File size: 27,514 Bytes
8ebda9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# coding=utf-8
# Copyright (c) 2019, NVIDIA CORPORATION.  All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""GPT-2 model."""

import torch
import torch.nn.functional as F
import math
import torch.nn as nn
from torch.nn import CrossEntropyLoss
# from ......configuration_transfo_xl import TransfoXLConfig 
from transformers import TransfoXLConfig

from transformers.modeling_utils import (
    PreTrainedModel
)


class PositionalEmbedding(torch.nn.Module):
    def __init__(self, hidden_size):
        super(PositionalEmbedding, self).__init__()

        self.hidden_size = hidden_size

        inv_freq = 1 / (10000 ** (torch.arange(0.0, hidden_size, 2.0) / hidden_size))
        self.register_buffer('inv_freq', inv_freq)

    def forward(self, pos_seq, bsz=None):
        sinusoid_inp = torch.ger(pos_seq, self.inv_freq)
        pos_emb = torch.cat([sinusoid_inp.sin(), sinusoid_inp.cos()], dim=-1)

        if bsz is not None:
            return pos_emb[None, :, :].expand(bsz, -1, -1)
        else:
            return pos_emb[None, :, :]

def ensure_divisibility(numerator, denominator):
    """Ensure that numerator is divisible by the denominator."""
    assert numerator % denominator == 0, '{} is not divisible by {}'.format(
        numerator, denominator)

def divide(numerator, denominator):
    """Ensure that numerator is divisible by the denominator and return
    the division value."""
    ensure_divisibility(numerator, denominator)
    return numerator // denominator

def scaled_init_method(sigma, num_layers):
    """Init method based on N(0, sigma/sqrt(2*num_layers)."""
    std = sigma / math.sqrt(2.0 * num_layers)
    def init_(tensor):
        return torch.nn.init.normal_(tensor, mean=0.0, std=std)

    return init_

def unscaled_init_method(sigma):
    """Init method based on N(0, sigma)."""
    def init_(tensor):
        return torch.nn.init.normal_(tensor, mean=0.0, std=sigma)

    return init_

@torch.jit.script
def gelu_impl(x):
     """OpenAI's gelu implementation."""
     return 0.5 * x * (1.0 + torch.tanh(0.7978845608028654 * x *
                                        (1.0 + 0.044715 * x * x)))

def gelu(x):
    return gelu_impl(x)

class GPT2SelfAttention(torch.nn.Module):
    """Parallel self-attention layer for GPT2.

    Self-attention layer takes input with size [b, s, h] where b is
    the batch size, s is the sequence lenght, and h is the hidden size
    and creates output of the same size.
    Arguments:
        hidden_size: total hidden size of the layer (h).
        num_attention_heads: number of attention heads (n). Note that we
                             require n to be divisible by number of GPUs
                             used to parallelize the model. Also, we
                             require hidden size to be divisible by n.
        dropout_prob: dropout probability for the attention scores.
        init_method: weight initialization.
        output_layer_init_method: output layer initialization. If None, use
                                  `init_method`.
    We use the following notation:
        h: hidden_size
        n: num_attention_heads
        p: number of partitions
        np: n/p
        hp: h/p
        hn: h/n
        b: batch size
        s: sequence length
    """
    def __init__(self, hidden_size, num_attention_heads,
                 attention_dropout_prob, output_dropout_prob,
                 init_method, output_layer_init_method=None, relative_encoding=False):
        super(GPT2SelfAttention, self).__init__()
        # Set output layer initialization if not provided.
        if output_layer_init_method is None:
            output_layer_init_method = init_method
        # Per attention head and per partition values.
        self.hidden_size_per_partition = hidden_size
        self.hidden_size_per_attention_head = divide(hidden_size,
                                                     num_attention_heads)
        self.num_attention_heads_per_partition = num_attention_heads
        self.relative_encoding = relative_encoding
        # Strided linear layer.
        self.query_key_value = torch.nn.Linear(hidden_size, 3*hidden_size, bias=True)

        if relative_encoding:
            self.relative = torch.nn.Linear(hidden_size, hidden_size, bias=True)
        # Dropout. Note that for a single iteration, this layer will generate
        # different outputs on different number of parallel partitions but
        # on average it should not be partition dependent.
        self.attention_dropout = torch.nn.Dropout(attention_dropout_prob)

