optimum-rbln 0.1.0__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- optimum/rbln/__init__.py +115 -0
- optimum/rbln/__version__.py +1 -0
- optimum/rbln/diffusers/__init__.py +64 -0
- optimum/rbln/diffusers/models/__init__.py +26 -0
- optimum/rbln/diffusers/models/autoencoder_kl.py +313 -0
- optimum/rbln/diffusers/models/controlnet.py +180 -0
- optimum/rbln/diffusers/models/unet_2d_condition.py +352 -0
- optimum/rbln/diffusers/pipelines/__init__.py +30 -0
- optimum/rbln/diffusers/pipelines/controlnet/__init__.py +24 -0
- optimum/rbln/diffusers/pipelines/controlnet/multicontrolnet.py +266 -0
- optimum/rbln/diffusers/pipelines/stable_diffusion/__init__.py +26 -0
- optimum/rbln/diffusers/pipelines/stable_diffusion/pipeline_controlnet_img2img.py +731 -0
- optimum/rbln/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py +106 -0
- optimum/rbln/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py +116 -0
- optimum/rbln/diffusers/pipelines/stable_diffusion_xl/__init__.py +2 -0
- optimum/rbln/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py +109 -0
- optimum/rbln/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py +111 -0
- optimum/rbln/modeling.py +0 -0
- optimum/rbln/modeling_alias.py +49 -0
- optimum/rbln/modeling_base.py +645 -0
- optimum/rbln/modeling_config.py +169 -0
- optimum/rbln/modeling_seq2seq.py +469 -0
- optimum/rbln/transformers/__init__.py +59 -0
- optimum/rbln/transformers/generation/__init__.py +24 -0
- optimum/rbln/transformers/generation/streamers.py +122 -0
- optimum/rbln/transformers/models/__init__.py +28 -0
- optimum/rbln/transformers/models/bart/__init__.py +24 -0
- optimum/rbln/transformers/models/bart/bart_architecture.py +377 -0
- optimum/rbln/transformers/models/clip/__init__.py +24 -0
- optimum/rbln/transformers/models/clip/modeling_clip.py +116 -0
- optimum/rbln/transformers/models/gpt2/__init__.py +24 -0
- optimum/rbln/transformers/models/gpt2/gpt2_architecture.py +253 -0
- optimum/rbln/transformers/models/gpt2/modeling_gpt2.py +700 -0
- optimum/rbln/transformers/models/llama/__init__.py +24 -0
- optimum/rbln/transformers/models/llama/llama_architecture.py +607 -0
- optimum/rbln/transformers/models/llama/modeling_llama.py +409 -0
- optimum/rbln/transformers/models/t5/__init__.py +24 -0
- optimum/rbln/transformers/models/t5/t5_architecture.py +439 -0
- optimum/rbln/transformers/models/wav2vec2/__init__.py +24 -0
- optimum/rbln/transformers/models/wav2vec2/modeling_wav2vec2.py +121 -0
- optimum/rbln/transformers/models/whisper/__init__.py +24 -0
- optimum/rbln/transformers/models/whisper/modeling_whisper.py +374 -0
- optimum/rbln/transformers/models/whisper/whisper_architecture.py +406 -0
- optimum/rbln/utils/__init__.py +25 -0
- optimum/rbln/utils/import_utils.py +28 -0
- optimum/rbln/utils/runtime_utils.py +71 -0
- optimum/rbln/utils/save_utils.py +92 -0
- optimum_rbln-0.1.0.dist-info/METADATA +144 -0
- optimum_rbln-0.1.0.dist-info/RECORD +51 -0
- optimum_rbln-0.1.0.dist-info/WHEEL +4 -0
- optimum_rbln-0.1.0.dist-info/licenses/LICENSE +201 -0
@@ -0,0 +1,116 @@
|
|
1
|
+
# Copyright 2024 Rebellions Inc.
|
2
|
+
|
3
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
4
|
+
# you may not use this file except in compliance with the License.
|
5
|
+
# You may obtain a copy of the License at:
|
6
|
+
|
7
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
8
|
+
|
9
|
+
# Unless required by applicable law or agreed to in writing, software
|
10
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
11
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
12
|
+
