File size: 56,608 Bytes
6dd9839
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
import torch
import torch.nn as nn
import math
from typing import List, Optional, Tuple, Union
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss, SiLU
from transformers.modeling_outputs import (
    BaseModelOutputWithPastAndCrossAttentions,
    MaskedLMOutput,
)
from transformers.modeling_utils import PreTrainedModel
from transformers import PretrainedConfig
from dataclasses import dataclass
from typing import Any, Dict

import torch.distributions as dists
from torch.nn import functional as F
from transformers.generation.configuration_utils import GenerationConfig
from transformers.utils import ModelOutput

try:
    from tqdm import trange
except ImportError:
    def trange(n, **kwargs):
        return range(n)

def rotate_half(x):
    x1, x2 = x.chunk(2, dim=-1)
    return torch.cat((-x2, x1), dim=-1)


def apply_rotary_pos_emb(x, cos, sin):
    cos = cos[:, :, : x.shape[-2], :]
    sin = sin[:, :, : x.shape[-2], :]

    return (x * cos) + (rotate_half(x) * sin)


def gelu(x):
    """
    This is the gelu implementation from the original ESM repo. Using F.gelu yields subtly wrong results.
    """
    return x * 0.5 * (1.0 + torch.erf(x / math.sqrt(2.0)))


class RotaryEmbedding(torch.nn.Module):
    """
    Rotary position embeddings based on those in
    [RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer). Query and keys are transformed by rotation
    matrices which depend on their relative positions.
    """

    def __init__(self, dim: int):
        super().__init__()
        # Generate and save the inverse frequency buffer (non trainable)
        inv_freq = 1.0 / (10000 ** (torch.arange(0, dim, 2).float() / dim))
        inv_freq = inv_freq
        self.register_buffer("inv_freq", inv_freq)

        self._seq_len_cached = None
        self._cos_cached = None
        self._sin_cached = None

    def _update_cos_sin_tables(self, x, seq_dimension=2):
        seq_len = x.shape[seq_dimension]

        # Reset the tables if the sequence length has changed,
        # or if we're on a new device (possibly due to tracing for instance)
        if seq_len != self._seq_len_cached or self._cos_cached.device != x.device:
            self._seq_len_cached = seq_len
            t = torch.arange(x.shape[seq_dimension], device=x.device).type_as(
                self.inv_freq
            )
            freqs = torch.outer(t, self.inv_freq)
            emb = torch.cat((freqs, freqs), dim=-1).to(x.device)

            self._cos_cached = emb.cos()[None, None, :, :]
            self._sin_cached = emb.sin()[None, None, :, :]

        return self._cos_cached, self._sin_cached

    def forward(
        self, q: torch.Tensor, k: torch.Tensor
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        self._cos_cached, self._sin_cached = self._update_cos_sin_tables(
            k, seq_dimension=-2
        )

        return (
            apply_rotary_pos_emb(q, self._cos_cached, self._sin_cached),
            apply_rotary_pos_emb(k, self._cos_cached, self._sin_cached),
        )


def create_position_ids_from_input_ids(
    input_ids, padding_idx, past_key_values_length=0
):
    """
    Replace non-padding symbols with their position numbers. Position numbers begin at padding_idx+1. Padding symbols
    are ignored. This is modified from fairseq's `utils.make_positions`.

    Args:
        x: torch.Tensor x:

    Returns: torch.Tensor
    """
    # The series of casts and type-conversions here are carefully balanced to both work with ONNX export and XLA.
    mask = input_ids.ne(padding_idx).int()
    incremental_indices = (
        torch.cumsum(mask, dim=1).type_as(mask) + past_key_values_length
    ) * mask
    return incremental_indices.long() + padding_idx


class EsmEmbeddings(nn.Module):
    """
    Same as BertEmbeddings with a tiny tweak for positional embeddings indexing.
    """

    def __init__(self, config):
        super().__init__()
        self.word_embeddings = nn.Embedding(
            config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id
        )

        if config.emb_layer_norm_before:
            self.layer_norm = nn.LayerNorm(
                config.hidden_size, eps=config.layer_norm_eps
            )
        else:
            self.layer_norm = None
        self.dropout = nn.Dropout(config.hidden_dropout_prob)
        # position_ids (1, len position emb) is contiguous in memory and exported when serialized
        self.position_embedding_type = getattr(
            config, "position_embedding_type", "absolute"
        )
        self.register_buffer(
            "position_ids",
            torch.arange(config.max_position_embeddings).expand((1, -1)),
            persistent=False,
        )

        self.padding_idx = config.pad_token_id
        self.position_embeddings = nn.Embedding(
            config.max_position_embeddings,
            config.hidden_size,
            padding_idx=self.padding_idx,
        )
        self.token_dropout = config.token_dropout
        self.mask_token_id = config.mask_token_id

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        position_ids=None,
        inputs_embeds=None,
        past_key_values_length=0,
    ):
        if position_ids is None:
            if input_ids is not None:
                # Create the position ids from the input token ids. Any padded tokens remain padded.
                position_ids = create_position_ids_from_input_ids(
                    input_ids, self.padding_idx, past_key_values_length
                )
            else:
                position_ids = self.create_position_ids_from_inputs_embeds(
                    inputs_embeds
                )

        if inputs_embeds is None:
            inputs_embeds = self.word_embeddings(input_ids)

        # Note that if we want to support ESM-1 (not 1b!) in future then we need to support an
        # embedding_scale factor here.
        embeddings = inputs_embeds

