-
Notifications
You must be signed in to change notification settings - Fork 0
/
quantization.py
285 lines (253 loc) · 10.2 KB
/
quantization.py
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
import os
import torch
from torch.ao.quantization import get_default_qconfig
from torch.ao.quantization.quantize_fx import prepare_fx, convert_fx
from torch.ao.quantization import get_default_qconfig_mapping
import argparse
from argparse import Namespace
from utils.args_utils import str2list, str2bool
import random
from time import time
import json
from torch.ao.quantization import QConfigMapping
from data.vizwiz_dataset import VizWizDataset
from data.vizwiz_dataloader import VizWizDataLoader
from models.End_ExpansionNet_v2 import (
End_ExpansionNet_v2_Encoder,
End_ExpansionNet_v2_Decoder,
E2E_ExpansionNet_Captioner
)
from utils import language_utils
from utils.language_utils import compute_num_pads, tokens2description
from utils.image_utils import preprocess_image
from utils.quantization_utils import (
calibrate_enc_dec,
prepare_model,
quantize_model,
quantize_encoder_decoder,
print_size_of_model
)
encoder_modules = [
"swin_transf",
"encoders",
"input_embedder_dropout",
"input_linear",
"out_enc_dropout",
"pos_encoder",
"enc_reduce_group",
"enc_reduce_norm",
"out_embedder",
]
decoder_modules = [
"decoders",
"log_softmax",
"out_embedder",
"out_dec_dropout",
"dec_reduce_group",
"pos_encoder",
"dec_reduce_group",
"dec_reduce_norm",
"vocab_linear",
]
def filter_state_dict(state_dict, list_to_include):
new_state_dict = {}
for key in state_dict.keys():
valid_key = False
for entry in list_to_include:
if entry in key:
valid_key = True
if valid_key:
new_state_dict[key] = state_dict[key]
return new_state_dict
def load_models(
model_args: Namespace,
dataset,
model_max_len: int,
img_size: int = 384,
device: str = "cpu",
):
encoder_model = End_ExpansionNet_v2_Encoder(
swin_img_size=img_size,
swin_patch_size=4,
swin_in_chans=3,
swin_embed_dim=192,
swin_depths=[2, 2, 18, 2],
swin_num_heads=[6, 12, 24, 48],
swin_window_size=12,
swin_mlp_ratio=4.0,
swin_qkv_bias=True,
swin_qk_scale=None,
swin_drop_rate=0.0,
swin_attn_drop_rate=0.0,
swin_drop_path_rate=0.1,
swin_norm_layer=torch.nn.LayerNorm,
swin_ape=False,
swin_patch_norm=True,
swin_use_checkpoint=False,
final_swin_dim=1536,
d_model=model_args.model_dim,
N_enc=model_args.N_enc,
N_dec=model_args.N_dec,
num_heads=8,
ff=2048,
num_exp_enc_list=[32, 64, 128, 256, 512],
num_exp_dec=16,
output_word2idx=dataset.caption_word2idx_dict,
output_idx2word=dataset.caption_idx2word_list,
max_seq_len=model_max_len,
drop_args=model_args.drop_args,
rank=device,
)
decoder_model = End_ExpansionNet_v2_Decoder(
d_model=512,
N_enc=3,
N_dec=3,
num_heads=8,
ff=2048,
num_exp_enc_list=[32, 64, 128, 256, 512],
num_exp_dec=16,
output_word2idx=dataset.caption_word2idx_dict,
output_idx2word=dataset.caption_idx2word_list,
max_seq_len=model_max_len,
drop_args=drop_args,
rank=device,
)
return encoder_model, decoder_model
def demo_quantized_model(encoder, decoder, sos_idx, eos_idx, device="cpu"):
demo_image_path = "./demo_material/micheal.jpg"
demo_image = preprocess_image(demo_image_path, img_size)
encoder.to(device)
decoder.to(device)
captioner = E2E_ExpansionNet_Captioner(beam_search_arg_defaults, split_encoder=True, encoder=encoder,
decoder=decoder, rank=device)
with torch.no_grad():
pred, _ = captioner(enc_x=demo_image.to(device),
enc_x_num_pads=[0], mode="beam_search")
pred = tokens2description(pred[0][0], dataset.caption_idx2word_list, sos_idx, eos_idx)
print(' \n\tDescription: ' + pred + '\n')
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="Image Captioning")
parser.add_argument("--model_dim", type=int, default=512)
parser.add_argument("--N_enc", type=int, default=3)
parser.add_argument("--N_dec", type=int, default=3)
parser.add_argument(
"--save_model_path",
type=str,
default="/usr0/home/nvaikunt/On_Device_Image_Captioning/pretrained_weights/rf_model.pth",
)
parser.add_argument("--eval_beam_sizes", type=str2list, default=[3])
parser.add_argument("--image_folder", type=str, default="./VizWizData")
parser.add_argument(
"--vocab_path", type=str, default="./vocab/coco_vocab_idx_dict.json"
)