        # Output.
        self.dense = torch.nn.Linear(hidden_size, hidden_size, bias=True)
        self.output_dropout = torch.nn.Dropout(output_dropout_prob)

    def _transpose_for_scores(self, tensor):
        """Transpose a 3D tensor [b, s, np*hn] into a 4D tensor with
        size [b, np, s, hn].
        """
        new_tensor_shape = tensor.size()[:-1] + \
                           (self.num_attention_heads_per_partition,
                            self.hidden_size_per_attention_head)
        tensor = tensor.view(*new_tensor_shape)
        return tensor.permute(0, 2, 1, 3)

    @staticmethod
    def _rel_shift(x, zero_triu=False):
        # ql x kl x bsz x h
        # bsz x h x ql x kl
        zero_pad = torch.zeros((*x.size()[:-2], x.size(-2), 1),
                               device=x.device, dtype=x.dtype)
        x_padded = torch.cat([zero_pad, x], dim=-1)

        x_padded = x_padded.view(*x.size()[:-2], x.size(-1) + 1, x.size(-2))

        x = x_padded[:, :, 1:].view_as(x)

        if zero_triu:
            ones = torch.ones((x.size(0), x.size(1)))
            x = x * torch.tril(ones, x.size(1) - x.size(0))[:, :, None, None]

        return x

    @staticmethod
    def _rel_shift_latest(x: torch.Tensor):
        ndims = x.dim()
        x_shape = x.size()
        row_dim = 2
        col_dim = row_dim + 1
        assert col_dim < ndims
        tgt_shape_1, tgt_shape_2 = [], []
        for i in range(ndims):
            if i == row_dim:
                tgt_shape_1.append(x_shape[col_dim])
                tgt_shape_2.append(x_shape[row_dim])
            elif i == col_dim:
                tgt_shape_1.append(x_shape[row_dim])
                tgt_shape_2.append(x_shape[col_dim] - 1)
            else:
                tgt_shape_1.append(x_shape[i])
                tgt_shape_2.append(x_shape[i])
        x = x.view(*tgt_shape_1)
        x = x[:, :, 1:, :]
        x = x.view(*tgt_shape_2)
        return x

    def forward(self, hidden_states, ltor_mask, position_embeddings=None, r_w_bias=None, r_r_bias=None, mem=None):
        # hidden_states: [b, s, h]
        # ltor_mask: [1, 1, s, s]

        # Attention heads. [b, s, hp]
        query_length = hidden_states.size(1)

        if mem is None:
            mixed_x_layer = self.query_key_value(hidden_states)
            (mixed_query_layer,
             mixed_key_layer,
             mixed_value_layer) = torch.chunk(mixed_x_layer, 3, dim=-1)
        else:
            cat = torch.cat((mem, hidden_states), 1)
            mixed_x_layer = self.query_key_value(cat)
            (mixed_query_layer,
             mixed_key_layer,
             mixed_value_layer) = torch.chunk(mixed_x_layer, 3, dim=-1)
            mixed_query_layer = mixed_query_layer[:, -query_length:]

        # Reshape and transpose [b, np, s, hn]
        query_layer = self._transpose_for_scores(mixed_query_layer)
        key_layer = self._transpose_for_scores(mixed_key_layer)
        value_layer = self._transpose_for_scores(mixed_value_layer)
        if self.relative_encoding:
            relative_layer = self.relative(position_embeddings)
            relative_layer = self._transpose_for_scores(relative_layer)  # 1 (bsz) x n_head x klen x d_head
            # Raw attention scores. [b, np, qs, ks]
            rw_head_q = query_layer + r_w_bias.unsqueeze(1)
            ac_score = torch.matmul(rw_head_q, key_layer.transpose(-1, -2))
            rr_head_q = query_layer + r_r_bias.unsqueeze(1)
            bd_score = torch.matmul(rr_head_q, relative_layer.transpose(-1, -2))
            bd_score = self._rel_shift(bd_score)  # qlen x klen x bsz x n_head
            # bd_score = bd_score.permute(2, 3, 0, 1) # bsz n_head qlen klen

            attention_scores = ac_score + bd_score
        else:
            # Raw attention scores. [b, np, s, s]
            attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2))
        attention_scores = attention_scores / math.sqrt(
            self.hidden_size_per_attention_head)

