FilterConsensusReads results too little reads compared raw fastq OR uBAM file

[teryyoung]

hello,
I runned fgbio's FilterConsensusReads to my RNAseq data, but the output results files are of too little size.
The raw read uBAM file had 3-4GB,
after first mapping and quality control, there had 2GB data,
this is one of the tag family size histogram file.

my issue is this, after I CallConsensusReads and FilterConsensusReads, is seems most of reads were filtered out, just 38kB data

I tried increase the umi from 5 to 6, and it didn't work, if someone know how to sovle this problem?