slurm_config: big
task_type: local_predict
dataset:
  split: test
  video_processor: VideoProcessor
  aligner: COINActionSegmentationAligner
  bert_name: bert-base-uncased
  test_path: data/coin/COIN.json
  meta_processor: COINActionSegmentationMetaProcessor
  vfeat_dir: data/feat/feat_coin_s3d
  text_processor: COINActionSegmentationTextProcessor
  num_iso_layer: 12
  sliding_window: 16
  sliding_window_size: 32
  max_video_len: 32
  max_len: 96
fairseq:
  dataset:
    batch_size: 1
    valid_subset: test
    num_workers: 2
  common_eval:
    path: runs/mtm/vlm/coin/checkpoint_best.pt
model:
  model_cls: MMFusionActionSegmentation
  mm_encoder_cls: MMBertForTokenClassification
  use_seg_emb: true
eval:
  save_path: runs/mtm/vlm/coin/eval
metric: COINActionSegmentationMetric
predictor: COINPredictor