File size: 6,107 Bytes
95dfa6c |
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 |
#!/usr/bin/env python
# coding: utf-8
# In[1]:
import os
import urllib
import traceback
import time
import sys
import numpy as np
import cv2
from rknn.api import RKNN
from math import exp
from sys import exit
batch_size = 1
# embed_seq_len = 590
vision_size = (512, 512)
vision_tokens = 257
prompt_tokens = 14
encoder_seq_len = vision_tokens + prompt_tokens
decoder_seq_len = 1
def convert_decoder():
rknn = RKNN(verbose=True)
ONNX_MODEL="decoder_model.onnx"
RKNN_MODEL=ONNX_MODEL.replace(".onnx",".rknn")
DATASET="dataset.txt"
QUANTIZE=False
# pre-process config
print('--> Config model')
rknn.config(quantized_algorithm='normal', quantized_method='channel', target_platform='rk3588', optimization_level=3, single_core_mode=True )
print('done')
# Load ONNX model
print('--> Loading model')
ret = rknn.load_onnx(model=ONNX_MODEL,
inputs=["encoder_attention_mask",
"encoder_hidden_states",
"inputs_embeds",
],
input_size_list=[[batch_size, encoder_seq_len],
[batch_size, encoder_seq_len, 768],
[batch_size, decoder_seq_len, 768]],
)
if ret != 0:
print('Load model failed!')
exit(ret)
print('done')
# Build model
print('--> Building model')
ret = rknn.build(do_quantization=QUANTIZE, dataset=DATASET, rknn_batch_size=None)
if ret != 0:
print('Build model failed!')
exit(ret)
print('done')
#export
print('--> Export RKNN model')
ret = rknn.export_rknn(RKNN_MODEL)
if ret != 0:
print('Export RKNN model failed!')
exit(ret)
print('done')
def convert_encoder():
rknn = RKNN(verbose=True)
ONNX_MODEL="encoder_model.onnx"
RKNN_MODEL=ONNX_MODEL.replace(".onnx",".rknn")
DATASET="dataset.txt"
QUANTIZE=False
# pre-process config
print('--> Config model')
rknn.config(quantized_algorithm='normal', quantized_method='channel', target_platform='rk3588', optimization_level=3, single_core_mode=True )
print('done')
# Load ONNX model
print('--> Loading model')
ret = rknn.load_onnx(model=ONNX_MODEL,
inputs=["attention_mask", "inputs_embeds"],
input_size_list=[[batch_size, encoder_seq_len], [batch_size, encoder_seq_len, 768]],
)
if ret != 0:
print('Load model failed!')
exit(ret)
print('done')
# Build model
print('--> Building model')
ret = rknn.build(do_quantization=QUANTIZE, dataset=DATASET, rknn_batch_size=None)
if ret != 0:
print('Build model failed!')
exit(ret)
print('done')
# Export RKNN model
print('--> Export RKNN model')
ret = rknn.export_rknn(RKNN_MODEL)
if ret != 0:
print('Export RKNN model failed!')
exit(ret)
print('done')
def convert_embed():
rknn = RKNN(verbose=True)
ONNX_MODEL="embed_tokens.onnx"
RKNN_MODEL=ONNX_MODEL.replace(".onnx",".rknn")
DATASET="dataset.txt"
QUANTIZE=False
# pre-process config
print('--> Config model')
rknn.config(quantized_algorithm='normal', quantized_method='channel', target_platform='rk3588', optimization_level=3, single_core_mode=True )
print('done')
# Load ONNX model
print('--> Loading model')
ret = rknn.load_onnx(model=ONNX_MODEL,
inputs=["input_ids"],
input_size_list=[[batch_size, embed_seq_len]],
)
if ret != 0:
print('Load model failed!')
exit(ret)
print('done')
# Build model
print('--> Building model')
ret = rknn.build(do_quantization=QUANTIZE, dataset=DATASET, rknn_batch_size=None)
if ret != 0:
print('Build model failed!')
exit(ret)
print('done')
# Export RKNN model
print('--> Export RKNN model')
ret = rknn.export_rknn(RKNN_MODEL)
if ret != 0:
print('Export RKNN model failed!')
exit(ret)
print('done')
def convert_vision():
rknn = RKNN(verbose=True)
ONNX_MODEL="vision_encoder.onnx"
RKNN_MODEL=ONNX_MODEL.replace(".onnx",".rknn")
DATASET="dataset.txt"
QUANTIZE=False
# pre-process config
print('--> Config model')
rknn.config(quantized_algorithm='normal', quantized_method='channel', target_platform='rk3588', optimization_level=3, single_core_mode=True )
print('done')
# Load ONNX model
print('--> Loading model')
ret = rknn.load_onnx(model=ONNX_MODEL,
inputs=["pixel_values"],
input_size_list=[[batch_size, 3, vision_size[0], vision_size[1]]],
)
if ret != 0:
print('Load model failed!')
exit(ret)
print('done')
# Build model
print('--> Building model')
ret = rknn.build(do_quantization=QUANTIZE, dataset=DATASET, rknn_batch_size=None)
if ret != 0:
print('Build model failed!')
exit(ret)
print('done')
# Export RKNN model
print('--> Export RKNN model')
ret = rknn.export_rknn(RKNN_MODEL)
if ret != 0:
print('Export RKNN model failed!')
exit(ret)
print('done')
import argparse
# python convert.py <decoder|encoder|vision|all>
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("model", type=str, help="Model to convert")
args = parser.parse_args()
if args.model == "decoder":
convert_decoder()
elif args.model == "encoder":
convert_encoder()
# elif args.model == "embed": # embed is faster with cpu
# convert_embed()
elif args.model == "vision":
convert_vision()
elif args.model == "all":
convert_decoder()
convert_encoder()
# convert_embed()
convert_vision()
else:
print("Invalid model")
exit(1) |