flax-community
/

bertin-roberta-large-spanish

Fill-Mask Transformers PyTorch JAX Safetensors Spanish roberta spanish Inference Endpoints

Model card Files Files and versions Community

versae commited on Jul 21, 2021

Commit

16447f8

•

1 Parent(s): daff9ab

Fix restoring steps

Files changed (1) hide show

run_mlm_flax_stream.py +1 -1

run_mlm_flax_stream.py CHANGED Viewed

@@ -691,7 +691,7 @@ if __name__ == "__main__":
     eval_samples = advance_iter_and_group_samples(training_iter, data_args.num_eval_samples, max_seq_length)
     steps = tqdm(range(num_train_steps), desc="Training...", position=0)
-    for step in range(saved_step, num_train_steps):
         if step < saved_step:
             steps.update(1)
             continue

     eval_samples = advance_iter_and_group_samples(training_iter, data_args.num_eval_samples, max_seq_length)
     steps = tqdm(range(num_train_steps), desc="Training...", position=0)
+    for step in range(0, num_train_steps):
         if step < saved_step:
             steps.update(1)
             continue