. This will list
            all subtasks. When a group config is created (as opposed to task
            config having `group` parameter set), this will be set to -1 to
            avoid recursive indexing. The whole list of subtasks will be loaded
            at evaluation.

        :param task_dir: str
            A directory to check for tasks

        :return
            Dictionary of task names as key and task metadata
        """
        tasks_and_groups = collections.defaultdict()
        for root, _, file_list in os.walk(task_dir):
            for f in file_list:
                if f.endswith(".yaml"):
                    yaml_path = os.path.join(root, f)
                    config = utils.load_yaml_config(yaml_path, mode="simple")
                    if self._config_is_python_task(config):
                        # This is a python class config
                        tasks_and_groups[config["task"]] = {
                            "type": "python_task",
                            "yaml_path": yaml_path,
                        }
                    elif self._config_is_group(config):
                        # This is a group config
                        tasks_and_groups[config["group"]] = {
                            "type": "group",
301
302
303
304
305
                            "task": -1,  # This signals that
                            # we don't need to know
                            # the task list for indexing
                            # as it can be loaded
                            # when called.
306
307
                            "yaml_path": yaml_path,
                        }
lintangsutawika's avatar
lintangsutawika committed
308

309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
                        # # Registered the level 1 tasks from a group config
                        # for config in config["task"]:
                        #     if isinstance(config, dict) and self._config_is_task(config):
                        #         task = config["task"]
                        #         tasks_and_groups[task] = {
                        #             "type": "task",
                        #             "yaml_path": yaml_path,
                        #             }

                    elif self._config_is_task(config):
                        # This is a task config
                        task = config["task"]
                        tasks_and_groups[task] = {
                            "type": "task",
                            "yaml_path": yaml_path,
324
                        }
325

lintangsutawika's avatar
lintangsutawika committed
326
327
328
329
                        if "tag" in config:
                            tag = config["tag"]
                            if isinstance(config["tag"], str):
                                tag = [tag]
330

lintangsutawika's avatar
lintangsutawika committed
331
332
333
334
                            for tag in tag:
                                if tag not in tasks_and_groups:
                                    tasks_and_groups[tag] = {
                                        "type": "tag",
335
336
337
338
                                        "task": [task],
                                        "yaml_path": -1,
                                    }
                                else:
lintangsutawika's avatar
lintangsutawika committed
339
                                    tasks_and_groups[tag]["task"].append(task)
340
341
342
343
                    else:
                        self.logger.debug(f"File {f} in {root} could not be loaded")

        return tasks_and_groups
lintangsutawika's avatar
lintangsutawika committed
344

345

346
347
348
349
350
351
352
def get_task_name_from_config(task_config: Dict[str, str]) -> str:
    if "task" in task_config:
        return task_config["task"]
    if "dataset_name" in task_config:
        return "{dataset_path}_{dataset_name}".format(**task_config)
    else:
        return "{dataset_path}".format(**task_config)
lintangsutawika's avatar
lintangsutawika committed
353

354

lintangsutawika's avatar
lintangsutawika committed
355
def get_task_name_from_object(task_object):
356
357
    if hasattr(task_object, "config"):
        return task_object._config["task"]
lintangsutawika's avatar
lintangsutawika committed
358
359
360
361
362
363
364
365
366

    # TODO: scrap this
    # this gives a mechanism for non-registered tasks to have a custom name anyways when reporting
    return (
        task_object.EVAL_HARNESS_NAME
        if hasattr(task_object, "EVAL_HARNESS_NAME")
        else type(task_object).__name__
    )

367
368

def get_task_dict(
369
    task_name_list: Union[str, List[Union[str, Dict, Task]]],
370
    task_manager: Optional[TaskManager] = None,
371
):
372
    """Creates a dictionary of task objects from either a name of task, config, or prepared Task object.
lintangsutawika's avatar
lintangsutawika committed
373

374
375
376
377
378
379
380
    :param task_name_list: List[Union[str, Dict, Task]]
        Name of model or LM object, see lm_eval.models.get_model
    :param task_manager: TaskManager = None
        A TaskManager object that stores indexed tasks. If not set,
        task_manager will load one. This should be set by the user
        if there are additional paths that want to be included
        via `include_path`
381

382
383
384
385
    :return
        Dictionary of task objects
    """
    task_name_from_string_dict = {}
386
387
388
    task_name_from_config_dict = {}
    task_name_from_object_dict = {}

389
    if isinstance(task_name_list, str):
lintangsutawika's avatar
lintangsutawika committed
390
        task_name_list = [task_name_list]
391
392
393
394
395
396
397
398
399
    elif isinstance(task_name_list, list):
        if not all([isinstance(task, (str, dict, Task)) for task in task_name_list]):
            raise TypeError(
                "Expected all list items to be of types 'str', 'dict', or 'Task', but at least one entry did not match."
            )
    else:
        raise TypeError(
            f"Expected a 'str' or 'list' but received {type(task_name_list)}."
        )
lintangsutawika's avatar
lintangsutawika committed
400

401
402
403
404
405
    string_task_name_list = [task for task in task_name_list if isinstance(task, str)]
    others_task_name_list = [task for task in task_name_list if ~isinstance(task, str)]
    if len(string_task_name_list) > 0:
        if task_manager is None:
            task_manager = TaskManager()
lintangsutawika's avatar
lintangsutawika committed
406

407
408
409
        task_name_from_string_dict = task_manager.load_task_or_group(
            string_task_name_list
        )
410

411
412
    for task_element in others_task_name_list:
        if isinstance(task_element, dict):
413
414
            task_name_from_config_dict = {
                **task_name_from_config_dict,
415
                **task_manager.load_config(config=task_element),
416
417
418
419
420
            }

        elif isinstance(task_element, Task):
            task_name_from_object_dict = {
                **task_name_from_object_dict,
lintangsutawika's avatar
lintangsutawika committed
421
                get_task_name_from_object(task_element): task_element,
422
            }
lintangsutawika's avatar
lintangsutawika committed
423

424
    if not set(task_name_from_string_dict.keys()).isdisjoint(
lintangsutawika's avatar
lintangsutawika committed
425
        set(task_name_from_object_dict.keys())
426
427
    ):
        raise ValueError
428

lintangsutawika's avatar
lintangsutawika committed
429
    return {
430
        **task_name_from_string_dict,
lintangsutawika's avatar
lintangsutawika committed
431
        **task_name_from_config_dict,
432
        **task_name_from_object_dict,
lintangsutawika's avatar
lintangsutawika committed
433
    }