ERROR happens when I try to run grpo.py with NuminaMath-TIR dataset during evaluation

[re-imagined]

command:

export PYTHONPATH="${PYTHONPATH}:/code/open-r1.git/src"

ACCELERATE_LOG_LEVEL=info accelerate launch \
--config_file /code/open-r1.git/recipes/accelerate_configs/zero2.yaml \
--num_processes  3 \
/code/open-r1.git/src/open_r1/grpo.py \
--model_name_or_path  $MODEL_PATH \
--model_revision  main \
--torch_dtype  bfloat16 \
--attn_implementation  flash_attention_2 \
--dataset_name $DATA_PATH \
--bf16  true \
--use_vllm  true \
--vllm_device  auto \
--vllm_gpu_memory_utilization  0.7 \
--do_eval  true \
--eval_strategy  steps \
--eval_steps  100 \
--gradient_accumulation_steps  16 \
--gradient_checkpointing  true \
--gradient_checkpointing_kwargs '{"use_reentrant": false}'  \
--learning_rate  2.0e-05 \
--log_level  info \
--logging_steps  5 \
--logging_strategy  steps \
--lr_scheduler_type  cosine \
--max_prompt_length  512 \
--max_completion_length  1024 \
--max_steps  -1 \
--num_generations  2 \
--num_train_epochs  1 \
--output_dir  $TRAIN_OUTPUT_PATH  \
--overwrite_output_dir  true \
--per_device_eval_batch_size  4    \
--per_device_train_batch_size  2 \
--report_to  none \
--save_strategy  "no" \
--seed  42 \
--warmup_ratio  0.1

error log:

2025-02-12T14:20:45.496251877Z ***** Running Evaluation *****
2025-02-12T14:20:45.496289060Z [INFO|trainer.py:4228] 2025-02-12 14:20:45,496 >>   Num examples = 99
2025-02-12T14:20:45.496296461Z [INFO|trainer.py:4231] 2025-02-12 14:20:45,496 >>   Batch size = 4
2025-02-12T14:20:45.501687663Z [rank2]: Traceback (most recent call last):
2025-02-12T14:20:45.501692127Z [rank2]:   File "/code/open-r1.git/src/open_r1/grpo.py", line 259, in <module>
2025-02-12T14:20:45.501694358Z [rank2]:     main(script_args, training_args, model_args)
2025-02-12T14:20:45.501702504Z [rank2]:   File "/code/open-r1.git/src/open_r1/grpo.py", line 213, in main
2025-02-12T14:20:45.501704855Z [rank2]:     train_result = trainer.train(resume_from_checkpoint=checkpoint)
2025-02-12T14:20:45.501706957Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 2171, in train
2025-02-12T14:20:45.501709146Z [rank2]:     return inner_training_loop(
2025-02-12T14:20:45.501710977Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 2598, in _inner_training_loop
2025-02-12T14:20:45.501713061Z [rank2]:     self._maybe_log_save_evaluate(
2025-02-12T14:20:45.501714874Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 3071, in _maybe_log_save_evaluate
2025-02-12T14:20:45.501717218Z [rank2]:     metrics = self._evaluate(trial, ignore_keys_for_eval)
2025-02-12T14:20:45.501719124Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 3025, in _evaluate
2025-02-12T14:20:45.501721188Z [rank2]:     metrics = self.evaluate(ignore_keys=ignore_keys_for_eval)
2025-02-12T14:20:45.501723024Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4073, in evaluate
2025-02-12T14:20:45.501725115Z [rank2]:     output = eval_loop(
2025-02-12T14:20:45.501726891Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4267, in evaluation_loop
2025-02-12T14:20:45.501729071Z [rank2]:     losses, logits, labels = self.prediction_step(model, inputs, prediction_loss_only, ignore_keys=ignore_keys)
2025-02-12T14:20:45.501731004Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4436, in prediction_step
2025-02-12T14:20:45.501732983Z [rank2]:     has_labels = False if len(self.label_names) == 0 else all(inputs.get(k) is not None for k in self.label_names)
2025-02-12T14:20:45.501734915Z [rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4436, in <genexpr>
2025-02-12T14:20:45.501736994Z [rank2]:     has_labels = False if len(self.label_names) == 0 else all(inputs.get(k) is not None for k in self.label_names)
2025-02-12T14:20:45.501738958Z [rank2]: AttributeError: 'list' object has no attribute 'get'
2025-02-12T14:20:45.502911925Z [rank0]: Traceback (most recent call last):
2025-02-12T14:20:45.502915986Z [rank0]:   File "/code/open-r1.git/src/open_r1/grpo.py", line 259, in <module>
2025-02-12T14:20:45.502918069Z [rank0]:     main(script_args, training_args, model_args)
2025-02-12T14:20:45.502919854Z [rank0]:   File "/code/open-r1.git/src/open_r1/grpo.py", line 213, in main
2025-02-12T14:20:45.502921873Z [rank0]:     train_result = trainer.train(resume_from_checkpoint=checkpoint)
2025-02-12T14:20:45.502923837Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 2171, in train
2025-02-12T14:20:45.502925908Z [rank0]:     return inner_training_loop(
2025-02-12T14:20:45.502929137Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 2598, in _inner_training_loop
2025-02-12T14:20:45.502931208Z [rank0]:     self._maybe_log_save_evaluate(
2025-02-12T14:20:45.502933181Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 3071, in _maybe_log_save_evaluate
2025-02-12T14:20:45.502935279Z [rank0]:     metrics = self._evaluate(trial, ignore_keys_for_eval)
2025-02-12T14:20:45.502937170Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 3025, in _evaluate
2025-02-12T14:20:45.502939127Z [rank0]:     metrics = self.evaluate(ignore_keys=ignore_keys_for_eval)
2025-02-12T14:20:45.502941384Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4073, in evaluate
2025-02-12T14:20:45.502943349Z [rank0]:     output = eval_loop(
2025-02-12T14:20:45.502945096Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4267, in evaluation_loop
2025-02-12T14:20:45.502947041Z [rank0]:     losses, logits, labels = self.prediction_step(model, inputs, prediction_loss_only, ignore_keys=ignore_keys)
2025-02-12T14:20:45.502949051Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4436, in prediction_step
2025-02-12T14:20:45.502951004Z [rank0]:     has_labels = False if len(self.label_names) == 0 else all(inputs.get(k) is not None for k in self.label_names)
2025-02-12T14:20:45.502953072Z [rank0]:   File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 4436, in <genexpr>
2025-02-12T14:20:45.502955101Z [rank0]:     has_labels = False if len(self.label_names) == 0 else all(inputs.get(k) is not None for k in self.label_names)
2025-02-12T14:20:45.502957174Z [rank0]: AttributeError: 'list' object has no attribute 'get'