projection_training_metadata.json

{
  "clip_model": "openai/clip-vit-base-patch32",
  "clap_model": "laion/larger_clap_music_and_speech",
  "embed_dim": 512,
  "training_dataset": "OpenSound/AudioCaps",
  "training_method": "clap_audio_to_clip_text",
  "num_samples": 10000,
  "epochs": 30,
  "batch_size": 256,
  "lr": 0.0001
}