File size: 35,532 Bytes
33d4721
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
import ast
from dataclasses import dataclass
from typing import Optional

import pandas as pd
from datasets import ClassLabel, Dataset, DatasetDict, Sequence
from sklearn.model_selection import train_test_split

from autotrain import logger


RESERVED_COLUMNS = ["autotrain_text", "autotrain_label", "autotrain_question", "autotrain_answer"]
LLM_RESERVED_COLUMNS = [
    "autotrain_prompt",
    "autotrain_context",
    "autotrain_rejected_text",
    "autotrain_prompt_start",
]


@dataclass
class TextBinaryClassificationPreprocessor:
    """
    A preprocessor class for binary text classification tasks.

    Attributes:
        train_data (pd.DataFrame): The training data.
        text_column (str): The name of the column containing text data.
        label_column (str): The name of the column containing label data.
        username (str): The username for the Hugging Face Hub.
        project_name (str): The project name for saving datasets.
        token (str): The authentication token for the Hugging Face Hub.
        valid_data (Optional[pd.DataFrame]): The validation data. Defaults to None.
        test_size (Optional[float]): The proportion of the dataset to include in the validation split. Defaults to 0.2.
        seed (Optional[int]): The random seed for splitting the data. Defaults to 42.
        convert_to_class_label (Optional[bool]): Whether to convert labels to class labels. Defaults to False.
        local (Optional[bool]): Whether to save the dataset locally. Defaults to False.

    Methods:
        __post_init__(): Validates the presence of required columns in the dataframes and checks for reserved column names.
        split(): Splits the training data into training and validation sets if validation data is not provided.
        prepare_columns(train_df, valid_df): Prepares the columns for training and validation dataframes.
        prepare(): Prepares the datasets for training and validation, converts labels if required, and saves or uploads the datasets.
    """

    train_data: pd.DataFrame
    text_column: str
    label_column: str
    username: str
    project_name: str
    token: str
    valid_data: Optional[pd.DataFrame] = None
    test_size: Optional[float] = 0.2
    seed: Optional[int] = 42
    convert_to_class_label: Optional[bool] = False
    local: Optional[bool] = False

    def __post_init__(self):
        # check if text_column and label_column are in train_data
        if self.text_column not in self.train_data.columns:
            raise ValueError(f"{self.text_column} not in train data")
        if self.label_column not in self.train_data.columns:
            raise ValueError(f"{self.label_column} not in train data")
        # check if text_column and label_column are in valid_data
        if self.valid_data is not None:
            if self.text_column not in self.valid_data.columns:
                raise ValueError(f"{self.text_column} not in valid data")
            if self.label_column not in self.valid_data.columns:
                raise ValueError(f"{self.label_column} not in valid data")

        # make sure no reserved columns are in train_data or valid_data
        for column in RESERVED_COLUMNS:
            if column in self.train_data.columns:
                raise ValueError(f"{column} is a reserved column name")
            if self.valid_data is not None:
                if column in self.valid_data.columns:
                    raise ValueError(f"{column} is a reserved column name")

    def split(self):
        if self.valid_data is not None:
            return self.train_data, self.valid_data
        else:
            train_df, valid_df = train_test_split(
                self.train_data,
                test_size=self.test_size,
                random_state=self.seed,
                stratify=self.train_data[self.label_column],
            )
            train_df = train_df.reset_index(drop=True)
            valid_df = valid_df.reset_index(drop=True)
            return train_df, valid_df

    def prepare_columns(self, train_df, valid_df):
        train_df.loc[:, "autotrain_text"] = train_df[self.text_column]
        train_df.loc[:, "autotrain_label"] = train_df[self.label_column]
        valid_df.loc[:, "autotrain_text"] = valid_df[self.text_column]
        valid_df.loc[:, "autotrain_label"] = valid_df[self.label_column]