        # Matt: ESM has the option to handle masking in MLM in a slightly unusual way. If the token_dropout
        # flag is False then it is handled in the same was as BERT/RoBERTa. If it is set to True, however,
        # masked tokens are treated as if they were selected for input dropout and zeroed out.
        # This "mask-dropout" is compensated for when masked tokens are not present, by scaling embeddings by
        # a factor of (fraction of unmasked tokens during training) / (fraction of unmasked tokens in sample).
        # This is analogous to the way that dropout layers scale down outputs during evaluation when not
        # actually dropping out values (or, equivalently, scale up their un-dropped outputs in training).
        if self.token_dropout:
            embeddings.masked_fill_(
                (input_ids == self.mask_token_id).unsqueeze(-1), 0.0
            )
            mask_ratio_train = (
                0.15 * 0.8
            )  # Hardcoded as the ratio used in all ESM model training runs
            src_lengths = attention_mask.sum(-1)
            mask_ratio_observed = (input_ids == self.mask_token_id).sum(
                -1
            ).float() / src_lengths
            embeddings = (
                embeddings
                * (1 - mask_ratio_train)
                / (1 - mask_ratio_observed)[:, None, None]
            ).to(embeddings.dtype)

        if self.position_embedding_type == "absolute":
            position_embeddings = self.position_embeddings(position_ids)
            embeddings += position_embeddings

        if self.layer_norm is not None:
            embeddings = self.layer_norm(embeddings)
        if attention_mask is not None:
            embeddings = (embeddings * attention_mask.unsqueeze(-1)).to(
                embeddings.dtype
            )
        # Matt: I think this line was copied incorrectly from BERT, disabling it for now.
        # embeddings = self.dropout(embeddings)
        return embeddings

    def create_position_ids_from_inputs_embeds(self, inputs_embeds):
        """
        We are provided embeddings directly. We cannot infer which are padded so just generate sequential position ids.

        Args:
            inputs_embeds: torch.Tensor

        Returns: torch.Tensor
        """
        input_shape = inputs_embeds.size()[:-1]
        sequence_length = input_shape[1]

        position_ids = torch.arange(
            self.padding_idx + 1,
            sequence_length + self.padding_idx + 1,
            dtype=torch.long,
            device=inputs_embeds.device,
        )
        return position_ids.unsqueeze(0).expand(input_shape)


class EsmSelfAttention(nn.Module):
    def __init__(self, config, position_embedding_type=None):
        super().__init__()
        if config.hidden_size % config.num_attention_heads != 0 and not hasattr(
            config, "embedding_size"
        ):
            raise ValueError(
                f"The hidden size ({config.hidden_size}) is not a multiple of the number of attention "
                f"heads ({config.num_attention_heads})"
            )

        self.num_attention_heads = config.num_attention_heads
        self.attention_head_size = int(config.hidden_size / config.num_attention_heads)
        self.all_head_size = self.num_attention_heads * self.attention_head_size

        self.query = nn.Linear(config.hidden_size, self.all_head_size)
        self.key = nn.Linear(config.hidden_size, self.all_head_size)
        self.value = nn.Linear(config.hidden_size, self.all_head_size)

        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
        self.position_embedding_type = position_embedding_type or getattr(
            config, "position_embedding_type", "absolute"
        )
        self.rotary_embeddings = None
        if (
            self.position_embedding_type == "relative_key"
            or self.position_embedding_type == "relative_key_query"
        ):
            self.max_position_embeddings = config.max_position_embeddings
            self.distance_embedding = nn.Embedding(
                2 * config.max_position_embeddings - 1, self.attention_head_size
            )
        elif self.position_embedding_type == "rotary":
            self.rotary_embeddings = RotaryEmbedding(dim=self.attention_head_size)

    def transpose_for_scores(self, x: torch.Tensor) -> torch.Tensor:
        new_x_shape = x.size()[:-1] + (
            self.num_attention_heads,
            self.attention_head_size,
        )
        x = x.view(new_x_shape)
        return x.permute(0, 2, 1, 3)

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask: Optional[torch.FloatTensor] = None,
        head_mask: Optional[torch.FloatTensor] = None,
        past_key_value: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
        output_attentions: Optional[bool] = False,
    ) -> Tuple[torch.Tensor]:
        mixed_query_layer = self.query(hidden_states)

        # If this is instantiated as a cross-attention module, the keys
        # and values come from an encoder; the attention mask needs to be
        # such that the encoder's padding tokens are not attended to.

        if past_key_value is not None:
            key_layer = self.transpose_for_scores(self.key(hidden_states))
            value_layer = self.transpose_for_scores(self.value(hidden_states))
            key_layer = torch.cat([past_key_value[0], key_layer], dim=2)
            value_layer = torch.cat([past_key_value[1], value_layer], dim=2)
        else:
            key_layer = self.transpose_for_scores(self.key(hidden_states))
            value_layer = self.transpose_for_scores(self.value(hidden_states))

        query_layer = self.transpose_for_scores(mixed_query_layer)

        # Matt: Our BERT model (which this code was derived from) scales attention logits down by sqrt(head_dim).
        # ESM scales the query down by the same factor instead. Modulo numerical stability these are equivalent,
        # but not when rotary embeddings get involved. Therefore, we scale the query here to match the original
        # ESM code and fix rotary embeddings.
        query_layer = query_layer * self.attention_head_size**-0.5

        if self.position_embedding_type == "rotary":
            query_layer, key_layer = self.rotary_embeddings(query_layer, key_layer)

        # Take the dot product between "query" and "key" to get the raw attention scores.
        attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2))

        if (
            self.position_embedding_type == "relative_key"
            or self.position_embedding_type == "relative_key_query"
        ):
            seq_length = hidden_states.size()[1]
            position_ids_l = torch.arange(
                seq_length, dtype=torch.long, device=hidden_states.device
            ).view(-1, 1)
            position_ids_r = torch.arange(
                seq_length, dtype=torch.long, device=hidden_states.device
            ).view(1, -1)
            distance = position_ids_l - position_ids_r
            positional_embedding = self.distance_embedding(
                distance + self.max_position_embeddings - 1
            )
            positional_embedding = positional_embedding.to(
                dtype=query_layer.dtype
            )  # fp16 compatibility

            if self.position_embedding_type == "relative_key":
                relative_position_scores = torch.einsum(
                    "bhld,lrd->bhlr", query_layer, positional_embedding
                )
                attention_scores = attention_scores + relative_position_scores
            elif self.position_embedding_type == "relative_key_query":
                relative_position_scores_query = torch.einsum(
                    "bhld,lrd->bhlr", query_layer, positional_embedding
                )
                relative_position_scores_key = torch.einsum(
                    "bhrd,lrd->bhlr", key_layer, positional_embedding
                )
                attention_scores = (
                    attention_scores
                    + relative_position_scores_query
                    + relative_position_scores_key
                )

        if attention_mask is not None:
            # Apply the attention mask is (precomputed for all layers in EsmModel forward() function)
            attention_scores = attention_scores + attention_mask