        # Apply the left to right attention mask.
        attention_scores = torch.mul(attention_scores, ltor_mask) - \
                           10000.0 * (1.0 - ltor_mask)

        # Attention probabilities. [b, np, s, s]
        attention_probs = torch.nn.Softmax(dim=-1)(attention_scores)
        # This is actually dropping out entire tokens to attend to, which might
        # seem a bit unusual, but is taken from the original Transformer paper.
        # with get_cuda_rng_tracker().fork():
        #     attention_probs = self.attention_dropout(attention_probs)

        # Context layer.
        # [b, np, s, hn]
        # print(f'attn_probs {attention_probs}, value_layer {value_layer}')
        context_layer = torch.matmul(attention_probs, value_layer.float())
        # [b, s, np, hn]
        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
        new_context_layer_shape = context_layer.size()[:-2] + \
                                  (self.hidden_size_per_partition,)
        # [b, s, hp]
        context_layer = context_layer.view(*new_context_layer_shape)

        # Output. [b, s, h]
        output = self.dense(context_layer)
        output = self.output_dropout(output)

        return output

class GPT2MLP(torch.nn.Module):
    """MLP for GPT2.

    MLP will take the input with h hidden state, project it to 4*h
    hidden dimension, perform gelu transformation, and project the
    state back into h hidden dimension. At the end, dropout is also
    applied.

    Arguments:
        hidden_size: The hidden size of the self attention.
        output_dropout_prob: dropout probability for the outputs
                             after self attention and final output.
        init_method: initialization method used for the weights. Note
                     that all biases are initialized to zero and
                     layernorm weight are initialized to one.
        output_layer_init_method: output layer initialization. If None,
                                  use `init_method`.
    """

    def __init__(self, hidden_size, output_dropout_prob, init_method,
                 output_layer_init_method=None):
        super(GPT2MLP, self).__init__()
        # Set output layer initialization if not provided.
        if output_layer_init_method is None:
            output_layer_init_method = init_method
        # Project to 4h.
        self.dense_h_to_4h = torch.nn.Linear(hidden_size, 4*hidden_size)
        # Project back to h.
        self.dense_4h_to_h = torch.nn.Linear(4*hidden_size, hidden_size)
        self.dropout = torch.nn.Dropout(output_dropout_prob)

    def forward(self, hidden_states):
        # [b, s, 4hp]
        intermediate_parallel = self.dense_h_to_4h(hidden_states)
        intermediate_parallel = gelu(intermediate_parallel)

        # [b, s, h]
        output = self.dense_4h_to_h(intermediate_parallel)
        output = self.dropout(output)
        return output


class GPT2TransformerLayer(torch.nn.Module):
    """A single layer transformer for GPT2.

    We use the following notation:
        h: hidden size
        n: number of attention heads
        b: batch size
        s: sequence length
    Transformore layer takes input with size [b, s, h] and returns an
    output of the same size.

    Arguments:
        hidden_size: The hidden size of the self attention.
        num_attention_heads: number of attention head in the self
                             attention.
        attention_dropout_prob: dropout probability of the attention
                                score in self attention.
        output_dropout_prob: dropout probability for the outputs
                             after self attention and final output.
        layernorm_epsilon: epsilon used in layernorm to avoid
                           division by zero.
        init_method: initialization method used for the weights. Note
                     that all biases are initialized to zero and
                     layernorm weight are initialized to one.
        output_layer_init_method: output layers (attention output and
                                  mlp output) initialization. If None,
                                  use `init_method`.
    """
    def __init__(self,
                 hidden_size,
                 num_attention_heads,
                 attention_dropout_prob,
                 output_dropout_prob,
                 layernorm_epsilon,
                 init_method,
                 output_layer_init_method=None,
                 relative_encoding=False):
        super(GPT2TransformerLayer, self).__init__()
        # Set output layer initialization if not provided.
        if output_layer_init_method is None:
            output_layer_init_method = init_method