        # drop text_column and label_column
        train_df = train_df.drop(columns=[self.text_column, self.label_column])
        valid_df = valid_df.drop(columns=[self.text_column, self.label_column])
        return train_df, valid_df

    def prepare(self):
        train_df, valid_df = self.split()
        train_df, valid_df = self.prepare_columns(train_df, valid_df)

        train_df.loc[:, "autotrain_label"] = train_df["autotrain_label"].astype(str)
        valid_df.loc[:, "autotrain_label"] = valid_df["autotrain_label"].astype(str)

        label_names = sorted(set(train_df["autotrain_label"].unique().tolist()))

        train_df = Dataset.from_pandas(train_df)
        valid_df = Dataset.from_pandas(valid_df)

        if self.convert_to_class_label:
            train_df = train_df.cast_column("autotrain_label", ClassLabel(names=label_names))
            valid_df = valid_df.cast_column("autotrain_label", ClassLabel(names=label_names))

        if self.local:
            dataset = DatasetDict(
                {
                    "train": train_df,
                    "validation": valid_df,
                }
            )
            dataset.save_to_disk(f"{self.project_name}/autotrain-data")
        else:
            train_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="train",
                private=True,
                token=self.token,
            )
            valid_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="validation",
                private=True,
                token=self.token,
            )

        if self.local:
            return f"{self.project_name}/autotrain-data"
        return f"{self.username}/autotrain-data-{self.project_name}"


class TextMultiClassClassificationPreprocessor(TextBinaryClassificationPreprocessor):
    """
    TextMultiClassClassificationPreprocessor is a class for preprocessing text data for multi-class classification tasks.

    This class inherits from TextBinaryClassificationPreprocessor and is designed to handle scenarios where the text data
    needs to be classified into more than two categories.

    Methods:
        Inherits all methods from TextBinaryClassificationPreprocessor.

    Attributes:
        Inherits all attributes from TextBinaryClassificationPreprocessor.
    """

    pass


class TextSingleColumnRegressionPreprocessor(TextBinaryClassificationPreprocessor):
    """
    A preprocessor class for single-column regression tasks, inheriting from TextBinaryClassificationPreprocessor.

    Methods
    -------
    split():
        Splits the training data into training and validation sets. If validation data is already provided, it returns
        the training and validation data as is. Otherwise, it performs a train-test split on the training data.

    prepare():
        Prepares the training and validation datasets by splitting the data, preparing the columns, and converting
        them to Hugging Face Datasets. The datasets are then either saved locally or pushed to the Hugging Face Hub,
        depending on the `local` attribute.
    """

    def split(self):
        if self.valid_data is not None:
            return self.train_data, self.valid_data
        else:
            train_df, valid_df = train_test_split(
                self.train_data,
                test_size=self.test_size,
                random_state=self.seed,
            )
            train_df = train_df.reset_index(drop=True)
            valid_df = valid_df.reset_index(drop=True)
            return train_df, valid_df

    def prepare(self):
        train_df, valid_df = self.split()
        train_df, valid_df = self.prepare_columns(train_df, valid_df)

        train_df = Dataset.from_pandas(train_df)
        valid_df = Dataset.from_pandas(valid_df)

        if self.local:
            dataset = DatasetDict(
                {
                    "train": train_df,
                    "validation": valid_df,
                }
            )
            dataset.save_to_disk(f"{self.project_name}/autotrain-data")
        else:
            train_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="train",
                private=True,
                token=self.token,
            )
            valid_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="validation",
                private=True,
                token=self.token,
            )

        if self.local:
            return f"{self.project_name}/autotrain-data"
        return f"{self.username}/autotrain-data-{self.project_name}"


class TextTokenClassificationPreprocessor(TextBinaryClassificationPreprocessor):
    """
    A preprocessor class for text token classification tasks, inheriting from TextBinaryClassificationPreprocessor.

    Methods
    -------
    split():
        Splits the training data into training and validation sets. If validation data is already provided, it returns
        the training and validation data as is. Otherwise, it splits the training data based on the test size and seed.

