predict problem

[leee-SeungHyeon]

python
run_llmmt.py
--model_name_or_path Qwen/Qwen2-7B
--do_predict
--low_cpu_mem_usage
--language_pairs ${TEST_PAIRS}
--mmt_data_path ./human_written_data/
--per_device_eval_batch_size 2
--output_dir ${OUTPUT_DIR}
--use_peft
--peft_model_id /home/xxxx/alma/ALMA/outputs/Qwen2-7B
--predict_with_generate
--max_new_tokens 256
--max_source_length 256
--fp16
--seed 42
--num_beams 5
--overwrite_cache
--overwrite_output_dir
--multi_gpu_one_model

The peft_model_id directory contains the following files:

When running the above files, the following error occurs:

Is there a solution for this?