# See the License for the specific language governing permissions and
|
13
|
+
# limitations under the License.
|
14
|
+
|
15
|
+
# Portions of this software are licensed under the Apache License,
|
16
|
+
# Version 2.0. See the NOTICE file distributed with this work for
|
17
|
+
# additional information regarding copyright ownership.
|
18
|
+
|
19
|
+
# All other portions of this software, including proprietary code,
|
20
|
+
# are the intellectual property of Rebellions Inc. and may not be
|
21
|
+
# copied, modified, or distributed without prior written permission
|
22
|
+
# from Rebellions Inc.
|
23
|
+
|
24
|
+
import logging
|
25
|
+
from typing import TYPE_CHECKING, Optional, Union
|
26
|
+
|
27
|
+
import torch
|
28
|
+
from transformers import AutoConfig, AutoModel, CLIPTextConfig, CLIPTextModel, CLIPTextModelWithProjection
|
29
|
+
from transformers.models.clip.modeling_clip import CLIPTextModelOutput
|
30
|
+
|
31
|
+
from ....modeling_base import RBLNModel
|
32
|
+
from ....modeling_config import RBLNConfig, RBLNRuntimeConfig
|
33
|
+
|
34
|
+
|
35
|
+
logger = logging.getLogger(__name__)
|
36
|
+
|
37
|
+
if TYPE_CHECKING:
|
38
|
+
from transformers import AutoFeatureExtractor, AutoProcessor, AutoTokenizer, CLIPTextModel
|
39
|
+
|
40
|
+
|
41
|
+
class _TextEncoder(torch.nn.Module):
|
42
|
+
def __init__(self, enc: "CLIPTextModel"):
|
43
|
+
super().__init__()
|
44
|
+
enc.config.return_dict = False
|
45
|
+
enc.config.output_hidden_states = True
|
46
|
+
self.enc = enc
|
47
|
+
|
48
|
+
def forward(self, inp):
|
49
|
+
enc_out = self.enc(inp)
|
50
|
+
return enc_out
|
51
|
+
|
52
|
+
|
53
|
+
class RBLNCLIPTextModel(RBLNModel):
|
54
|
+
model_type = "rbln_clip"
|
55
|
+
auto_model_class = AutoModel # feature extraction
|
56
|
+
original_model_class = CLIPTextModel
|
57
|
+
original_config_class = CLIPTextConfig
|
58
|
+
|
59
|
+
def __post_init__(self, **kwargs):
|
60
|
+
self.dtype = torch.float32
|
61
|
+
|
62
|
+
@classmethod
|
63
|
+
def from_pretrained(cls, *args, **kwargs):
|
64
|
+
configtmp = AutoConfig.from_pretrained
|
65
|
+
modeltmp = AutoModel.from_pretrained
|
66
|
+
AutoConfig.from_pretrained = cls.original_config_class.from_pretrained
|
67
|
+
AutoModel.from_pretrained = cls.original_model_class.from_pretrained
|
68
|
+
rt = super().from_pretrained(*args, **kwargs)
|
69
|
+
AutoConfig.from_pretrained = configtmp
|
70
|
+
AutoModel.from_pretrained = modeltmp
|
71
|
+
return rt
|
72
|
+
|
73
|
+
@classmethod
|
74
|
+
def wrap_model_if_needed(cls, model: torch.nn.Module) -> torch.nn.Module:
|
75
|
+
return _TextEncoder(model).eval()
|
76
|
+
|
77
|
+
@classmethod
|
78
|
+
def _get_rbln_config(
|
79
|
+
cls,
|
80
|
+
preprocessors: Union["AutoFeatureExtractor", "AutoProcessor", "AutoTokenizer"],
|
81
|
+
model_config: "CLIPTextConfig",
|
82
|
+
rbln_batch_size: Optional[int] = None,
|
83
|
+
rbln_img_width: Optional[int] = None,
|
84
|
+
rbln_img_height: Optional[int] = None,
|
85
|
+
) -> RBLNConfig:
|
86
|
+
model_config.return_dict = False
|
87
|
+
if rbln_batch_size is None:
|
88
|
+
rbln_batch_size = 1
|
89
|
+
|
90
|
+
rbln_runtime_config = RBLNRuntimeConfig(
|
91
|
+
input_info=[
|
92
|
+
(
|
93
|
+
"input_ids",
|
94
|
+
[
|
95
|
+
rbln_batch_size,
|
96
|
+
model_config.max_position_embeddings,
|
97
|
+
],
|
98
|
+
"int64",
|
99
|
+
),
|
100
|
+
],
|
101
|
+
)
|
102
|
+
|
103
|
+
rbln_config = RBLNConfig.from_rbln_runtime_configs([rbln_runtime_config])
|
104
|
+
return rbln_config
|
105
|
+
|
106
|
+
def forward(self, input_ids: "torch.Tensor", **kwargs):
|
107
|
+
text_output = super().forward(input_ids)
|
