README.md

metadata

configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: label
      dtype: string
    - name: prompt_similarity
      dtype: float32
    - name: identity_consistency
      dtype: float32
    - name: negative_prompt_similarity
      dtype: float32
    - name: target_prompt_similarity
      dtype: float32
    - name: aesthetic_score
      dtype: float32
  splits:
    - name: train
      num_bytes: 112
      num_examples: 4
  download_size: 3513
  dataset_size: 112

    method: cte_cold_emb_qv 

    num_inference_steps: 30

    {metric} : {mean} 

{metric} : {mean} {metric} : {mean} {metric} : {mean} {metric} : {mean}