slurm_config: big
task_type: local_predict
dataset:
  split: test
  video_processor: VideoProcessor
  aligner: DSAligner
  bert_name: bert-base-uncased
  meta_processor: MSRVTTMetaProcessor
  test_path: data/msrvtt/MSRVTT_JSFUSION_test.csv
  vfeat_dir: data/feat/feat_vtt_s3d
  text_processor: MSRVTTTextProcessor
  num_iso_layer: 12
  max_video_len: 32
  max_len: 96
fairseq:
  dataset:
    batch_size: 256
    valid_subset: test
    num_workers: 2
  common_eval:
    path: runs/mtm/vlm/vtt/checkpoint_last.pt
model:
  model_cls: MMFusionJoint
  mm_encoder_cls: MMBertForJoint
  use_seg_emb: true
eval:
  save_path: runs/mtm/vlm/vtt/eval
metric: RetrievalMetric
predictor: RetrievalPredictor