Unable to jit script the EncoderClassifier #1691

I'm using the EncoderClassifier (took example from huggingface) this way:
EncoderClassifier.from_hparams(source='speechbrain/lang-id-commonlanguage_ecapa')
and I want to export it to jit.
First I followed the instructions (3. Post Training Static Quantization), and quantized the model (the size reduced from 80 to 20 Mb btw) and then I'm trying to save it for jit-compilation:
torchscript_model = torch.jit.script(model_static_quantized)
but I'm getting an error:
torch.jit.frontend.UnsupportedNodeError: try blocks aren't supported: File "/.../lib/python3.9/site-packages/packaging/version.py", line 48 a valid PEP 440 version or a legacy version. """ try: ~~~ <--- HERE return Version(version) except InvalidVersion:

Okay, I followed to the training recipe and removed the usages of this method (packaging.parse - it just checks the torch version, I left the code suitable for my version) but error remained the same.
I also removed pretrainer part from train_ecapa_tdnn.yaml and from train.py - nothing changed.
What am I doing wrong?