Open mohiburnabil opened 10 months ago
Each split file contains a list, which is composed of the information of each data sample: [{'name': xx, 'gloss': 'A B C', 'text': 'a b c', 'num_frames': 100}, {...}, ..., {...}]
Excuse me, I want to ask you a question. What is the function of pami0-pami2? {'name': 'train/27January_2013_Sunday_tagesschau-8842', 'signer': 'Signer01', 'gloss': 'BLEIBEN WIND', 'text': 'es bleibt windig .', 'num_frames': 43, 'alignments': {'pami0': '0 406 406 406 406 407 407 408 408 408 2986 2986 2986 2986 2987 2987 2987 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 2988 3172 3172 3172 3173 3173 3173 3174 3174 3174', 'pami1': '0 0 20 20 20 28 28 3 3 0 0 0 0 0 0 0 0 36 36 12 12 30 0 0 0 0 0 0 0 0 0 0 0 0 14 14 23 23 23 23 23 23 23', 'pami2': '0 0 0 13 13 13 13 13 0 0 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 45 0 0 0 0 0 0 0 0 0'}, 'sign_features': tensor([])}
dev: data/phoenix-2014t/phoenix-2014t_cleaned.dev test: data/phoenix-2014t/phoenix-2014t_cleaned.test train: data/phoenix-2014t/phoenix-2014t_cleaned.train
What is the process to generate the .test .train .dev files?