hehuan710

https://github.com/jacobgil/keras-cam/blob/2b7ada2c5c819808a174444e82622315a07fa11e/cam.py#L34 Should this be the case here? **for i, w in enumerate(class_weights[:,np.argmax(predictions)]):**

https://github.com/albertogaspar/dts/blob/f10ab160d620486397bb799cacd11592bb5734b0/dts/models/TCN.py#L241 Whether it should be corrected like this ： “if inputs.shape[-1] == 1:”

Incomplete information read by cfgrib

3

### **Cfgrib can't read the time dimension, but another software can**   [ER03_202104022000_07203.zip](https://github.com/ecmwf/cfgrib/files/6249702/ER03_202104022000_07203.zip)

您好，我注意到输入数据的最后一维是1，那就既可以使用CNN1d又可以使用CNN2d，是使用CNN1d准确率下降了吗？

bz2直接读取问题

1

请问可以直接读取bz2中的grib文件吗

为None时正常运行 grid0 = meb.interp_sg_idw(sta,grid0,nearNum = 4,background=0,effectR=20) C:\ProgramData\Anaconda3\lib\site-packages\meteva\base\fun\interpolating.py in interp_gg_linear(grd, grid, used_coords, outer_value) 460 if (grd is None): 461 return None --> 462 levels = grd["level"].values 463 times = grd["time"].values 464...

module 'meteva' has no attribute 'base'

2

import meteva 时报错