        # Normalize the attention scores to probabilities.
        attention_probs = nn.functional.softmax(attention_scores, dim=-1)

        # This is actually dropping out entire tokens to attend to, which might
        # seem a bit unusual, but is taken from the original Transformer paper.
        attention_probs = self.dropout(attention_probs)

        # Mask heads if we want to
        if head_mask is not None:
            attention_probs = attention_probs * head_mask

        context_layer = torch.matmul(attention_probs, value_layer)

        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
        new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,)
        context_layer = context_layer.view(new_context_layer_shape)

        outputs = (
            (context_layer, attention_probs) if output_attentions else (context_layer,)
        )

        return outputs


class EsmSelfOutput(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
        self.dropout = nn.Dropout(config.hidden_dropout_prob)

    def forward(self, hidden_states, input_tensor):
        hidden_states = self.dense(hidden_states)
        hidden_states = self.dropout(hidden_states)
        hidden_states += input_tensor
        return hidden_states


class EsmAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.self = EsmSelfAttention(config)
        self.output = EsmSelfOutput(config)
        self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)

    def forward(
        self,
        hidden_states,
        attention_mask=None,
        head_mask=None,
        past_key_value=None,
        output_attentions=False,
    ):
        hidden_states_ln = self.LayerNorm(hidden_states)
        self_outputs = self.self(
            hidden_states_ln,
            attention_mask,
            head_mask,
            past_key_value,
            output_attentions,
        )
        attention_output = self.output(self_outputs[0], hidden_states)
        outputs = (attention_output,) + self_outputs[
            1:
        ]  # add attentions if we output them
        return outputs


class AdaLNEsmAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.self = EsmSelfAttention(config)
        self.output = EsmSelfOutput(config)
        self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
          # 用 protein_cond 生成 attention 前 AdaLN 的 scale / shift
        self.adaln_modulation = nn.Linear(config.hidden_size, 2 *config.hidden_size)

    def forward(
        self,
        hidden_states,
        protein_cond=None,
        attention_mask=None,
        head_mask=None,
        past_key_value=None,
        output_attentions=False,
    ):
        # protein_cond 应该是一个 每个样本一个向量,形状大概是 [batch,hidden_size].
        # AdaLN 用的是 pooled protein condition,不是整条 protein token序列。
        if protein_cond is None:
              raise ValueError("protein_cond is required when AdaLN is enabled.")
        hidden_states_ln = self.LayerNorm(hidden_states)

        scale, shift = self.adaln_modulation(protein_cond).chunk(2,dim=-1)
        scale = scale.unsqueeze(1)
        shift = shift.unsqueeze(1)
        hidden_states_ln = hidden_states_ln * (1.0 + scale) + shift

        self_outputs = self.self(
            hidden_states_ln,
            attention_mask,
            head_mask,
            past_key_value,
            output_attentions,
        )
        attention_output = self.output(self_outputs[0], hidden_states)
        outputs = (attention_output,) + self_outputs[1:]
        return outputs


class ProteinConditioningAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.num_attention_heads = config.num_attention_heads
        self.attention_head_size = int(config.hidden_size /config.num_attention_heads)
        self.all_head_size = self.num_attention_heads *self.attention_head_size

        self.query = nn.Linear(config.hidden_size, self.all_head_size)
        self.key = nn.Linear(config.hidden_size, self.all_head_size)
        self.value = nn.Linear(config.hidden_size, self.all_head_size)
        self.out_proj = nn.Linear(config.hidden_size, config.hidden_size)

        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
        self.rna_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
        self.protein_norm = nn.LayerNorm(config.hidden_size,eps=config.layer_norm_eps)

        self.gate_proj = nn.Linear(config.hidden_size,config.hidden_size)

    def transpose_for_scores(self, x):
        new_x_shape = x.size()[:-1] + (
            self.num_attention_heads,
            self.attention_head_size,
        )
        x = x.view(new_x_shape)
        return x.permute(0, 2, 1, 3)

    def forward(
        self,
        rna_hidden_states,
        protein_hidden_states,
        protein_attention_mask=None,
        output_attentions=False,
    ):
        rna_hidden_states_ln = self.rna_norm(rna_hidden_states)
        protein_hidden_states_ln = self.protein_norm(protein_hidden_states)

        query_layer = self.transpose_for_scores(self.query(rna_hidden_states_ln))
        key_layer =self.transpose_for_scores(self.key(protein_hidden_states_ln))
        value_layer =self.transpose_for_scores(self.value(protein_hidden_states_ln))

        query_layer = query_layer * self.attention_head_size**-0.5
        attention_scores = torch.matmul(query_layer,
        key_layer.transpose(-1, -2))

        if protein_attention_mask is not None:
            protein_attention_mask = protein_attention_mask[:, None,None, :].to(dtype=attention_scores.dtype,
                device=attention_scores.device)
            attention_scores = attention_scores + (
                1.0 - protein_attention_mask
            ) * torch.finfo(attention_scores.dtype).min

        attention_probs = nn.functional.softmax(attention_scores, dim=-1)
        attention_probs = self.dropout(attention_probs)

        context_layer = torch.matmul(attention_probs, value_layer)
        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
        new_context_layer_shape = context_layer.size()[:-2] +(self.all_head_size,)
        context_layer = context_layer.view(new_context_layer_shape)

        protein_update = self.out_proj(context_layer)

        gate = torch.sigmoid(self.gate_proj(rna_hidden_states_ln))
        conditioned_hidden_states = rna_hidden_states + gate * protein_update

        outputs = (
            (conditioned_hidden_states, attention_probs)
            if output_attentions
            else (conditioned_hidden_states,)
        )
        return outputs
    

class EsmIntermediate(nn.Module):
    def __init__(self, config):
        super().__init__()

        self.dense = nn.Linear(
            config.hidden_size,
            int(config.intermediate_size * 2),
            bias=config.add_bias_fnn,
        )
        self.activation_fn = SiLU()