# parser.add_argument('--pretrain_checkpoint', type=str, default="/home/arpitsah/Desktop/Fall-2023/odml/On_Device_Image_Captioning/pretrained_weightscheckpoint_2023-10-12-13:36:34_epoch4it1968bs8_xe_.pth")
parser.add_argument("--vizwiz", type=str2bool, default=True)
parser.add_argument("--batch_size", type=int, default=1)
parser.add_argument("--num_accum", type=int, default=1)
parser.add_argument("--device", type=str, default="cpu")
parser.add_argument(
"--save_path",
type=str,
default="/usr0/home/nvaikunt/On_Device_Image_Captioning/pretrained_weights",
) # default='./github_ignore_material/saves/')
parser.add_argument(
"--static", type=str2bool, default=True
) # default='./github_ignore_material/saves/')
parser.add_argument(
"--qat", type=str2bool, default=False
) # default='./github_ignore_material/saves/')
parser.add_argument("--calibration_steps", type=int, default=1000)
parser.add_argument("--static_qconfig", type=str, default="x86")
parser.add_argument("--demo", type=str2bool, default=False)
args = parser.parse_args()
print("save_model_path: " + str(args.save_model_path))
drop_args = Namespace(enc=0.0, dec=0.0, enc_input=0.0, dec_input=0.0, other=0.0)
model_args = Namespace(
model_dim=args.model_dim,
N_enc=args.N_enc,
N_dec=args.N_dec,
dropout=0.0,
drop_args=drop_args,
vizwiz=args.vizwiz,
image_folder=args.image_folder,
)
quant_args = Namespace(
static=args.static,
calibration_steps=args.calibration_steps,
static_qconfig_str=args.static_qconfig,
)
dataset = None
if args.vizwiz:
if os.path.isfile(args.vocab_path):
with open("./vocab/coco_vocab_idx_dict.json", "r") as vocab_json:
coco_vocab_idx_dict = json.load(vocab_json)
else:
coco_vocab_idx_dict = None
# Currently testing with val_split, normally should set to 1 with train being True
split = 2
dataset = VizWizDataset(
split,
train=False,
val=True,
coco_vocab_dict=coco_vocab_idx_dict,
vizwiz_annotations_dir="/usr0/home/nvaikunt/On_Device_Image_Captioning/VizWizData/annotations",
)
ckpt_path = args.save_model_path
checkpoint = torch.load(ckpt_path, map_location=args.device)
state_dict = checkpoint["model_state_dict"]
del checkpoint
model_max_len = dataset.max_seq_len + 20
img_size = 384
# device = "cpu"
device = args.device
beam_search_arg_defaults = {'sos_idx': dataset.get_sos_token_idx(),
'eos_idx': dataset.get_eos_token_idx(),
'beam_size': 5,
'beam_max_seq_len': model_max_len,
'sample_or_max': 'max',
'how_many_outputs': 1, }
encoder_model, decoder_model = load_models(model_args, dataset,
model_max_len, img_size=img_size, device=device)
encoder_state_dict = filter_state_dict(state_dict, encoder_modules)
decoder_state_dict = filter_state_dict(state_dict, decoder_modules)
encoder_model.load_state_dict(encoder_state_dict)
decoder_model.load_state_dict(decoder_state_dict)
encoder_model.to(device)
decoder_model.to(device)
image_folder = args.image_folder
array_of_init_seeds = [random.random() for _ in range(1 * 2)]
data_loader = VizWizDataLoader(vizwiz_dataset=dataset,
batch_size=4,
num_procs=1,
array_of_init_seeds=array_of_init_seeds,
dataloader_mode='caption_wise',
resize_image_size=img_size,
rank=device,
image_folder=image_folder,
verbose=True)
if quant_args.static:
static_qconfig_str = quant_args.static_qconfig_str
qconfig_mapping = get_default_qconfig_mapping(static_qconfig_str)
else:
qconfig_mapping = QConfigMapping().set_global(torch.ao.quantization.default_dynamic_qconfig)
quantized_encoder, quantized_decoder = quantize_encoder_decoder(encoder_model, decoder_model,
data_loader, 3, qconfig_mapping, device,
static=quant_args.static)
# Save models
orig_file_name = ckpt_path.split("/")[-1]
if args.static:
model_type = "static"
else:
model_type = "dynamic"
encoder_save_file = f"{model_type}_quantized_encoder_{orig_file_name}"
decoder_save_file = f"{model_type}_quantized_decoder_{orig_file_name}"
torch.save(quantized_encoder.state_dict(), os.path.join(args.save_path, encoder_save_file))
torch.save(quantized_decoder.state_dict(), os.path.join(args.save_path, decoder_save_file))
# Print Info
print_size_of_model(encoder_model)
print_size_of_model(decoder_model)
print_size_of_model(quantized_encoder)
print_size_of_model(quantized_decoder)
if args.demo:
demo_quantized_model(quantized_encoder, quantized_decoder,
sos_idx=dataset.get_sos_token_idx(), eos_idx=dataset.get_eos_token_idx())