leizhu1989

> 您好，我想请教下，我直接跑demo，用的是efficient b0，loss下降到30左右就降不下去了，不知这是不是正常的情况？ 你好，请问你训练多少个epoch，loss下降明显的，我训练50个epoch，loss依然很高，用的是300W数据集： Epoch:[50 ][26 /26 ] Loss: 23002.686 lr 1e-06 average_time:3.894s remain_time:12.627h Test epochs: 5 Loss 46399478.200 mean error and failure rate mean error : 1.664 failure rate:...

thank you ,I should make clear my question. I want to know whether the featuremap of the last layer of convolutional output can be directly drawn into an density maps...

thank you ，I got it

I have sovled the problem ,it can be trained now

and cudnn 7.5

I have solve the problem,it should be add cufft to cuda library

I should read all files carefully, thank you for your reply

> hi , 兄dei 以下是我寫的腳本能清理wiki資料成, train.json 要的形式，希望能幫上忙。 它只紀錄前10筆, 看個人需求記得做更改。 > > ``` > import glob > import json > > names =glob.glob("./data/wiki_zh/*/*") > for i, name in enumerate(names): > articles=[]...

@steven8274 hello! may I ask how to understand the "All signals used as input to the model are subject to a random gainchosen from a uniform distribution ranging from -25...

@steven8274 thank you for you reply , I still don't quite understand what signal to add，example： import numpy as np min_gain_dB = -25 max_gain_dB = 0 random_gain_dB = np.random.uniform(min_gain_dB, max_gain_dB)...