        # Layernorm on the input data.
        self.input_layernorm = torch.nn.LayerNorm(hidden_size, eps=layernorm_epsilon)

        # Self attention.
        self.attention = GPT2SelfAttention(
            hidden_size,
            num_attention_heads,
            attention_dropout_prob,
            output_dropout_prob,
            init_method,
            output_layer_init_method=output_layer_init_method,
            relative_encoding=relative_encoding)

        # Layernorm on the input data.
        self.post_attention_layernorm = torch.nn.LayerNorm(hidden_size,
                                                  eps=layernorm_epsilon)

        # MLP
        self.mlp = GPT2MLP(
            hidden_size,
            output_dropout_prob,
            init_method,
            output_layer_init_method=output_layer_init_method)

    def forward(self, hidden_states, ltor_mask, position_embeddings=None, r_w_bias=None, r_r_bias=None, mem=None):
        # hidden_states: [b, s, h]
        # ltor_mask: [1, 1, s, s]

        # Layer norm at the begining of the transformer layer.
        layernorm_output = self.input_layernorm(hidden_states)
        mem = self.input_layernorm(mem) if mem is not None else None
        # Self attention.
        attention_output = self.attention(layernorm_output, ltor_mask, position_embeddings, r_w_bias, r_r_bias, mem)
        # Residual connection.
        # print(f'hz {hidden_states.shape}, attn {attention_output.shape}')
        layernorm_input = hidden_states + attention_output
        # Layer norm post the self attention.
        layernorm_output = self.post_attention_layernorm(layernorm_input)
        # MLP.
        mlp_output = self.mlp(layernorm_output)
        # Second residual connection.
        output = layernorm_input + mlp_output

        return output

class GPT2TransformerForLatent(torch.nn.Module):
    """GPT-2 transformer.

    This module takes input from embedding layer and it's output can
    be used directly by a logit layer. It consists of L (num-layers)
    blocks of:
        layer norm
        self attention
        residual connection
        layer norm
        mlp
        residual connection
    followed by a final layer norm.

    Arguments:
        num_layers: Number of transformer layers.
        hidden_size: The hidden size of the self attention.
        num_attention_heads: number of attention head in the self
                             attention.
        attention_dropout_prob: dropout probability of the attention
                                score in self attention.
        output_dropout_prob: dropout probability for the outputs
                             after self attention and final output.
        checkpoint_activations: if True, checkpoint activations.
        checkpoint_num_layers: number of layers to checkpoint. This
                               is basically the chunk size in checkpoitning.
        layernorm_epsilon: epsilon used in layernorm to avoid
                           division by zero.
        init_method_std: standard deviation of the init method which has
                         the form N(0, std).
        use_scaled_init_for_output_weights: If Ture use 1/sqrt(2*num_layers)
                                            scaling for the output weights (
                                            output of self attention and mlp).
    """
    def __init__(self,
                 num_layers,
                 hidden_size,
                 num_attention_heads,
                 max_sequence_length,
                 max_memory_length,
                 embedding_dropout_prob,
                 attention_dropout_prob,
                 output_dropout_prob,
                 checkpoint_activations,
                 latent_size = 64,
                 checkpoint_num_layers=1,
                 layernorm_epsilon=1.0e-5,
                 init_method_std=0.02,
                 use_scaled_init_for_output_weights=True,
                 relative_encoding=False):
        super(GPT2TransformerForLatent, self).__init__()
        # Store activation checkpoiting flag.
        self.checkpoint_activations = checkpoint_activations
        self.checkpoint_num_layers = checkpoint_num_layers
        self.max_memory_length = max_memory_length

        self.latent_size = latent_size
        # self.linear = nn.Linear(self.latent_size, hidden_size * num_layers, bias=False).float() # different latent vector for each layer 
        # self.linear_emb = nn.Linear(self.latent_size, hidden_size * num_layers, bias=False).float()
        self.linear_emb = nn.Linear(self.latent_size, hidden_size, bias=False).float()
        