108
|
+
return CLIPTextModelOutput(
|
109
|
+
text_embeds=text_output[0],
|
110
|
+
last_hidden_state=text_output[1],
|
111
|
+
hidden_states=text_output[2:],
|
112
|
+
)
|
113
|
+
|
114
|
+
|
115
|
+
class RBLNCLIPTextModelWithProjection(RBLNCLIPTextModel):
|
116
|
+
original_model_class = CLIPTextModelWithProjection
|
@@ -0,0 +1,24 @@
|
|
1
|
+
# Copyright 2024 Rebellions Inc.
|
2
|
+
|
3
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
4
|
+
# you may not use this file except in compliance with the License.
|
5
|
+
# You may obtain a copy of the License at:
|
6
|
+
|
7
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
8
|
+
|
9
|
+
# Unless required by applicable law or agreed to in writing, software
|
10
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
11
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
12
|
+
# See the License for the specific language governing permissions and
|
13
|
+
# limitations under the License.
|
14
|
+
|
15
|
+
# Portions of this software are licensed under the Apache License,
|
16
|
+
# Version 2.0. See the NOTICE file distributed with this work for
|
17
|
+
# additional information regarding copyright ownership.
|
18
|
+
|
19
|
+
# All other portions of this software, including proprietary code,
|
20
|
+
# are the intellectual property of Rebellions Inc. and may not be
|
21
|
+
# copied, modified, or distributed without prior written permission
|
22
|
+
# from Rebellions Inc.
|
23
|
+
|
24
|
+
from .modeling_gpt2 import RBLNGPT2LMHeadModel
|
@@ -0,0 +1,253 @@
|
|
1
|
+
# Copyright 2024 Rebellions Inc.
|
2
|
+
|
3
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
4
|
+
# you may not use this file except in compliance with the License.
|
5
|
+
# You may obtain a copy of the License at:
|
6
|
+
|
7
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
8
|
+
|
9
|
+
# Unless required by applicable law or agreed to in writing, software
|
10
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
11
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
12
|
+
# See the License for the specific language governing permissions and
|
13
|
+
# limitations under the License.
|
14
|
+
|
15
|
+
# Portions of this software are licensed under the Apache License,
|
16
|
+
# Version 2.0. See the NOTICE file distributed with this work for
|
17
|
+
# additional information regarding copyright ownership.
|
18
|
+
|
19
|
+
# All other portions of this software, including proprietary code,
|
20
|
+
# are the intellectual property of Rebellions Inc. and may not be
|
21
|
+
# copied, modified, or distributed without prior written permission
|
22
|
+
# from Rebellions Inc.
|
23
|
+
|
24
|
+
from typing import Optional, Tuple, Union
|
25
|
+
|
26
|
+
import torch
|
27
|
+
import torch.nn as nn
|
28
|
+
from transformers.modeling_outputs import (
|
29
|
+
BaseModelOutputWithPast,
|
30
|
+
BaseModelOutputWithPastAndCrossAttentions,
|
31
|
+
)
|
32
|
+
from transformers.models.gpt2.modeling_gpt2 import GPT2Attention, GPT2Block, GPT2Model
|
33
|
+
|
34
|
+
|
35
|
+
class _GPT2Attention(GPT2Attention):
|
36
|
+
def _attn(self, query, key, value, attention_mask=None, head_mask=None):
|
37
|
+
attn_weights = torch.matmul(query, key.transpose(-1, -2))
|
38
|
+
|
39
|
+
if self.scale_attn_weights:
|
40
|
+
attn_weights = attn_weights / torch.full(
|
41
|
+
[], value.size(-1) ** 0.5, dtype=attn_weights.dtype, device=attn_weights.device
|
42
|
+
)
|
43
|
+
|
44
|
+
# Layer-wise attention scaling
|
45
|
+
if self.scale_attn_by_inverse_layer_idx:
|
46
|
+
attn_weights = attn_weights / float(self.layer_idx + 1)