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        hidden_states = self.dense(hidden_states)

        # GLU
        x1, x2 = hidden_states.split(int(hidden_states.size(-1) / 2), -1)
        hidden_states = self.activation_fn(x1) * x2

        return hidden_states


class EsmOutput(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.dense = nn.Linear(
            config.intermediate_size, config.hidden_size, bias=config.add_bias_fnn
        )
        self.dropout = nn.Dropout(config.hidden_dropout_prob)

    def forward(self, hidden_states, input_tensor):
        hidden_states = self.dense(hidden_states)
        hidden_states = self.dropout(hidden_states)
        hidden_states += input_tensor
        return hidden_states


class EsmLayer(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.chunk_size_feed_forward = config.chunk_size_feed_forward
        self.seq_len_dim = 1
        self.attention = EsmAttention(config)
        self.AdaLN_attention = AdaLNEsmAttention(config)
        self.intermediate = EsmIntermediate(config)
        self.output = EsmOutput(config)
        self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)

        self.use_FiLM = config.use_FiLM
        self.use_AdaLN = config.use_AdaLN
        self.use_gated_bias = config.use_gated_bias
        self.use_protein_conditioning_attention = config.use_protein_conditioning_attention
        protein_dim = getattr(config, "protein_dim", config.hidden_size)
        if protein_dim == config.hidden_size:
            self.protein_proj = nn.Identity()
        else:
            self.protein_proj = nn.Linear(protein_dim, config.hidden_size)

        self.ffn_adaln_modulation = nn.Linear(config.hidden_size, 2 *config.hidden_size)

        if self.use_protein_conditioning_attention:
            self.protein_conditioning_attention = ProteinConditioningAttention(config)

        self.film_modulation = nn.Linear(config.hidden_size, 2 * config.hidden_size)


        self.gated_bias_proj = nn.Linear(config.hidden_size, config.hidden_size)
        self.gated_bias_gate = nn.Linear(config.hidden_size, config.hidden_size)



    def forward(
        self,
        hidden_states,
        protein_cond=None,
        attention_mask=None,
        head_mask=None,
        protein_attention_mask=None,
        past_key_value=None,
        output_attentions=False,
    ):
        
        # protein_cond default shape: [B, Lp, Dp]
        # protein_hidden_states: token-level protein condition, shape [B, Lp, H]
        # pooled_protein_cond: pooled protein condition for AdaLN / FiLM / gated bias, shape [B, H]

        # decoder uni-directional self-attention cached key/values tuple is at positions 1,2
        self_attn_past_key_value = (
            past_key_value[:2] if past_key_value is not None else None
        )

        protein_hidden_states = None
        pooled_protein_cond = None

        if protein_cond is not None:
            protein_hidden_states = self.protein_proj(protein_cond)   # [B, Lp, H]

            if protein_hidden_states.dim() == 3:


                if protein_attention_mask is None:
                    pooled_protein_cond = protein_hidden_states.mean(dim=1)   #[B, H]
                else:
                    # protein_attention_mask: [B, Lp], padding = 0, valid = 1
                    mask = protein_attention_mask.to(protein_hidden_states.dtype).unsqueeze(-1)   #[B, Lp, 1]
                    denom = mask.sum(dim=1).clamp_min(1.0)   # [B, 1]
                    pooled_protein_cond = (protein_hidden_states * mask).sum(dim=1) / denom

            elif protein_hidden_states.dim() == 2:
                pooled_protein_cond = protein_hidden_states
                protein_hidden_states = protein_hidden_states.unsqueeze(1)
            else:
                raise ValueError("protein_cond must have shape [B, Lp, Dp] or [B, Dp].")

        if self.use_AdaLN:
            self_attention_outputs = self.AdaLN_attention(
                hidden_states,
                protein_cond=pooled_protein_cond,
                attention_mask=attention_mask,
                head_mask=head_mask,
                output_attentions=output_attentions,
                past_key_value=self_attn_past_key_value,
            )
            attention_output = self_attention_outputs[0]

        else:
            self_attention_outputs = self.attention(
                hidden_states,
                attention_mask,
                head_mask,
                output_attentions=output_attentions,
                past_key_value=self_attn_past_key_value,
            )
            attention_output = self_attention_outputs[0]

        if self.use_protein_conditioning_attention:
            if protein_hidden_states is None:
                raise ValueError("protein_hidden_states is required when protein conditioning attention is enabled."
                )
            protein_attention_outputs = self.protein_conditioning_attention(
                attention_output,
                protein_hidden_states,
                protein_attention_mask=protein_attention_mask,
                output_attentions=output_attentions,
            )
            attention_output = protein_attention_outputs[0]


        outputs = self_attention_outputs[
            1:
        ]  # add self attentions if we output attention weights

        # gated_bias
        if self.use_gated_bias:
            if pooled_protein_cond is None:
                raise ValueError("pooled_protein_cond is required when gated bias is enabled.")
            bias = self.gated_bias_proj(pooled_protein_cond)
            gate = torch.sigmoid(self.gated_bias_gate(pooled_protein_cond))
            bias = bias.unsqueeze(1)
            gate = gate.unsqueeze(1)
            attention_output = attention_output + gate * bias

        # FiLM
        if self.use_FiLM:
            if pooled_protein_cond is None:
                raise ValueError("pooled_protein_cond is required when FiLM is enabled.")
            gamma, beta = self.film_modulation(pooled_protein_cond).chunk(2,dim=-1)
            gamma = gamma.unsqueeze(1)
            beta = beta.unsqueeze(1)
            attention_output = attention_output * (1.0 + gamma) + beta