        # torch.nn.init.normal_(self.linear.weight, mean=0.0, std=init_method_std)
        torch.nn.init.normal_(self.linear_emb.weight, mean=0.0, std=init_method_std)


        output_layer_init_method = None
        if use_scaled_init_for_output_weights:
            output_layer_init_method = scaled_init_method(init_method_std,
                                                      num_layers)
        # Embeddings dropout
        self.embedding_dropout = torch.nn.Dropout(embedding_dropout_prob)
        self.relative_encoding = relative_encoding
        if relative_encoding:
            # Relative position embedding
            self.position_embeddings = PositionalEmbedding(hidden_size)
            # Per attention head and per partition values.
            self.hidden_size_per_attention_head = divide(hidden_size,
                                                         num_attention_heads)
            self.num_attention_heads_per_partition = num_attention_heads
            self.r_w_bias = torch.nn.Parameter(
                torch.Tensor(self.num_attention_heads_per_partition, self.hidden_size_per_attention_head))
            self.r_r_bias = torch.nn.Parameter(
                torch.Tensor(self.num_attention_heads_per_partition, self.hidden_size_per_attention_head))

            # Always initialize bias to zero.
            with torch.no_grad():
                self.r_w_bias.zero_()
                self.r_r_bias.zero_()
        else:
            # Position embedding (serial).
            self.position_embeddings = torch.nn.Embedding(max_sequence_length,
                                                          hidden_size)
            # Initialize the position embeddings.
            torch.nn.init.normal_(self.position_embeddings.weight, mean=0.0, std=init_method_std)

        def get_layer():
            return GPT2TransformerLayer(
                hidden_size,
                num_attention_heads,
                attention_dropout_prob,
                output_dropout_prob,
                layernorm_epsilon,
                unscaled_init_method(init_method_std),
                output_layer_init_method=output_layer_init_method,
                relative_encoding=relative_encoding)

        # Transformer layers.
        self.layers = torch.nn.ModuleList(
            [get_layer() for _ in range(num_layers)])

        # Final layer norm before output.
        self.final_layernorm = torch.nn.LayerNorm(hidden_size, eps=layernorm_epsilon)


    def forward(self, hidden_states, attention_mask, latent_state, mems):
        batch_size, query_length, hidden_size = hidden_states.size()
        # memory_length = self.latent_size
        memory_length = mems[0].size(1) if mems else 0

        # key_length = query_length + memory_length+1
        # attention_mask = attention_mask[:, :, :, -query_length-memory_length-1:]
        key_length = query_length + memory_length
        attention_mask = attention_mask[:, :, :, -query_length - memory_length:]

        if latent_state is not None: 
            latent_emb = self.linear_emb(latent_state)
            # latent_emb = torch.split(latent_emb.unsqueeze(1), hidden_size, dim=2)
            latent_emb = latent_emb.unsqueeze(1)
        # print(f'latent_state {latent_state.half()}\n linear_emb {self.linear_emb.weight} \n latent_emb {latent_emb}')
        # torch.save(latent_state, '/cognitive_comp/wanghao/experiments/fengshen/latent_state.pt')
        # torch.save(self.linear_emb, '/cognitive_comp/wanghao/experiments/fengshen/weight.pt')


        position_sequence = torch.arange(key_length - 1, -1, -1.0, device=hidden_states.device,
                                         dtype=hidden_states.dtype)
        position_embeddings = self.position_embeddings(position_sequence)

        # print(f'pos {position_embeddings.shape}, latent {latent_emb.shape}')
        # if latent_state is not None:
        #     position_embeddings += latent_emb.unsqueeze(0)
        # Apply dropout
        position_embeddings = self.embedding_dropout(position_embeddings)

        # print(f'latent_emb {latent_emb.shape}, {hidden_states.shape}')
        if latent_state is not None:
            hidden_states = hidden_states + latent_emb 
        hidden_states = self.embedding_dropout(hidden_states)