|
47
|
+
|
48
|
+
# -------------------
|
49
|
+
# Below are deleted since "where" op does not supported on RBLN graph.
|
50
|
+
# -------------------
|
51
|
+
# if not self.is_cross_attention:
|
52
|
+
# # if only "normal" attention layer implements causal mask
|
53
|
+
# query_length, key_length = query.size(-2), key.size(-2)
|
54
|
+
# causal_mask = self.bias[:, :, key_length - query_length : key_length, :key_length]
|
55
|
+
# mask_value = torch.finfo(attn_weights.dtype).min
|
56
|
+
# # Need to be a tensor, otherwise we get error: `RuntimeError: expected scalar type float but found double`.
|
57
|
+
# # Need to be on the same device, otherwise `RuntimeError: ..., x and y to be on the same device`
|
58
|
+
# mask_value = torch.full([], mask_value, dtype=attn_weights.dtype, device=attn_weights.device)
|
59
|
+
# attn_weights = torch.where(causal_mask, attn_weights.to(attn_weights.dtype), mask_value)
|
60
|
+
|
61
|
+
if attention_mask is not None:
|
62
|
+
# Apply the attention mask
|
63
|
+
attn_weights = attn_weights + attention_mask
|
64
|
+
|
65
|
+
attn_weights = nn.functional.softmax(attn_weights, dim=-1)
|
66
|
+
|
67
|
+
# Downcast (if necessary) back to V's dtype (if in mixed-precision) -- No-Op otherwise
|
68
|
+
attn_weights = attn_weights.type(value.dtype)
|
69
|
+
# attn_weights = self.attn_dropout(attn_weights)
|
70
|
+
|
71
|
+
# Mask heads if we want to
|
72
|
+
if head_mask is not None:
|
73
|
+
attn_weights = attn_weights * head_mask
|
74
|
+
|
75
|
+
attn_output = torch.matmul(attn_weights, value)
|
76
|
+
|
77
|
+
return attn_output, attn_weights
|
78
|
+
|
79
|
+
def forward(
|
80
|
+
self,
|
81
|
+
hidden_states: Optional[Tuple[torch.FloatTensor]],
|
82
|
+
layer_past: Optional[Tuple[torch.Tensor]] = None,
|
83
|
+
attention_mask: Optional[torch.FloatTensor] = None,
|
84
|
+
head_mask: Optional[torch.FloatTensor] = None,
|
85
|
+
encoder_hidden_states: Optional[torch.Tensor] = None,
|
86
|
+
encoder_attention_mask: Optional[torch.FloatTensor] = None,
|
87
|
+
use_cache: Optional[bool] = False,
|
88
|
+
output_attentions: Optional[bool] = False,
|
89
|
+
cache_position: Optional[torch.LongTensor] = None,
|
90
|
+
) -> Tuple[Union[torch.Tensor, Tuple[torch.Tensor]], ...]:
|
91
|
+
query, key, value = self.c_attn(hidden_states).split(self.split_size, dim=2)
|
92
|
+
|
93
|
+
query = self._split_heads(query, self.num_heads, self.head_dim)
|
94
|
+
key = self._split_heads(key, self.num_heads, self.head_dim)
|
95
|
+
value = self._split_heads(value, self.num_heads, self.head_dim)
|
96
|
+
|
97
|
+
if layer_past is not None:
|
98
|
+
past_key, past_value = layer_past
|
99
|
+
query_length = query.shape[-2]
|
100
|
+
|
101
|
+
key = torch.slice_scatter(past_key, key, dim=2, start=cache_position, end=cache_position + query_length)
|
102
|
+
value = torch.slice_scatter(
|
103
|
+
past_value, value, dim=2, start=cache_position, end=cache_position + query_length
|
104
|
+
)
|
105
|
+
|
106
