Forums - quant 16 bitwidth faild

7 posts / 0 new
Last post
quant 16 bitwidth faild
MarkWu
Join Date: 8 May 22
Posts: 37
Posted: Tue, 2023-06-06 00:59

Hi,

I read SNPE-SDK doc and found this sentence:

[ --bitwidth=<val> ]
                        Use the --bitwidth option to select the bitwidth to use when quantizing the weights/activation/bias, either 8 (default) or 16. Can't mix with
                        --weights_bitwidth or --act_bitwidth or --bias_bitdwith

But when I used the quantization command below, I reported an error

command : snpe-dlc-quant --input_dlc=xxx.dlc --input_list=xxx.txt --output_dlc=xxx.dlc --bitwidth=16

ERROR INFO : IrQuantizer: Unsupported weight bitwidth: 16

BR.

  • Up0
  • Down0
yunxqin
Join Date: 2 Mar 23
Posts: 44
Posted: Tue, 2023-06-06 20:28

Dear developer,

  • Your snpe version is not supported bitwidth=8.

BR.

 

Yunxiang

  • Up0
  • Down0
MarkWu
Join Date: 8 May 22
Posts: 37
Posted: Wed, 2023-06-07 20:36

Dear Yunxiang

I use --bitwidth=16 because the 8-bit quantization model has insufficient accuracy.

uesd snpe version : snpe-2.10.0.4541

BR.

  • Up0
  • Down0
yunxqin
Join Date: 2 Mar 23
Posts: 44
Posted: Thu, 2023-06-08 00:17

Dear developer,

You can use the commands  '--skip_quantization  -enabel_htp' to improve the accuracy. And it's for fp16 quantization.

Br.

Yunxiang

  • Up0
  • Down0
MarkWu
Join Date: 8 May 22
Posts: 37
Posted: Mon, 2023-06-12 01:49

Dear Yunxiang

I quantified the model according to your method, but failed to load the quantified model on the Snapdragon 888 chip. How can I solve this?

error message: 

error_code=1002; error_message=Layer parameter value is invalid. error_code=1002; error_message=Layer parameter value is invalid. No backend could validate Op=Conv_0 Type=Conv2d error code=3110; error_component=Model Validation; line_no=662; thread_id=526580776192; error_component=Model Validation; line_no=272; thread_id=539432203520

SNPE version: snpe-2.10.0.4541

BR.

  • Up0
  • Down0
yunxqin
Join Date: 2 Mar 23
Posts: 44
Posted: Mon, 2023-06-12 20:01

Dear developer,

Would you mind providing the model you used? You can upload the model to git and I will run it on Snapdragon 888 .

Br.

Yunxiang

  • Up0
  • Down0
MarkWu
Join Date: 8 May 22
Posts: 37
Posted: Tue, 2023-06-13 03:14

Dear Yunxiang

sorry, i can't providing the model.

The type of model is object detection.

BR.

 

  • Up0
  • Down0
or Register

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries (“Qualcomm”). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.