sound_v2/audio_caption_json/data.2.json

{
    "keys": {
        "reason_seq": "sound_v2/64splits/reason_tokens.3.pt",
        "semantic_seq": "sound_v2/64splits/semantic_tokens.3.pt",
        "caption_seq": "sound_v2/64splits/caption.3.pt"
    },
    "type": [
        "audio",
        "audio",
        "text"
    ],
    "features": {},
    "loss_key": [
        "caption_seq"
    ],
    "task": "audio_caption",
    "repeat_num": 1
}