{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreiaipzchfbowdfgucjsmdq2pztnwol6lx73o63slbkqnuq6hfj4qia",
"uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mh64yezjomt2"
},
"path": "/t/best-practices-to-create-an-audio-dataset/174312#post_1",
"publishedAt": "2026-03-16T09:13:22.000Z",
"site": "https://discuss.huggingface.co",
"textContent": "Hello everyone,\n\nI have a question about the percentage breakdown of test, training, and validation data.\n\nIs there some kind of guideline, such as\n\nusing 10% for testing, 5% for evaluation, and 85% for training?\n\nAnd my bonus question: does it make a difference if the test and evaluation files are the same? I mean the same MP3 files.\n\nThanks in advance",
"title": "Best practices to create an audio dataset"
}