        # latent_mem = self.linear(latent_state.half())
        # latent_mem = torch.split(latent_mem.unsqueeze(1), hidden_size, dim=2)

        if self.max_memory_length > 0:
            mem_layers = [hidden_states.detach()]
        else:
            mem_layers = []

        for i, layer in enumerate(self.layers):
            args = [hidden_states, attention_mask]
            if self.relative_encoding:
                args += [position_embeddings, self.r_w_bias, self.r_r_bias]

            mem_i = mems[i] if mems else None
            # print(f'mems {len(mems)} {mems[0].shape}')
            # mem_i = torch.cat((latent_mem[i], mems[i]), 1) if mems else latent_mem[i]
            # print(f'mem_i {mem_i.shape}, {mem_i}')
            hidden_states = layer(*args, mem=mem_i)

            if latent_state is not None:
                hidden_states = hidden_states + latent_emb

            if self.max_memory_length > 0:
                mem_layers.append(hidden_states.detach())
        # print(f'mem_layers {len(mem_layers)} mems {len(mems)}')
        # Final layer norm.
        output = self.final_layernorm(hidden_states)
        if self.max_memory_length > 0:
            mem_layers = self.update_mems(mem_layers, mems)

        return (output, mem_layers)

    def update_mems(self, hiddens, mems):
        memory_length = mems[0].size(1) if mems else 0
        query_length = hiddens[0].size(1)
        new_memory_length = min(self.max_memory_length, memory_length + query_length)
        new_mems = []
        with torch.no_grad():
            for i in range(len(hiddens)):
                if new_memory_length <= query_length:
                    new_mems.append(hiddens[i][:, -new_memory_length:])
                else:
                    new_mems.append(torch.cat((mems[i][:, -new_memory_length+query_length:], hiddens[i]), dim=1))
        return new_mems


class GPT2ModelForLatent(PreTrainedModel):
    """GPT-2 Language model.

    The output of the forward method are the logits (parallel or
    serial depending on the `parallel_output` flag.
    """

    def _init_weights(self, module):
        """ Initialize the weights """
        pass  # to bypass the not implement error 

    def __init__(self, config:TransfoXLConfig):
        super().__init__(config)
        self.config = config

        self.word_embeddings = torch.nn.Embedding(config.vocab_size, config.hidden_size)

        # Transformer
        self.transformer = GPT2TransformerForLatent(config.num_layers,
                                                    config.hidden_size,
                                                    config.num_attention_heads,
                                                    config.max_sequence_length,
                                                    config.max_memory_length,
                                                    config.embedding_dropout_prob,
                                                    config.attention_dropout_prob,
                                                    config.output_dropout_prob,
                                                    config.checkpoint_activations,
                                                    config.latent_size,
                                                    config.checkpoint_num_layers,
                                                    relative_encoding=config.relative_encoding)


    def forward(self, input_ids, attention_mask, latent_state, mems=None, labels=None, label_ignore=None):
        embeddings = self.word_embeddings(input_ids)

        # Transformer.
        logits, hidden_layers = self.transformer(embeddings, attention_mask, latent_state, mems)
        lm_logits = F.linear(logits,
                            self.word_embeddings.weight)

        outputs = (lm_logits, hidden_layers) # (bz, sql, vocab), ()
        if labels is not None:
            shift_logits = lm_logits[..., :-1, :].contiguous()
            shift_labels = labels[..., 1:].contiguous()

            loss_fct = CrossEntropyLoss(ignore_index=label_ignore, reduce=False)
            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)),
                            shift_labels.view(-1))
            loss = torch.sum(loss.view(-1, shift_labels.shape[-1]), -1)
            outputs = (loss,) + outputs

        return outputs

    def get_attn_mask(self, seq_length):
        # mem_length = self.config.max_memory_length + 1
        mem_length = self.config.max_memory_length
        attention_mask = torch.ones((1, seq_length, seq_length + mem_length))
        attention_mask = torch.tril(torch.triu(attention_mask, 1 - seq_length + mem_length), mem_length)
        attention_mask = attention_mask.unsqueeze(1)
        return attention_mask