NeMo
PyTorch
English
seq2seq
masked language modeling

Can NeMo t5 checkpoints easily be converted back to Huggingface compatible t5?

#3
by Lauler - opened

I see that you've got scripts for converting from HF to NeMo for training: https://github.com/NVIDIA/NeMo/blob/main/scripts/nlp_language_modeling/hf_t5-v1_1_to_nemo.py .

Is it straightforward to convert back to HF format once training is complete with NeMo?

Sign up or log in to comment