[BUG] Riva deploy NeMo trained Punctuation_and_Capitalisation

When converting Punctuation_and_Capitalisation RMIR to model, some artifacts are not renamed correctly (they retain the nemo: artifact tag), eg. as seen in RMIR/manifest.yaml

  tokenizer-287d39d680584cbda278b45d25435eb5_tokenizer.vocab_file: {artifact_type: File,
    conf_path: 287d39d680584cbda278b45d25435eb5_tokenizer.vocab_file, content_callback: BinaryContentCallback,
    description: '', encryption: false, module: tokenizer, path_type: TAR_PATH}

and RMIR/artifacts/config.yaml

global_variables: { ...
  tokenizer: wordpiece, vocab: 'nemo:287d39d680584cbda278b45d25435eb5_tokenizer.vocab_file'}
tokenizer: { ...
  tokenizer: wordpiece, unk_token: '[UNK]', vocab: 'nemo:287d39d680584cbda278b45d25435eb5_tokenizer.vocab_file'}

ends up in punctuation_tokenizer/config.pbtxt as

parameters {
  key: "vocab"
  value {
    string_value: "/data/models/punctuation_tokenizer/1/nemo:287d39d680584cbda278b45d25435eb5_tokenizer.vocab_file"
backend: "riva_nlp_tokenizer"

this leads to Triton crashing with

I0120 20:57:15.955918 2464 tokenizer_library.cc:21] TRITONBACKEND_ModelInstanceInitialize: punctuation_tokenizer_0 (device 0)
terminate called after throwing an instance of 'riva::nlp::tokenizer::VocabularyException'
  what():  Error occurred processing vocabulary file

Is perhaps there a better way of reporting bugs, issues, and contributing? For example is there a GitHub repo for the servicemaker? This would be a much better alternative than reverse engineering the code deep inside a container. Thanks for the awesome toolset.