I also see difference in the evaluation results, when running huggingface trainer, with different evaluation batch size [per_device_eval_batch_size]
1 Like
I also see difference in the evaluation results, when running huggingface trainer, with different evaluation batch size [per_device_eval_batch_size]