6955fab0f4f0a38ead24cec84152d82769a0cfa2,scripts/bert/run_pretraining_hvd.py,,,#,248

Before Change

            get_dataset_fn = get_pretrain_data_npz
            if args.cased:
                raise UserWarning("argument cased is valid only when --raw is set")
            if args.max_seq_length:
                raise UserWarning("argument max_seq_length is valid only when --raw is set")
            if args.short_seq_prob:
                raise UserWarning("argument short_seq_prob is valid only when --raw is set")
            if args.masked_lm_prob:
                raise UserWarning("argument masked_lm_prob is valid only when --raw is set")
            if args.max_predictions_per_seq:
                raise UserWarning("argument max_predictions_per_seq is valid only when "

After Change

        else:
            tokenizer = nlp.data.BERTTokenizer(vocab=vocab, lower=not args.cased)

        cache_dir = os.path.join(args.ckpt_dir, "data_eval_cache")
        cache_file = os.path.join(cache_dir, "part-000.npz")
        nlp.utils.mkdir(cache_dir)

        // generate dev dataset from the raw text if needed

Instances