dataset_path: WildVision/wildvision-arena-data
dataset_kwargs:
  token: True
output_type: generate_until
doc_to_visual: !function utils.wild_vision_doc_to_visual
doc_to_text: !function utils.wild_vision_doc_to_text
doc_to_target: !function utils.wild_vision_doc_to_target
process_results_use_image: true
generation_kwargs:
  max_new_tokens: 4096
  temperature: 0
  top_p: 1.0
  num_beams: 1
  do_sample: false
# The return value of process_results will be used by metrics
process_results: !function utils.wild_vision_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
metric_list:
  - metric: raw_scores
    aggregation: !function utils.wild_vision_aggregation_raw_scores
    higher_is_better: true
  - metric: elo_scores
    aggregation: !function utils.wild_vision_aggregation_elo_scores
    higher_is_better: true
  - metric: win_rates
    aggregation: !function utils.wild_vision_aggregation_win_rates
    higher_is_better: true
  - metric: judgements_better
    aggregation: !function utils.wild_vision_aggregation_judgements_better
    higher_is_better: true
  - metric: judgements_better_plus
    aggregation: !function utils.wild_vision_aggregation_judgements_better_plus
    higher_is_better: true
  - metric: judgements_worse
    aggregation: !function utils.wild_vision_aggregation_judgements_worse
    higher_is_better: false
  - metric: judgements_worse_plus
    aggregation: !function utils.wild_vision_aggregation_judgements_worse_plus
    higher_is_better: false
  - metric: judgements_tie
    aggregation: !function utils.wild_vision_aggregation_judgements_tie
    higher_is_better: false
  - metric: judgements_unclear
    aggregation: !function utils.wild_vision_aggregation_judgements_unclear
    higher_is_better: false
metadata:
  judge_model: gpt-4o
  baseline_model: claude-3-sonnet-20240229
