Hello,
Great work! I see that in the Hugging face repo for the SthV2 pretrained model you provide only the params file (13.6GB), while in the OXE pretrained you also provide tokenizer.model and vqgan. If i want to use the SthV2 pretraining without any post-training on OXE should i just transfer the tokenizer.model and vqgan files and use the SthV2 params file?
Hello,
Great work! I see that in the Hugging face repo for the SthV2 pretrained model you provide only the
paramsfile (13.6GB), while in the OXE pretrained you also providetokenizer.modelandvqgan. If i want to use the SthV2 pretraining without any post-training on OXE should i just transfer thetokenizer.modelandvqganfiles and use the SthV2paramsfile?