TimeVQVAE-saved_models-20230203.zip (2 GB)
Pretrained models of TimeVQVAE
This is a part of the TimeVQVAE GitHub repository, https://github.com/ML4ITS/TimeVQVAE.
It provides the pretrained models of TimeVQVAE for all the UCR archive datasets.
For each dataset, there are 7 pretrained models:
- two encoders (one for LF and another for HF), named encoder_l and encoder_h.
- two decoders (one for LF and another for HF), named decoder_l and decoder_h.
- two vector-quantizers (one for LF and another for HF), named vq_model_l and vq_model_h.
- maskgit (i.e., the prior model), named maskgit.
Note that the file name format is "model_type-dataset_name".