--batch_size

Switch

--batch_size N

Description

Batch size to use when generating embeddings with --bert_model/--emb_model.

Argument and Default Value

An integer. Default: 32 (dlac.GPU_BATCH_SIZE).

Details

Verified end-to-end: ran successfully at --batch_size 8 producing real embedding features for a small corpus.

Other Switches

Required Switches:

• --bert_model or --emb_model

Example Commands

dlatkInterface.py -d dla_tutorial -t msgs -c user_id --add_bert --batch_size 16