Wals — Roberta Sets 136zip Best

tokenizer = RobertaTokenizer.from_pretrained('roberta-base') model = RobertaForSequenceClassification.from_pretrained('roberta-base', num_labels=136) # 136 features

with zipfile.ZipFile('wals_roberta_sets_136zip_best.zip', 'r') as zip_ref: zip_ref.extractall('wals_data/') wals roberta sets 136zip best

to run the WALS optimization before feeding the latent factors into the RoBERTa layers. Optimization ("Best" Settings) Latent Factors tokenizer = RobertaTokenizer