librispeech_tts_r.json

{
    "model_name_or_path": "ylacombe/musicgen-melody",

    "push_to_hub": false,
    "hub_model_id": "stable-speech-mini",
    "report_to": ["wandb"],
    "overwrite_output_dir": true,
    "output_dir": "/home/yoach/dataspeech/artefacts/training/",

    "train_dataset_name": "blabble-io/libritts_r+blabble-io/libritts_r+blabble-io/libritts_r",
    "train_metadata_dataset_name": "stable-speech/libritts-r-tags-and-text-generated+stable-speech/libritts-r-tags-and-text-generated+stable-speech/libritts-r-tags-and-text-generated",
    "train_dataset_config_name": "clean+clean+other",
    "train_split_name": "train.clean.360+train.clean.100+train.other.500",

    "eval_dataset_name": "blabble-io/libritts_r+blabble-io/libritts_r",
    "eval_metadata_dataset_name": "stable-speech/libritts-r-tags-and-text-generated+stable-speech/libritts-r-tags-and-text-generated",
    "eval_dataset_config_name": "clean+other",
    "eval_split_name": "test.clean+test.other",

    "target_audio_column_name": "audio", 
    "description_column_name": "text",
    "prompt_column_name": "text_description",

    "max_train_samples": 1000,
    "max_eval_samples": 200,

    
    "max_duration_in_seconds": 20,
    "min_duration_in_seconds": 1.0,

    "add_audio_samples_to_wandb": true,
    "id_column_name": "id",

    "preprocessing_num_workers": 24,

    "pad_token_id": 2048,
    "decoder_start_token_id": 2048,

    "do_train": true,
    "num_train_epochs": 20,
    "gradient_accumulation_steps": 1,
    "gradient_checkpointing": true,
    "per_device_train_batch_size": 2,
    "learning_rate": 1e-6,
    "adam_beta1": 0.9,
    "adam_beta2": 0.95,
    "weight_decay": 0.1,

    "logging_steps": 25,


    "do_eval": true, 
    "predict_with_generate": true,
    "include_inputs_for_metrics": true,
    "evaluation_strategy": "epoch",
    "per_device_eval_batch_size": 2,
    "generation_max_length": 400,

    "fp16": true,
    "seed": 456,

    "dataloader_num_workers":8
}