        # FFN
        layer_output = self.feed_forward_chunk(attention_output, protein_cond=pooled_protein_cond)

        outputs = (layer_output,) + outputs

        return outputs

    def feed_forward_chunk(self, attention_output, protein_cond=None):
        attention_output_ln = self.LayerNorm(attention_output)

        if self.use_AdaLN:
            if protein_cond is None:
                raise ValueError("protein_cond is required when AdaLN is enabled.")
            scale, shift = self.ffn_adaln_modulation(protein_cond).chunk(2,dim=-1)
            scale = scale.unsqueeze(1)
            shift = shift.unsqueeze(1)
            attention_output_ln = attention_output_ln * (1.0 + scale) + shift

        intermediate_output = self.intermediate(attention_output_ln)
        layer_output = self.output(intermediate_output, attention_output)
        return layer_output


class EsmEncoder(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.config = config
        self.layer = nn.ModuleList(
            [EsmLayer(config) for _ in range(config.num_hidden_layers)]
        )
        self.emb_layer_norm_after = nn.LayerNorm(
            config.hidden_size, eps=config.layer_norm_eps
        )

    def forward(
        self,
        hidden_states,
        protein_cond=None,
        attention_mask=None,
        head_mask=None,
        protein_attention_mask=None,
        past_key_values=None,
        return_dict=True,
    ):
        for i, layer_module in enumerate(self.layer):
            layer_head_mask = head_mask[i] if head_mask is not None else None
            past_key_value = past_key_values[i] if past_key_values is not None else None
            layer_outputs = layer_module(
                hidden_states=hidden_states,
                protein_cond=protein_cond,
                attention_mask=attention_mask,
                head_mask=layer_head_mask,
                protein_attention_mask=protein_attention_mask,
                past_key_value=past_key_value,
                output_attentions=False
            )
            hidden_states = layer_outputs[0]
        if self.emb_layer_norm_after:
            hidden_states = self.emb_layer_norm_after(hidden_states)
        if not return_dict:
            return tuple(
                v
                for v in [
                    hidden_states,
                ]
                if v is not None
            )
        return BaseModelOutputWithPastAndCrossAttentions(
            last_hidden_state=hidden_states,
        )


class EsmConfig(PretrainedConfig):
    r"""
    This is the configuration class to store the configuration of a [`ESMModel`]. It is used to instantiate a ESM model
    according to the specified arguments, defining the model architecture. Instantiating a configuration with the
    defaults will yield a similar configuration to that of the ESM
    [facebook/esm-1b](https://huggingface.co/facebook/esm-1b) architecture.

    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
    documentation from [`PretrainedConfig`] for more information.


    Args:
        vocab_size (`int`, *optional*):
            Vocabulary size of the ESM model. Defines the number of different tokens that can be represented by the
            `inputs_ids` passed when calling [`ESMModel`].
        mask_token_id (`int`, *optional*):
            The index of the mask token in the vocabulary. This must be included in the config because of the
            "mask-dropout" scaling trick, which will scale the inputs depending on the number of masked tokens.
        pad_token_id (`int`, *optional*):
            The index of the padding token in the vocabulary. This must be included in the config because certain parts
            of the ESM code use this instead of the attention mask.
        hidden_size (`int`, *optional*, defaults to 768):
            Dimensionality of the encoder layers and the pooler layer.
        num_hidden_layers (`int`, *optional*, defaults to 12):
            Number of hidden layers in the Transformer encoder.
        num_attention_heads (`int`, *optional*, defaults to 12):
            Number of attention heads for each attention layer in the Transformer encoder.
        intermediate_size (`int`, *optional*, defaults to 3072):
            Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
        hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
        attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
            The dropout ratio for the attention probabilities.
        max_position_embeddings (`int`, *optional*, defaults to 1026):
            The maximum sequence length that this model might ever be used with. Typically set this to something large
            just in case (e.g., 512 or 1024 or 2048).
        initializer_range (`float`, *optional*, defaults to 0.02):
            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
        layer_norm_eps (`float`, *optional*, defaults to 1e-12):
            The epsilon used by the layer normalization layers.
        position_embedding_type (`str`, *optional*, defaults to `"absolute"`):
            Type of position embedding. Choose one of `"absolute"`, `"relative_key"`, `"relative_key_query", "rotary"`.
            For positional embeddings use `"absolute"`. For more information on `"relative_key"`, please refer to
            [Self-Attention with Relative Position Representations (Shaw et al.)](https://arxiv.org/abs/1803.02155).
            For more information on `"relative_key_query"`, please refer to *Method 4* in [Improve Transformer Models
            with Better Relative Position Embeddings (Huang et al.)](https://arxiv.org/abs/2009.13658).
        is_decoder (`bool`, *optional*, defaults to `False`):
            Whether the model is used as a decoder or not. If `False`, the model is used as an encoder.
        use_cache (`bool`, *optional*, defaults to `True`):
            Whether or not the model should return the last key/values attentions (not used by all models). Only
            relevant if `config.is_decoder=True`.
        emb_layer_norm_before (`bool`, *optional*):
            Whether to apply layer normalization after embeddings but before the main stem of the network.
        token_dropout (`bool`, defaults to `False`):
            When this is enabled, masked tokens are treated as if they had been dropped out by input dropout.

