No training configuration found in save file

[YS-Github-bug]

i met a problem in installing this software.
I first create a conda environment using python=3.6.
Then I pip install packages:pandas==0.24.2 keras==2.0.5 tensorflow==1.9.0 h5py==2.10.0
Following this, pip install spliceai and also run conda install mkl-service

i try to test the install succeed, using "spliceai -I input.vcf -O output.vcf -R hg38_p14.fa -A grch38".

I get the following errors and didn't get expected output. So that i didn't get any predicted values.

UserWarning: No training configuration found in save file: the model was not compiled. Compile it manually.
warnings.warn('No training configuration found in save file: '
WARNING:root:Skipping record (ref issue): 2 152389953 . T A,C,G . . .

WARNING:root:Skipping record (ref issue): 2 179446218 . ATACT A . . .

WARNING:root:Skipping record (ref issue): 2 179446218 . ATACT AT,ATA . . .

WARNING:root:Skipping record (ref issue): 2 179642185 . G A . . .

WARNING:root:Skipping record (ref issue): 19 38958362 . C T . . .

Do you have any good solutions?

[kishorejaganathan]

Skipping record (ref issue) happens when the reference nucleotide provided in the VCF does not match with what is inferred from the fasta file. For example, in your fasta file, 2:152389953 does not have T (I have checked this in the genome browser and it is an A). It is possible your VCF is hg19.