From 3d47a7f8ab8504a98fba8c8ed1397c28e68cca8d Mon Sep 17 00:00:00 2001 From: Rabeeh KARIMI Date: Wed, 14 Aug 2019 10:58:26 +0200 Subject: [PATCH] loads the tokenizer for each checkpoint, to solve the reproducability issue --- examples/run_glue.py | 1 + 1 file changed, 1 insertion(+) diff --git a/examples/run_glue.py b/examples/run_glue.py index f017db2f6f..c221b0eb7a 100644 --- a/examples/run_glue.py +++ b/examples/run_glue.py @@ -463,6 +463,7 @@ def main(): for checkpoint in checkpoints: global_step = checkpoint.split('-')[-1] if len(checkpoints) > 1 else "" model = model_class.from_pretrained(checkpoint) + tokenizer = tokenizer_class.from_pretrained(checkpoint) model.to(args.device) result = evaluate(args, model, tokenizer, prefix=global_step) result = dict((k + '_{}'.format(global_step), v) for k, v in result.items())