    Examples:

    ```python
    >>> from transformers import EsmModel, EsmConfig

    >>> # Initializing a ESM facebook/esm-1b style configuration >>> configuration = EsmConfig()

    >>> # Initializing a model from the configuration >>> model = ESMModel(configuration)

    >>> # Accessing the model configuration >>> configuration = model.config
    ```"""
    model_type = "esm"


class EsmPreTrainedModel(PreTrainedModel):
    """
    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
    models.
    """

    config_class = EsmConfig
    base_model_prefix = "esm"
    _no_split_modules = ["EsmLayer", "EsmFoldTriangularSelfAttentionBlock"]

    # Copied from transformers.models.bert.modeling_bert.BertPreTrainedModel._init_weights
    def _init_weights(self, module):
        """Initialize the weights"""
        if isinstance(module, nn.Linear):
            # Slightly different from the TF version which uses truncated_normal for initialization
            # cf https://github.com/pytorch/pytorch/pull/5617
            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
            if module.bias is not None:
                module.bias.data.zero_()
        elif isinstance(module, nn.Embedding):
            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
            if module.padding_idx is not None:
                module.weight.data[module.padding_idx].zero_()
        elif isinstance(module, nn.LayerNorm):
            module.bias.data.zero_()
            module.weight.data.fill_(1.0)


class EsmModel(EsmPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.config = config
        self.embeddings = EsmEmbeddings(config)
        self.encoder = EsmEncoder(config)
        self.post_init()
        self._reset_conditioning_parameters()

    def _reset_conditioning_parameters(self):
        for layer in self.encoder.layer:
            nn.init.zeros_(layer.AdaLN_attention.adaln_modulation.weight)
            nn.init.zeros_(layer.AdaLN_attention.adaln_modulation.bias)

            nn.init.zeros_(layer.ffn_adaln_modulation.weight)
            nn.init.zeros_(layer.ffn_adaln_modulation.bias)

            nn.init.zeros_(layer.film_modulation.weight)
            nn.init.zeros_(layer.film_modulation.bias)

            nn.init.zeros_(layer.gated_bias_proj.weight)
            nn.init.zeros_(layer.gated_bias_proj.bias)
            nn.init.zeros_(layer.gated_bias_gate.weight)
            nn.init.zeros_(layer.gated_bias_gate.bias)

            if hasattr(layer, "protein_conditioning_attention"):
                nn.init.zeros_(layer.protein_conditioning_attention.out_proj.weight)
                nn.init.zeros_(layer.protein_conditioning_attention.out_proj.bias)

    def forward(
        self,
        protein_cond: Optional[torch.Tensor] = None,
        protein_attention_mask: Optional[torch.Tensor] = None,
        input_ids: Optional[torch.Tensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.Tensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
    ) -> Union[BaseModelOutputWithPastAndCrossAttentions]:
        if input_ids is not None and inputs_embeds is not None:
            raise ValueError(
                "You cannot specify both input_ids and inputs_embeds at the same time"
            )
        elif input_ids is not None:
            input_shape = input_ids.size()
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.size()[:-1]
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        batch_size, seq_length = input_shape
        device = input_ids.device if input_ids is not None else inputs_embeds.device
        if attention_mask is None:
            attention_mask = torch.ones(
                ((batch_size, seq_length)), device=device
            )

        extended_attention_mask: torch.Tensor = self.get_extended_attention_mask(
            attention_mask, input_shape
        )
        head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)

        embedding_output = self.embeddings(
            input_ids=input_ids,
            position_ids=position_ids,
            attention_mask=attention_mask,
            inputs_embeds=inputs_embeds,
        )
        encoder_outputs = self.encoder(
            hidden_states=embedding_output,
            protein_cond=protein_cond,
            attention_mask=extended_attention_mask,
            head_mask=head_mask,
            protein_attention_mask=protein_attention_mask,
        )
        sequence_output = encoder_outputs[0]

        return BaseModelOutputWithPastAndCrossAttentions(
            last_hidden_state=sequence_output,
            hidden_states=encoder_outputs.hidden_states,
            attentions=encoder_outputs.attentions,
            cross_attentions=encoder_outputs.cross_attentions,
        )


class EsmLMHead(nn.Module):
    """ESM Head for masked language modeling."""

    def __init__(self, config):
        super().__init__()
        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
        self.layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)

        self.decoder = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
        self.bias = nn.Parameter(torch.zeros(config.vocab_size))

    def forward(self, features, **kwargs):
        x = self.dense(features)
        x = gelu(x)
        x = self.layer_norm(x)

        # project back to size of vocabulary with bias
        x = self.decoder(x) + self.bias
        return x


class EsmForMaskedLM(EsmPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.esm = EsmModel(config)
        self.lm_head = EsmLMHead(config)
        self.lm_head.apply(self._init_weights)
    def forward(
        self,
        protein_cond: Optional[torch.Tensor] = None,
        protein_attention_mask: Optional[torch.Tensor] = None,
        input_ids: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
    ) -> Union[Tuple, MaskedLMOutput]:
        outputs = self.esm(
            protein_cond=protein_cond,
            protein_attention_mask=protein_attention_mask,
            input_ids=input_ids,
            attention_mask=attention_mask,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
        )
        sequence_output = outputs[0]
        prediction_scores = self.lm_head(sequence_output)
        masked_lm_loss = None
        return MaskedLMOutput(
            loss=masked_lm_loss,
            logits=prediction_scores,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )


# ==================== ProRiboGen: Masked Diffusion Generation ====================
# 以下板块是坏的,现在先不要用。
# 以下板块是坏的,现在先不要用。
# 以下板块是坏的,现在先不要用。
# 重要的事情说三遍!
'''

def _top_p_logits(logits, top_p):
    sorted_logits, sorted_indices = torch.sort(logits, descending=True)
    cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
    sorted_indices_to_remove = cumulative_probs > top_p
    sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
    sorted_indices_to_remove[..., 0] = 0
    mask = torch.zeros_like(logits, dtype=torch.bool, device=logits.device)
    mask = mask.scatter_(-1, sorted_indices, sorted_indices_to_remove)
    logits = logits.masked_fill(mask, torch.finfo(logits.dtype).min)
    return logits


def _top_k_logits(logits, top_k):
    if top_k is None or top_k == 0:
        return logits
    top_k = min(top_k, logits.size(-1))
    indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
    logits = logits.masked_fill(indices_to_remove, torch.finfo(logits.dtype).min)
    return logits