|
+
present = (key, value)
|
107
|
+
attn_output, _ = _GPT2Attention._attn(self, query, key, value, attention_mask, head_mask)
|
108
|
+
|
109
|
+
attn_output = self._merge_heads(attn_output, self.num_heads, self.head_dim)
|
110
|
+
|
111
|
+
attn_output = self.c_proj(attn_output)
|
112
|
+
attn_output = self.resid_dropout(attn_output)
|
113
|
+
|
114
|
+
outputs = (attn_output, present)
|
115
|
+
|
116
|
+
return outputs
|
117
|
+
|
118
|
+
|
119
|
+
class _GPT2Block(GPT2Block):
|
120
|
+
def forward(
|
121
|
+
self,
|
122
|
+
hidden_states: Optional[Tuple[torch.FloatTensor]],
|
123
|
+
layer_past: Optional[Tuple[torch.Tensor]] = None,
|
124
|
+
attention_mask: Optional[torch.FloatTensor] = None,
|
125
|
+
head_mask: Optional[torch.FloatTensor] = None,
|
126
|
+
encoder_hidden_states: Optional[torch.Tensor] = None,
|
127
|
+
encoder_attention_mask: Optional[torch.FloatTensor] = None,
|
128
|
+
use_cache: Optional[bool] = False,
|
129
|
+
output_attentions: Optional[bool] = False,
|
130
|
+
cache_position: Optional[torch.LongTensor] = None,
|
131
|
+
) -> Union[Tuple[torch.Tensor], Optional[Tuple[torch.Tensor, Tuple[torch.FloatTensor, ...]]]]:
|
132
|
+
residual = hidden_states
|
133
|
+
hidden_states = self.ln_1(hidden_states)
|
134
|
+
|
135
|
+
attn_outputs = _GPT2Attention.forward(
|
136
|
+
self.attn,
|
137
|
+
hidden_states,
|
138
|
+
layer_past=layer_past,
|
139
|
+
attention_mask=attention_mask,
|
140
|
+
head_mask=head_mask,
|
141
|
+
cache_position=cache_position,
|
142
|
+
)
|
143
|
+
|
144
|
+
attn_output = attn_outputs[0] # output_attn: a, present, (attentions)
|
145
|
+
outputs = attn_outputs[1:]
|
146
|
+
# residual connection
|
147
|
+
hidden_states = attn_output + residual
|
148
|
+
|
149
|
+
residual = hidden_states
|
150
|
+
hidden_states = self.ln_2(hidden_states)
|
151
|
+
feed_forward_hidden_states = self.mlp(hidden_states)
|
152
|
+
# residual connection
|
153
|
+
hidden_states = residual + feed_forward_hidden_states
|
154
|
+
|
155
|
+
outputs = (hidden_states,) + outputs
|
156
|
+
return outputs # hidden_states, present, (attentions, cross_attentions)
|
157
|
+
|
158
|
+
|
159
|
+
class _GPT2Model(GPT2Model):
|
160
|
+
def forward(
|
161
|
+
self,
|
162
|
+
input_ids: Optional[torch.LongTensor] = None,
|
163
|
+
past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
|
164
|
+
attention_mask: Optional[torch.FloatTensor] = None,
|
165
|
+
position_ids: Optional[torch.LongTensor] = None,
|
166
|
+
head_mask: Optional[torch.FloatTensor] = None,
|
167
|
+
cache_position: Optional[torch.LongTensor] = None,
|
168
|
+
) -> Union[Tuple, BaseModelOutputWithPastAndCrossAttentions]:
|
169
|
+
input_shape = input_ids.size()
|
170
|
+
|
171
|
+
if position_ids is None:
|
172
|
+
# force dtype to torch.long -> torch.int32 (to match cache_position)
|
173
|
+
position_ids = torch.arange(0, input_shape[-1], dtype=torch.int32) + cache_position
|
174
|
+
position_ids = position_ids.unsqueeze(0)
