ser推理模型推理图片时,使用cpu推理,半精度推理,报错 #12748
Unanswered
xiaoxianyu12
asked this question in
Q&A
Replies: 1 comment
-
看着像 |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem
--kie_algorithm=LayoutXLM
--ser_model_dir=../inference/ser_vi_layoutxlm/HRA2103COA_TLZ-100_UV928_6200EL_UV928RIANLON
--image_dir=../test_data/testSpeed/
--ser_dict_path=../train_data/HRA2103COA_TLZ-100_UV928_6200EL_UV928RIANLON.txt
--vis_font_path=../doc/fonts/simfang.ttf
--ocr_order_method="tb-yx"
--output=../outputInfer/ser/HRA2103COA_TLZ-100_UV928_6200EL_UV928RIANLON/testSpeed
--det_model_dir=../inference/ch_PP-OCRv4_det_server_infer
--rec_model_dir=../inference/ch_PP-OCRv4_rec_server_infer
--use_gpu =False
--use_tensorrt=True
--precision='fp16'
--- marked 696 operators to bfloat16
--- added 498 quantize ops before bfloat16 op
--- added 374 dequantize ops after bfloat16 op
--- squashed 0 scale with dequantize op
--- squashed 0 scale with quantize op
--- squashed 408 dequantize-quantize pairs
--- squashed 0 requantize ops
--- squashed 0 requantize ops
--- squashed 1 dequant with ops
--- squashed 0 quantize op
--- squashed 0 quantize with bfloat16 conv2d op
--- squashed 0 quantize with bfloat16 fused_conv2d op
Traceback (most recent call last):
File "kie/predict_kie_token_ser.py", line 188, in
main(parse_args())
File "kie/predict_kie_token_ser.py", line 156, in main
ser_res, _, elapse = ser_predictor(img)
File "kie/predict_kie_token_ser.py", line 120, in call
self.predictor.run()
NotImplementedError: (Unimplemented) There are no kernels which are registered in the dequantize operator.
[Hint: Expected kernels_iter != all_op_kernels.end(), but received kernels_iter == all_op_kernels.end().] (at ../paddle/fluid/framework/operator.cc:2273)
[operator < dequantize > error]
Beta Was this translation helpful? Give feedback.
All reactions