evaluation_val/art/choose_hypothesis_options/results.json

{
  "dataset_name": "art",
  "dataset_config_name": null,
  "template_name": "choose_hypothesis_options",
  "evaluation": {
    "accuracy": 0.6298955613577023
  },
  "arguments": "Namespace(config_name=None, dataset_config_name=None, dataset_name='art', debug=False, dtype='bfloat16', max_length=2048, model_name_or_path='/gpfsscratch/rech/six/commun/commun/experiments/muennighoff/bloomckpt/176bt0/p31lossseq_global_step498', output_dir='/gpfsscratch/rech/six/commun/commun/experiments/muennighoff/bloomckpt/176bt0/p31lossseq_global_step498/evaluation', pad_to_max_length=False, per_device_eval_batch_size=4, prefixlm=False, split='validation', target_max_length=256, template_config_name=None, template_name='choose_hypothesis_options', tokenizer_name=None, use_slow_tokenizer=False)"
}