|
175
|
+
|
176
|
+
# GPT2Attention mask.
|
177
|
+
# Here we assume mask is causal mask, (batch, 1, query_length, key_length + query_length)
|
178
|
+
attention_mask = (1.0 - attention_mask) * torch.finfo(self.dtype).min
|
179
|
+
|
180
|
+
# Prepare head mask if needed
|
181
|
+
# 1.0 in head_mask indicate we keep the head
|
182
|
+
# attention_probs has shape bsz x n_heads x N x N
|
183
|
+
# head_mask has shape n_layer x batch x n_heads x N x N
|
184
|
+
head_mask = self.get_head_mask(head_mask, self.config.n_layer)
|
185
|
+
|
186
|
+
inputs_embeds = self.wte(input_ids)
|
187
|
+
position_embeds = self.wpe(position_ids)
|
188
|
+
hidden_states = inputs_embeds + position_embeds
|
189
|
+
|
190
|
+
hidden_states = self.drop(hidden_states)
|
191
|
+
|
192
|
+
output_shape = (-1,) + input_shape[1:] + (hidden_states.size(-1),)
|
193
|
+
|
194
|
+
presents = ()
|
195
|
+
for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
|
196
|
+
outputs = _GPT2Block.forward(
|
197
|
+
block,
|
198
|
+
hidden_states,
|
199
|
+
layer_past=layer_past,
|
200
|
+
attention_mask=attention_mask,
|
201
|
+
head_mask=head_mask[i],
|
202
|
+
cache_position=cache_position,
|
203
|
+
)
|
204
|
+
hidden_states = outputs[0]
|
205
|
+
|
206
|
+
presents = presents + (outputs[1],)
|
207
|
+
|
208
|
+
hidden_states = self.ln_f(hidden_states)
|
209
|
+
hidden_states = hidden_states.view(output_shape)
|
210
|
+
return BaseModelOutputWithPast(last_hidden_state=hidden_states, past_key_values=presents)
|
211
|
+
|
212
|
+
|
213
|
+
class GPT2LMHeadModelWrapper(torch.nn.Module):
|
214
|
+
def __init__(self, gpt):
|
215
|
+
super().__init__()
|
216
|
+
self.model = gpt
|
217
|
+
|
218
|
+
def forward(
|
219
|
+
self,
|
220
|
+
input_ids: torch.Tensor,
|
221
|
+
past_key_values: torch.Tensor,
|
222
|
+
attention_mask: torch.Tensor,
|
223
|
+
cache_position: torch.LongTensor,
|
224
|
+
):
|
225
|
+
kv_cache = []
|
226
|
+
for i in range(self.model.config.n_layer):
|
227
|
+
kv_cache.append((past_key_values[i, 0], past_key_values[i, 1]))
|
228
|
+
|
229
|
+
transformer_outputs = _GPT2Model.forward(
|
230
|
+
self.model.transformer,
|
231
|
+
input_ids=input_ids,
|
232
|
+
past_key_values=kv_cache,
|
233
|
+
attention_mask=attention_mask,
|
234
|
+
cache_position=cache_position,
|
235
|
+
)
|
236
|
+
|
237
|
+
hidden_states = transformer_outputs[0]
|
238
|
+
|
239
|
+
# TODO : Use query_length here to pick last logit
|
240
|
+
# batch_size, sequence_length = hidden_states.shape[:2]
|
241
|
+
# hidden_states = hidden_states.view(batch_size * sequence_length, -1)
|
242
|
+
# hidden_states = torch.nn.functional.embedding(query_length, hidden_states)
|
243
|
+
# hidden_states = hidden_states.view(batch_size, 1, -1)
|
244
|
+
|
245
|
+
lm_logits = self.model.lm_head(hidden_states)
|
246
|
+
kv_cache = transformer_outputs[1]
|
247
|
+
|
248
|
+
past_key_values = []
|
249
|
+
for i in range(self.model.config.n_layer):
|
250
|
+
past_key_values.append(torch.stack(kv_cache[i], dim=0))
|
251
|
+
past_key_values = torch.stack(past_key_values, dim=0)
|
252
|
+
|
253
|
+
return lm_logits, past_key_values
|