    prepare():
        Prepares the training and validation data for token classification. This includes splitting the data, preparing
        columns, evaluating text and label columns, and converting them to datasets. The datasets are then either saved
        locally or pushed to the Hugging Face Hub based on the configuration.
    """

    def split(self):
        if self.valid_data is not None:
            return self.train_data, self.valid_data
        else:
            train_df, valid_df = train_test_split(
                self.train_data,
                test_size=self.test_size,
                random_state=self.seed,
            )
            train_df = train_df.reset_index(drop=True)
            valid_df = valid_df.reset_index(drop=True)
            return train_df, valid_df

    def prepare(self):
        train_df, valid_df = self.split()
        train_df, valid_df = self.prepare_columns(train_df, valid_df)
        try:
            train_df.loc[:, "autotrain_text"] = train_df["autotrain_text"].apply(lambda x: ast.literal_eval(x))
            valid_df.loc[:, "autotrain_text"] = valid_df["autotrain_text"].apply(lambda x: ast.literal_eval(x))
        except ValueError:
            logger.warning("Unable to do ast.literal_eval on train_df['autotrain_text']")
            logger.warning("assuming autotrain_text is already a list")
        try:
            train_df.loc[:, "autotrain_label"] = train_df["autotrain_label"].apply(lambda x: ast.literal_eval(x))
            valid_df.loc[:, "autotrain_label"] = valid_df["autotrain_label"].apply(lambda x: ast.literal_eval(x))
        except ValueError:
            logger.warning("Unable to do ast.literal_eval on train_df['autotrain_label']")
            logger.warning("assuming autotrain_label is already a list")

        label_names_train = sorted(set(train_df["autotrain_label"].explode().unique().tolist()))
        label_names_valid = sorted(set(valid_df["autotrain_label"].explode().unique().tolist()))
        label_names = sorted(set(label_names_train + label_names_valid))

        train_df = Dataset.from_pandas(train_df)
        valid_df = Dataset.from_pandas(valid_df)

        if self.convert_to_class_label:
            train_df = train_df.cast_column("autotrain_label", Sequence(ClassLabel(names=label_names)))
            valid_df = valid_df.cast_column("autotrain_label", Sequence(ClassLabel(names=label_names)))

        if self.local:
            dataset = DatasetDict(
                {
                    "train": train_df,
                    "validation": valid_df,
                }
            )
            dataset.save_to_disk(f"{self.project_name}/autotrain-data")
        else:
            train_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="train",
                private=True,
                token=self.token,
            )
            valid_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="validation",
                private=True,
                token=self.token,
            )

        if self.local:
            return f"{self.project_name}/autotrain-data"
        return f"{self.username}/autotrain-data-{self.project_name}"


@dataclass
class LLMPreprocessor:
    """
    A class used to preprocess data for large language model (LLM) training.

    Attributes
    ----------
    train_data : pd.DataFrame
        The training data.
    username : str
        The username for the Hugging Face Hub.
    project_name : str
        The name of the project.
    token : str
        The token for authentication.
    valid_data : Optional[pd.DataFrame], optional
        The validation data, by default None.
    test_size : Optional[float], optional
        The size of the test split, by default 0.2.
    seed : Optional[int], optional
        The random seed, by default 42.
    text_column : Optional[str], optional
        The name of the text column, by default None.
    prompt_column : Optional[str], optional
        The name of the prompt column, by default None.
    rejected_text_column : Optional[str], optional
        The name of the rejected text column, by default None.
    local : Optional[bool], optional
        Whether to save the dataset locally, by default False.

    Methods
    -------
    __post_init__()
        Validates the provided columns and checks for reserved column names.
    split()
        Splits the data into training and validation sets.
    prepare_columns(train_df, valid_df)
        Prepares the columns for training and validation datasets.
    prepare()
        Prepares the datasets and pushes them to the Hugging Face Hub or saves them locally.
    """

    train_data: pd.DataFrame
    username: str
    project_name: str
    token: str
    valid_data: Optional[pd.DataFrame] = None
    test_size: Optional[float] = 0.2
    seed: Optional[int] = 42
    text_column: Optional[str] = None
    prompt_column: Optional[str] = None
    rejected_text_column: Optional[str] = None
    local: Optional[bool] = False

    def __post_init__(self):
        if self.text_column is None:
            raise ValueError("text_column must be provided")

        # check if text_column and rejected_text_column are in train_data
        if self.prompt_column is not None and self.prompt_column not in self.train_data.columns:
            self.prompt_column = None
        if self.rejected_text_column is not None and self.rejected_text_column not in self.train_data.columns:
            self.rejected_text_column = None