def _sample_tokens(logits, temperature=0.0, top_p=None, top_k=None, alg="origin"):
    if temperature > 0:
        logits = logits / temperature
    if top_p is not None and top_p < 1:
        logits = _top_p_logits(logits, top_p)
    if top_k is not None:
        logits = _top_k_logits(logits, top_k)
    probs = torch.softmax(logits.float(), dim=-1)
    if temperature > 0:
        x0 = dists.Categorical(probs=probs).sample()
    else:
        _, x0 = probs.max(dim=-1)
    confidence = torch.gather(probs, -1, x0.unsqueeze(-1)).squeeze(-1)

    if alg == "topk_margin":
        sorted_probs, _ = torch.sort(probs, dim=-1, descending=True)
        confidence = sorted_probs[..., 0] - sorted_probs[..., 1]
    elif alg == "entropy":
        log_probs = torch.log(probs.clamp(min=1e-10))
        confidence = (probs * log_probs).sum(dim=-1)

    return confidence, x0


@dataclass
class ProRiboGenOutput(ModelOutput):
    """Output type for ProRiboGen diffusion generation."""
    sequences: torch.LongTensor = None
    history: Optional[Tuple[torch.LongTensor]] = None


class MDMGenerationConfig(GenerationConfig):
    """
    Configuration for Masked Diffusion Model generation.

    Args:
        steps (`int`, defaults to 50):
            Number of diffusion denoising steps. More steps generally yields higher quality.
        alg (`str`, defaults to `"random"`):
            Token unmasking order algorithm. One of:
            - `"random"`: randomly select positions to unmask at each step.
            - `"maskgit_plus"`: unmask positions with highest prediction confidence.
            - `"entropy"`: unmask positions with lowest prediction entropy.
            - `"topk_margin"`: unmask positions with highest top-1 vs top-2 probability margin.
            - `"origin"`: stochastically unmask each position with probability proportional to schedule.
            - `"p2"`: progressive prediction with remasking based on confidence.
        temperature (`float`, defaults to 1.0):
            Sampling temperature applied to the model logits. Higher values produce more diverse outputs.
        top_p (`float`, defaults to 0.9):
            Top-p (nucleus) sampling cutoff. Set to 1.0 to disable.
        top_k (`int`, defaults to 0):
            Top-k sampling cutoff. Set to 0 to disable.
        alg_temp (`float`, defaults to 0.9):
            Temperature for the confidence-based unmasking order (Gumbel-TopK). Set to 0 for deterministic order.
        output_history (`bool`, defaults to `False`):
            Whether to return the sequence state at each diffusion step.
    """
    def __init__(self, **kwargs):
        if "do_sample" not in kwargs:
            kwargs["do_sample"] = True
        super().__init__(**kwargs)
        self.temperature: float = kwargs.pop("temperature", 1.0)
        self.top_p: Optional[float] = kwargs.pop("top_p", 0.9)
        self.top_k: Optional[int] = kwargs.pop("top_k", 0)
        self.eps: float = kwargs.pop("eps", 1e-3)
        self.steps: int = kwargs.pop("steps", 50)
        self.alg: str = kwargs.pop("alg", "random")
        self.alg_temp: Optional[float] = kwargs.pop("alg_temp", 0.9)
        self.output_history: bool = kwargs.pop("output_history", False)
        self.mask_token_id = kwargs.pop("mask_token_id", None)
        self.num_return_sequences = kwargs.pop("num_return_sequences", 1)


class MDMGenerationMixin:
    """Mixin that adds masked diffusion generation to any MaskedLM model."""

    @staticmethod
    def _expand_inputs_for_generation(
        expand_size: int = 1,
        input_ids: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.LongTensor] = None,
    ):
        if expand_size == 1:
            return input_ids, attention_mask
        if input_ids is not None:
            input_ids = input_ids.repeat_interleave(expand_size, dim=0)
        if attention_mask is not None:
            attention_mask = attention_mask.repeat_interleave(expand_size, dim=0)
        return input_ids, attention_mask

    @torch.no_grad()
    def diffusion_generate(
        self,
        inputs: Optional[torch.Tensor] = None,
        generation_config: Optional[MDMGenerationConfig] = None,
        **kwargs,
    ) -> Union[ProRiboGenOutput, torch.LongTensor]:
        """
        Generate RNA sequences using masked diffusion.

        Args:
            inputs (`torch.LongTensor`):
                Input token IDs, typically all `<mask>` tokens representing the desired output length.
                Shape: `(1, sequence_length)`.
            generation_config (`MDMGenerationConfig`, *optional*):
                Generation configuration. If not provided, defaults are used.

        Returns:
            `ProRiboGenOutput` with `sequences` tensor of shape `(num_return_sequences, sequence_length)`.
        """
        if generation_config is None:
            generation_config = MDMGenerationConfig()
        generation_config.update(**kwargs)

        input_ids = inputs
        attention_mask = kwargs.get("attention_mask", None)

        if input_ids is None:
            raise ValueError("`inputs` must be provided for diffusion generation.")

        if generation_config.max_new_tokens is not None:
            generation_config.max_length = (
                input_ids.shape[-1] + generation_config.max_new_tokens
            )
        elif not hasattr(generation_config, "max_length") or generation_config.max_length is None:
            generation_config.max_length = input_ids.shape[-1]

        input_ids, attention_mask = self._expand_inputs_for_generation(
            expand_size=generation_config.num_return_sequences,
            input_ids=input_ids,
            attention_mask=attention_mask,
        )

        mask_token_id = generation_config.mask_token_id
        if mask_token_id is None:
            raise ValueError("`mask_token_id` must be set in the generation config.")

