Chris Seymour

Is it possible that a none hdf5 file could be named `*.fast5` under your directory structure or are you using any unusual characters in the filenames?

The filenames look okay - what filesystem are you using?

Can you run `stat --file-system --format=%T .` in directory with the fast5 in.

Can you try with hdf5 file locking disabled - ```bash $ export HDF5_USE_FILE_LOCKING=FALSE $ bonito basecaller ... ```

Yes, that is right @HaoDreamlong

@lpryszcz is correct - `ctc-crf` models (v3+) don't output a probability distribution over the alphabet.

Hey @danrdanny - have you also compared aligned lengths?

@danrdanny @gallardo-seq I've failed to find datasets to reproduce this so far but I do have some new models which may help. Are either of you able to test these...

A fast5 file would really help thanks - https://nanoporetech.ent.box.com/f/7c4375e2b71b48258ebed29f198b89ab

@Sumsarium @callumparr models pushed to rerio this morning.