        # make sure no reserved columns are in train_data or valid_data
        for column in RESERVED_COLUMNS + LLM_RESERVED_COLUMNS:
            if column in self.train_data.columns:
                raise ValueError(f"{column} is a reserved column name")
            if self.valid_data is not None:
                if column in self.valid_data.columns:
                    raise ValueError(f"{column} is a reserved column name")

    def split(self):
        if self.valid_data is not None:
            return self.train_data, self.valid_data
        # no validation is done in llm training if validation data is not provided
        return self.train_data, self.train_data
        # else:
        #     train_df, valid_df = train_test_split(
        #         self.train_data,
        #         test_size=self.test_size,
        #         random_state=self.seed,
        #     )
        #     train_df = train_df.reset_index(drop=True)
        #     valid_df = valid_df.reset_index(drop=True)
        #     return train_df, valid_df

    def prepare_columns(self, train_df, valid_df):
        drop_cols = [self.text_column]
        train_df.loc[:, "autotrain_text"] = train_df[self.text_column]
        valid_df.loc[:, "autotrain_text"] = valid_df[self.text_column]
        if self.prompt_column is not None:
            drop_cols.append(self.prompt_column)
            train_df.loc[:, "autotrain_prompt"] = train_df[self.prompt_column]
            valid_df.loc[:, "autotrain_prompt"] = valid_df[self.prompt_column]
        if self.rejected_text_column is not None:
            drop_cols.append(self.rejected_text_column)
            train_df.loc[:, "autotrain_rejected_text"] = train_df[self.rejected_text_column]
            valid_df.loc[:, "autotrain_rejected_text"] = valid_df[self.rejected_text_column]

        # drop drop_cols
        train_df = train_df.drop(columns=drop_cols)
        valid_df = valid_df.drop(columns=drop_cols)
        return train_df, valid_df

    def prepare(self):
        train_df, valid_df = self.split()
        train_df, valid_df = self.prepare_columns(train_df, valid_df)
        train_df = Dataset.from_pandas(train_df)
        valid_df = Dataset.from_pandas(valid_df)
        if self.local:
            dataset = DatasetDict(
                {
                    "train": train_df,
                    "validation": valid_df,
                }
            )
            dataset.save_to_disk(f"{self.project_name}/autotrain-data")
        else:
            train_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="train",
                private=True,
                token=self.token,
            )
            valid_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="validation",
                private=True,
                token=self.token,
            )
        if self.local:
            return f"{self.project_name}/autotrain-data"
        return f"{self.username}/autotrain-data-{self.project_name}"


@dataclass
class Seq2SeqPreprocessor:
    """
    Seq2SeqPreprocessor is a class for preprocessing sequence-to-sequence training data.

    Attributes:
        train_data (pd.DataFrame): The training data.
        text_column (str): The name of the column containing the input text.
        label_column (str): The name of the column containing the labels.
        username (str): The username for pushing data to the hub.
        project_name (str): The name of the project.
        token (str): The token for authentication.
        valid_data (Optional[pd.DataFrame]): The validation data. Default is None.
        test_size (Optional[float]): The proportion of the dataset to include in the validation split. Default is 0.2.
        seed (Optional[int]): The random seed for splitting the data. Default is 42.
        local (Optional[bool]): Whether to save the dataset locally or push to the hub. Default is False.

    Methods:
        __post_init__(): Validates the presence of required columns in the training and validation data.
        split(): Splits the training data into training and validation sets if validation data is not provided.
        prepare_columns(train_df, valid_df): Prepares the columns for training and validation data.
        prepare(): Prepares the dataset for training by splitting, preparing columns, and converting to Dataset objects.
    """

    train_data: pd.DataFrame
    text_column: str
    label_column: str
    username: str
    project_name: str
    token: str
    valid_data: Optional[pd.DataFrame] = None
    test_size: Optional[float] = 0.2
    seed: Optional[int] = 42
    local: Optional[bool] = False

    def __post_init__(self):
        # check if text_column and label_column are in train_data
        if self.text_column not in self.train_data.columns:
            raise ValueError(f"{self.text_column} not in train data")
        if self.label_column not in self.train_data.columns:
            raise ValueError(f"{self.label_column} not in train data")
        # check if text_column and label_column are in valid_data
        if self.valid_data is not None:
            if self.text_column not in self.valid_data.columns:
                raise ValueError(f"{self.text_column} not in valid data")
            if self.label_column not in self.valid_data.columns:
                raise ValueError(f"{self.label_column} not in valid data")