        x = F.pad(
            input_ids,
            (0, generation_config.max_length - input_ids.shape[1]),
            value=mask_token_id,
        )

        steps = generation_config.steps
        eps = generation_config.eps
        alg = generation_config.alg
        alg_temp = generation_config.alg_temp
        temperature = generation_config.temperature
        top_p = generation_config.top_p
        top_k = generation_config.top_k

        histories = [] if generation_config.output_history else None
        fix_mask = (x != mask_token_id)
        gen_attention_mask = (
            (x != self.config.pad_token_id).long()
            if self.config.pad_token_id is not None
            else None
        )
        timesteps = torch.linspace(1, eps, steps + 1, device=x.device)

        for i in trange(steps, desc="Diffusion"):
            mask_index = (x == mask_token_id)
            if not mask_index.any():
                break

            outputs = self(input_ids=x, attention_mask=gen_attention_mask)
            logits = outputs.logits
            mask_logits = logits[mask_index]
            t = timesteps[i]
            s = timesteps[i + 1]

            if alg == "origin":
                p_transfer = 1 - s / t if i < steps - 1 else 1
                x0 = torch.full_like(
                    x[mask_index], fill_value=mask_token_id, device=x.device, dtype=torch.long
                )
                transfer_index = torch.rand(*x0.shape, device=x.device) < p_transfer
                _, sampled = _sample_tokens(
                    mask_logits[transfer_index], temperature=temperature,
                    top_p=top_p, top_k=top_k, alg=alg
                )
                x0[transfer_index] = sampled
                x[mask_index] = x0

            elif alg == "p2":
                kappa_t = (i + 1) / steps
                confidence, x0 = _sample_tokens(
                    mask_logits, temperature, top_p, top_k, alg=alg
                )
                full_conf = torch.full_like(x, float("inf"), dtype=confidence.dtype)
                full_conf[mask_index] = confidence
                full_conf[fix_mask] = float("inf")
                num_positions = (~fix_mask).sum(dim=1, keepdim=True)
                num_to_mask = (num_positions.float() * (1 - kappa_t)).long()
                sorted_idx = torch.argsort(full_conf, dim=-1, descending=False)
                max_mask = num_to_mask.max()
                arange_mask = torch.arange(max_mask, device=x.device).unsqueeze(0) < num_to_mask
                to_mask_idx = sorted_idx[:, :max_mask][arange_mask]
                to_mask = torch.zeros_like(x, dtype=torch.bool)
                batch_idx = (
                    torch.arange(x.size(0), device=x.device)
                    .unsqueeze(1).expand(-1, max_mask)[arange_mask]
                )
                to_mask[batch_idx, to_mask_idx] = True
                x[to_mask] = mask_token_id
                mask_candidates = mask_index & ~to_mask
                x_proposals = torch.full_like(x, fill_value=mask_token_id)
                x_proposals[mask_index] = x0
                x[mask_candidates] = x_proposals[mask_candidates]

            elif alg in ["maskgit_plus", "entropy", "topk_margin"]:
                confidence, x0 = _sample_tokens(
                    mask_logits, temperature=temperature, top_p=top_p, top_k=top_k, alg=alg
                )
                confidence = confidence.to(mask_logits.dtype)
                num_mask_tokens = mask_index.sum(dim=1)
                if i < steps - 1:
                    n_transfer = (num_mask_tokens.float() * (1 - s / t)).long()
                else:
                    n_transfer = num_mask_tokens
                full_confidence = torch.full_like(x, -torch.inf, dtype=logits.dtype)
                full_confidence[mask_index] = confidence
                max_transfer = n_transfer.max().item()
                if max_transfer > 0:
                    if alg_temp is None or alg_temp == 0:
                        _, all_indices = torch.topk(full_confidence, max_transfer, dim=1)
                    else:
                        scaled = full_confidence / alg_temp
                        uniform = torch.rand_like(scaled).clamp_(1e-20, 1 - 1e-20)
                        scores = scaled + (-torch.log(-torch.log(uniform)))
                        _, all_indices = torch.topk(scores, max_transfer, dim=1)
                    valid_mask = (
                        torch.arange(max_transfer, device=x.device).unsqueeze(0)
                        < n_transfer.unsqueeze(1)
                    )
                    valid_indices = all_indices[valid_mask]
                    valid_batch = (
                        torch.arange(x.size(0), device=x.device)
                        .unsqueeze(1).expand_as(all_indices)[valid_mask]
                    )
                    x_ = torch.full_like(x, fill_value=mask_token_id)
                    x_[mask_index] = x0.clone()
                    x[valid_batch, valid_indices] = x_[valid_batch, valid_indices]

            elif alg == "random":
                _, x0 = _sample_tokens(
                    mask_logits, temperature=temperature, top_p=top_p, top_k=top_k, alg=alg
                )
                num_mask_tokens = mask_index.sum(dim=1)
                if i < steps - 1:
                    n_transfer = (num_mask_tokens.float() * (1 - s / t)).long()
                else:
                    n_transfer = num_mask_tokens
                max_transfer = n_transfer.max().item()
                if max_transfer > 0:
                    x_ = torch.full_like(x, fill_value=mask_token_id)
                    x_[mask_index] = x0.clone()
                    for b in range(x.size(0)):
                        positions = mask_index[b].nonzero(as_tuple=True)[0]
                        n = n_transfer[b].item()
                        if len(positions) > 0 and n > 0:
                            n = min(n, len(positions))
                            sel = torch.randperm(len(positions), device=x.device)[:n]
                            x[b, positions[sel]] = x_[b, positions[sel]]
            else:
                raise NotImplementedError(f"Algorithm '{alg}' is not implemented.")

            if histories is not None:
                histories.append(x.clone())

        if generation_config.return_dict_in_generate:
            return ProRiboGenOutput(sequences=x, history=histories)
        return x


class ProRiboGenForMaskedLM(EsmForMaskedLM, MDMGenerationMixin):
    """
    ProRiboGen: discrete diffusion language model on RNA tokens (ESM backbone).

    在本项目中与 RBP 蛋白条件结合,用于 RBP–RNA 条件下的序列建模与生成。

    继承 `EsmForMaskedLM`,并提供 `diffusion_generate()` 做迭代 mask 扩散解码。

    `diffusion_generate()` 用于(无条件或带条件的)RNA 序列生成;`forward()` 用于标准 MLM 前向。
    """

    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        labels: Optional[torch.LongTensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, MaskedLMOutput]:
        return super().forward(
            input_ids=input_ids,
            attention_mask=attention_mask,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            labels=labels,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

'''