Log which device is used for training a tagger

stanfordnlp · Jan 11, 2025 · a8b4190 · a8b4190
1 parent 37b0416
commit a8b4190
Showing 1 changed file with 2 additions and 0 deletions.
diff --git a/stanza/models/tagger.py b/stanza/models/tagger.py
@@ -307,6 +307,8 @@ def train(args):
     global_start_time = time.time()
     format_str = 'Finished STEP {}/{}, loss = {:.6f} ({:.3f} sec/batch), lr: {:.6f}'
 
+    logger.debug("Training model on device %s", next(trainer.model.parameters()).device)
+
     if args['adapt_eval_interval']:
         args['eval_interval'] = utils.get_adaptive_eval_interval(dev_data.num_examples, 2000, args['eval_interval'])
         logger.info("Evaluating the model every {} steps...".format(args['eval_interval']))