        # make sure no reserved columns are in train_data or valid_data
        for column in RESERVED_COLUMNS:
            if column in self.train_data.columns:
                raise ValueError(f"{column} is a reserved column name")
            if self.valid_data is not None:
                if column in self.valid_data.columns:
                    raise ValueError(f"{column} is a reserved column name")

    def split(self):
        if self.valid_data is not None:
            return self.train_data, self.valid_data
        else:
            train_df, valid_df = train_test_split(
                self.train_data,
                test_size=self.test_size,
                random_state=self.seed,
            )
            train_df = train_df.reset_index(drop=True)
            valid_df = valid_df.reset_index(drop=True)
            return train_df, valid_df

    def prepare_columns(self, train_df, valid_df):
        train_df.loc[:, "autotrain_text"] = train_df[self.text_column]
        train_df.loc[:, "autotrain_label"] = train_df[self.label_column]
        valid_df.loc[:, "autotrain_text"] = valid_df[self.text_column]
        valid_df.loc[:, "autotrain_label"] = valid_df[self.label_column]

        # drop text_column and label_column
        train_df = train_df.drop(columns=[self.text_column, self.label_column])
        valid_df = valid_df.drop(columns=[self.text_column, self.label_column])
        return train_df, valid_df

    def prepare(self):
        train_df, valid_df = self.split()
        train_df, valid_df = self.prepare_columns(train_df, valid_df)

        train_df = Dataset.from_pandas(train_df)
        valid_df = Dataset.from_pandas(valid_df)

        if self.local:
            dataset = DatasetDict(
                {
                    "train": train_df,
                    "validation": valid_df,
                }
            )
            dataset.save_to_disk(f"{self.project_name}/autotrain-data")
        else:
            train_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="train",
                private=True,
                token=self.token,
            )
            valid_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="validation",
                private=True,
                token=self.token,
            )
        if self.local:
            return f"{self.project_name}/autotrain-data"
        return f"{self.username}/autotrain-data-{self.project_name}"


@dataclass
class SentenceTransformersPreprocessor:
    """
    A preprocessor class for preparing datasets for sentence transformers.

    Attributes:
        train_data (pd.DataFrame): The training data.
        username (str): The username for the Hugging Face Hub.
        project_name (str): The project name for the Hugging Face Hub.
        token (str): The token for authentication with the Hugging Face Hub.
        valid_data (Optional[pd.DataFrame]): The validation data. Default is None.
        test_size (Optional[float]): The proportion of the dataset to include in the validation split. Default is 0.2.
        seed (Optional[int]): The random seed for splitting the data. Default is 42.
        local (Optional[bool]): Whether to save the dataset locally or push to the Hugging Face Hub. Default is False.
        sentence1_column (Optional[str]): The name of the first sentence column. Default is "sentence1".
        sentence2_column (Optional[str]): The name of the second sentence column. Default is "sentence2".
        sentence3_column (Optional[str]): The name of the third sentence column. Default is "sentence3".
        target_column (Optional[str]): The name of the target column. Default is "target".
        convert_to_class_label (Optional[bool]): Whether to convert the target column to class labels. Default is False.

    Methods:
        __post_init__(): Ensures no reserved columns are in train_data or valid_data.
        split(): Splits the train_data into training and validation sets if valid_data is not provided.
        prepare_columns(train_df, valid_df): Prepares the columns for training and validation datasets.
        prepare(): Prepares the datasets and either saves them locally or pushes them to the Hugging Face Hub.
    """

    train_data: pd.DataFrame
    username: str
    project_name: str
    token: str
    valid_data: Optional[pd.DataFrame] = None
    test_size: Optional[float] = 0.2
    seed: Optional[int] = 42
    local: Optional[bool] = False
    sentence1_column: Optional[str] = "sentence1"
    sentence2_column: Optional[str] = "sentence2"
    sentence3_column: Optional[str] = "sentence3"
    target_column: Optional[str] = "target"
    convert_to_class_label: Optional[bool] = False

