AssertionError assert(len(tmp_ids) == 1) when using RoBERTa

[steveazzolin]

Hi,

I was trying to run scripts/run_optiprompt.pt with the model roberta-base when I encountered the following problem:

Traceback (most recent call last): File "/nfs/data_chaos/sazzolin/OptiPrompt/code/run_optiprompt.py", line 176, in best_result, result_rel = evaluate(model, valid_samples_batches, valid_sentences_batches, filter_indices, index_list) File "/nfs/data_chaos/sazzolin/OptiPrompt/code/utils.py", line 110, in evaluate log_probs, cor_b, tot_b, pred_b, topk_preds, loss, common_vocab_loss = model.run_batch(sentences_b, samples_b, training=False, filter_indices=filter_indices, index_list=index_list, vocab_to_common_vocab=vocab_to_common_vocab) File "/nfs/data_chaos/sazzolin/OptiPrompt/code/models.py", line 345, in run_batch tokens_tensor, segments_tensor, attention_mask_tensor, masked_indices_list, tokenized_text_list, mlm_labels_tensor, mlm_label_ids = self._get_input_tensors_batch_train(sentences_list, samples_list) File "/nfs/data_chaos/sazzolin/OptiPrompt/code/models.py", line 151, in _get_input_tensors_batch_train tokens_tensor, segments_tensor, masked_indices, tokenized_text, mlm_labels_tensor, mlm_label_id = self.__get_input_tensors(sentences, mlm_label=samples['obj_label']) File "/nfs/data_chaos/sazzolin/OptiPrompt/code/models.py", line 298, in __get_input_tensors assert(len(tmp_ids) == 1) AssertionError

Nonetheless, the code works just fine when using bert-base-cased.

This is the complete python call:

python code/run_optiprompt.py \
            --relation_profile relation_metainfo/LAMA_relations.jsonl \
            --relation ${REL} \
            --common_vocab_filename common_vocabs/common_vocab_cased.txt \
            --model_name roberta_base \
            --do_train \
            --train_data data/autoprompt_data/${REL}/train.jsonl \
            --dev_data data/autoprompt_data/${REL}/dev.jsonl \
            --do_eval \
            --test_data data/LAMA-TREx/${REL}.jsonl \
            --output_dir ${DIR} \
            --random_init none \
            --seed ${SEED} \
            --output_predictions