Mike Yi SUN

PaddleClas运行特征提取示例报错。

2

环境： PaddlePaddle docker：2.3.2-gpu-cuda11.2-cudnn8 PaddleClas branch: release/2.5, installed from repo by `python setup.py install` 问题描述： 运行特征提取示例报错。 运行命令： ```bash cd deploy # Under /pth/to/PaddleClas/deploy/ python python/predict_rec.py -c configs/inference_rec.yaml -o Global.rec_inference_model_dir="../../deploy_cus/models/PPLCNetV2_base_ssld_infer ``` 模型文件确认无误。...

Any chance to release your trained models (checkpoints)?

8

Hi jungomi, I'm also trying to implement math formula recognition in PyTorch. Currently I'm using Lua+Torch for img2markup project and PyTorch for Show, Attend, and Tell. They have similar performance...

Different behavior on RV1126 by export NN_LAYER_DUMP=1

1

## Problem description We try to evaluate our int8 .rknn model on RV1126 from rknn_toolkit 1.7.1 on Ubuntu 18.04. The model (attached below) fails to run on RV1126 and the...

I tried to evaluate a .rknn model onto my RV1126 with rknn_toolkit 1.7.1 from Ubuntu 18.04 but encountered a NPUTransfer failed error. The error log is as follows. Extra information:...