    def __post_init__(self):
        # make sure no reserved columns are in train_data or valid_data
        for column in RESERVED_COLUMNS + LLM_RESERVED_COLUMNS:
            if column in self.train_data.columns:
                raise ValueError(f"{column} is a reserved column name")
            if self.valid_data is not None:
                if column in self.valid_data.columns:
                    raise ValueError(f"{column} is a reserved column name")

    def split(self):
        if self.valid_data is not None:
            return self.train_data, self.valid_data
        else:
            train_df, valid_df = train_test_split(
                self.train_data,
                test_size=self.test_size,
                random_state=self.seed,
            )
            train_df = train_df.reset_index(drop=True)
            valid_df = valid_df.reset_index(drop=True)
            return train_df, valid_df

    def prepare_columns(self, train_df, valid_df):
        train_df.loc[:, "autotrain_sentence1"] = train_df[self.sentence1_column]
        train_df.loc[:, "autotrain_sentence2"] = train_df[self.sentence2_column]
        valid_df.loc[:, "autotrain_sentence1"] = valid_df[self.sentence1_column]
        valid_df.loc[:, "autotrain_sentence2"] = valid_df[self.sentence2_column]
        keep_cols = ["autotrain_sentence1", "autotrain_sentence2"]

        if self.sentence3_column is not None:
            train_df.loc[:, "autotrain_sentence3"] = train_df[self.sentence3_column]
            valid_df.loc[:, "autotrain_sentence3"] = valid_df[self.sentence3_column]
            keep_cols.append("autotrain_sentence3")

        if self.target_column is not None:
            train_df.loc[:, "autotrain_target"] = train_df[self.target_column]
            valid_df.loc[:, "autotrain_target"] = valid_df[self.target_column]
            keep_cols.append("autotrain_target")

        train_df = train_df[keep_cols]
        valid_df = valid_df[keep_cols]

        return train_df, valid_df

    def prepare(self):
        train_df, valid_df = self.split()
        train_df, valid_df = self.prepare_columns(train_df, valid_df)

        if self.convert_to_class_label:
            label_names = sorted(set(train_df["autotrain_target"].unique().tolist()))

        train_df = Dataset.from_pandas(train_df)
        valid_df = Dataset.from_pandas(valid_df)

        if self.convert_to_class_label:
            train_df = train_df.cast_column("autotrain_target", ClassLabel(names=label_names))
            valid_df = valid_df.cast_column("autotrain_target", ClassLabel(names=label_names))

        if self.local:
            dataset = DatasetDict(
                {
                    "train": train_df,
                    "validation": valid_df,
                }
            )
            dataset.save_to_disk(f"{self.project_name}/autotrain-data")
        else:
            train_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="train",
                private=True,
                token=self.token,
            )
            valid_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="validation",
                private=True,
                token=self.token,
            )
        if self.local:
            return f"{self.project_name}/autotrain-data"
        return f"{self.username}/autotrain-data-{self.project_name}"


@dataclass
class TextExtractiveQuestionAnsweringPreprocessor:
    """
    Preprocessor for text extractive question answering tasks.

    Attributes:
        train_data (pd.DataFrame): The training data.
        text_column (str): The name of the text column in the data.
        question_column (str): The name of the question column in the data.
        answer_column (str): The name of the answer column in the data.
        username (str): The username for the Hugging Face Hub.
        project_name (str): The project name for the Hugging Face Hub.
        token (str): The token for authentication with the Hugging Face Hub.
        valid_data (Optional[pd.DataFrame]): The validation data. Default is None.
        test_size (Optional[float]): The proportion of the dataset to include in the validation split. Default is 0.2.
        seed (Optional[int]): The random seed for splitting the data. Default is 42.
        local (Optional[bool]): Whether to save the dataset locally or push to the Hugging Face Hub. Default is False.

