Looks like you are trying to predict ORFs within assembled transcripts (that is what TransDecoder does). If you successfully ran TransDecoder.LongOrfs -t target_transcripts.fasta, then TransDecoder produces a few files, including longest_orfs.pep which already has the amino acid sequences of the predicted ORFs. But if you just want to translate coding sequences to proteins, use something like EMBOSS transeq.