dataset_path: MMMU/MMMU
validation_split: validation
output_type: generate_until
doc_to_visual: !function utils.mmmu_doc_to_visual
doc_to_text: !function utils.mmmu_doc_to_text
doc_to_target: "answer"
# The return value of process_results will be used by metrics
# process_results: !function utils.mmmu_process_results
# Note that the metric name can be either a registed metric function (such as the case for GQA) or a key name returned by process_results
process_results: !function utils.mmmu_process_results
# Note that the metric name can be either a registed metric function (such as the case for GQA) or a key name returned by process_results
generation_kwargs:
  until:
    - "<|endoftext|>"
  temperature: 1.0
  do_sample: false
  max_gen_toks: 512
  top_p: 0.94
  repetition_penalty: 1.0
  image_aspect_ratio: original
metric_list:
  - metric: mmmu_acc
    aggregation: !function utils.mmmu_aggregate_results
    higher_is_better: true