    Methods:
        __post_init__(): Validates the columns in the training and validation data and converts the answer column to a dictionary.
        split(): Splits the training data into training and validation sets if validation data is not provided.
        prepare_columns(train_df, valid_df): Prepares the columns for training and validation data.
        prepare(): Prepares the dataset for training by splitting, preparing columns, and converting to Hugging Face Dataset format.
    """

    train_data: pd.DataFrame
    text_column: str
    question_column: str
    answer_column: str
    username: str
    project_name: str
    token: str
    valid_data: Optional[pd.DataFrame] = None
    test_size: Optional[float] = 0.2
    seed: Optional[int] = 42
    local: Optional[bool] = False

    def __post_init__(self):
        # check if text_column, question_column, and answer_column are in train_data
        if self.text_column not in self.train_data.columns:
            raise ValueError(f"{self.text_column} not in train data")
        if self.question_column not in self.train_data.columns:
            raise ValueError(f"{self.question_column} not in train data")
        if self.answer_column not in self.train_data.columns:
            raise ValueError(f"{self.answer_column} not in train data")
        # check if text_column, question_column, and answer_column are in valid_data
        if self.valid_data is not None:
            if self.text_column not in self.valid_data.columns:
                raise ValueError(f"{self.text_column} not in valid data")
            if self.question_column not in self.valid_data.columns:
                raise ValueError(f"{self.question_column} not in valid data")
            if self.answer_column not in self.valid_data.columns:
                raise ValueError(f"{self.answer_column} not in valid data")

        # make sure no reserved columns are in train_data or valid_data
        for column in RESERVED_COLUMNS:
            if column in self.train_data.columns:
                raise ValueError(f"{column} is a reserved column name")
            if self.valid_data is not None:
                if column in self.valid_data.columns:
                    raise ValueError(f"{column} is a reserved column name")

        # convert answer_column to dict
        try:
            self.train_data.loc[:, self.answer_column] = self.train_data[self.answer_column].apply(
                lambda x: ast.literal_eval(x)
            )
        except ValueError:
            logger.warning("Unable to do ast.literal_eval on train_data[answer_column]")
            logger.warning("assuming answer_column is already a dict")

        if self.valid_data is not None:
            try:
                self.valid_data.loc[:, self.answer_column] = self.valid_data[self.answer_column].apply(
                    lambda x: ast.literal_eval(x)
                )
            except ValueError:
                logger.warning("Unable to do ast.literal_eval on valid_data[answer_column]")
                logger.warning("assuming answer_column is already a dict")

    def split(self):
        if self.valid_data is not None:
            return self.train_data, self.valid_data
        else:
            train_df, valid_df = train_test_split(
                self.train_data,
                test_size=self.test_size,
                random_state=self.seed,
            )
            train_df = train_df.reset_index(drop=True)
            valid_df = valid_df.reset_index(drop=True)
            return train_df, valid_df

    def prepare_columns(self, train_df, valid_df):
        train_df.loc[:, "autotrain_text"] = train_df[self.text_column]
        train_df.loc[:, "autotrain_question"] = train_df[self.question_column]
        train_df.loc[:, "autotrain_answer"] = train_df[self.answer_column]
        valid_df.loc[:, "autotrain_text"] = valid_df[self.text_column]
        valid_df.loc[:, "autotrain_question"] = valid_df[self.question_column]
        valid_df.loc[:, "autotrain_answer"] = valid_df[self.answer_column]

        # drop all other columns
        train_df = train_df.drop(
            columns=[
                x for x in train_df.columns if x not in ["autotrain_text", "autotrain_question", "autotrain_answer"]
            ]
        )
        valid_df = valid_df.drop(
            columns=[
                x for x in valid_df.columns if x not in ["autotrain_text", "autotrain_question", "autotrain_answer"]
            ]
        )
        return train_df, valid_df

    def prepare(self):
        train_df, valid_df = self.split()
        train_df, valid_df = self.prepare_columns(train_df, valid_df)

        train_df = Dataset.from_pandas(train_df)
        valid_df = Dataset.from_pandas(valid_df)

        if self.local:
            dataset = DatasetDict(
                {
                    "train": train_df,
                    "validation": valid_df,
                }
            )
            dataset.save_to_disk(f"{self.project_name}/autotrain-data")
        else:
            train_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="train",
                private=True,
                token=self.token,
            )
            valid_df.push_to_hub(
                f"{self.username}/autotrain-data-{self.project_name}",
                split="validation",
                private=True,
                token=self.token,
            )
        if self.local:
            return f"{self.project_name}/autotrain-data"
        return f"{self.username}/autotrain-data-{self.project_name}"