"Erratic Equulites"

v5.0.0

*Upgrade to Pytorch version 1.2

• Improved training stability: gradient capping and warm-up
• Merged mod-base and canonical entry points
  • Custom model definitions should now take an
    alphabet_info argument rather than outsize
• Improved RNA support: tools can reverse references and basecalls
• Basecaller changes:
  • chunk size argument now matches guppy
  • CPU calling enabled
  • lower memory usage
• Multi-GPU training enabled
• Bug fixes

“Delinquent Demoiselle"

• Alternative "ab initio” entry point to training.
• Examples
  • R9.4.1 DNA
  • R9.4.1 RNA
  • R10 RNA
• Bug fixes

"Crabby Clarias"

Training and calling of modified bases.

“Beleaguered Brevoortia"

Since v3.0.1

v3.1.0

• Basecaller script that uses GPU
• Training walk-through
• Tweaks to optimisation parameters

v3.0.2

• Improved training parameters
• Use orthonormal initialisation of starting weights

“Angsty Acanthonus” point release 1

Improve and simplify how version information is stored.

"Angsty Acanthonus"

Initial public release of Taiyaki

Taiyaki is ONT research software for training RNN models for basecalling Oxford Nanopore reads.

© 2019 Oxford Nanopore Technologies Ltd.
Taiyaki is distributed under the terms of the Oxford Nanopore Technologies' Public Licence.