lnp_ml/models/pretrain/moe/pretrain_history.json

238 lines
4.1 KiB
JSON

{
"train": [
{
"loss": 0.8444833619870162,
"n_samples": 8236
},
{
"loss": 0.7221717920159761,
"n_samples": 8236
},
{
"loss": 0.6476956897131617,
"n_samples": 8236
},
{
"loss": 0.6058183341653907,
"n_samples": 8236
},
{
"loss": 0.5818977655282931,
"n_samples": 8236
},
{
"loss": 0.5606534384650591,
"n_samples": 8236
},
{
"loss": 0.5438616421108051,
"n_samples": 8236
},
{
"loss": 0.5296446231454828,
"n_samples": 8236
},
{
"loss": 0.5034642058301169,
"n_samples": 8236
},
{
"loss": 0.4926449143521817,
"n_samples": 8236
},
{
"loss": 0.4809608229043821,
"n_samples": 8236
},
{
"loss": 0.46792712891605537,
"n_samples": 8236
},
{
"loss": 0.46016837027595137,
"n_samples": 8236
},
{
"loss": 0.4539264049589084,
"n_samples": 8236
},
{
"loss": 0.4343785792082592,
"n_samples": 8236
},
{
"loss": 0.4022773470098154,
"n_samples": 8236
},
{
"loss": 0.391768712971962,
"n_samples": 8236
},
{
"loss": 0.3946328157646787,
"n_samples": 8236
},
{
"loss": 0.38186461452780096,
"n_samples": 8236
},
{
"loss": 0.37831240947412137,
"n_samples": 8236
},
{
"loss": 0.37218554509734,
"n_samples": 8236
},
{
"loss": 0.37003324522793785,
"n_samples": 8236
},
{
"loss": 0.37280535147194727,
"n_samples": 8236
},
{
"loss": 0.366068535345512,
"n_samples": 8236
},
{
"loss": 0.3588095046959785,
"n_samples": 8236
},
{
"loss": 0.3447240012661238,
"n_samples": 8236
},
{
"loss": 0.34654789203402486,
"n_samples": 8236
},
{
"loss": 0.3341197949726064,
"n_samples": 8236
},
{
"loss": 0.3334885352210202,
"n_samples": 8236
}
],
"val": [
{
"loss": 0.8233604396717749,
"n_samples": 1454
},
{
"loss": 0.7986104544943924,
"n_samples": 1454
},
{
"loss": 0.6878670279392678,
"n_samples": 1454
},
{
"loss": 1.4778851701927973,
"n_samples": 1454
},
{
"loss": 0.7188797573439849,
"n_samples": 1454
},
{
"loss": 0.708491360878846,
"n_samples": 1454
},
{
"loss": 0.7465513021450751,
"n_samples": 1454
},
{
"loss": 0.7036250697697373,
"n_samples": 1454
},
{
"loss": 0.6551542343594841,
"n_samples": 1454
},
{
"loss": 0.6759496378767441,
"n_samples": 1454
},
{
"loss": 1.0201101731893285,
"n_samples": 1454
},
{
"loss": 0.6907692915635719,
"n_samples": 1454
},
{
"loss": 0.7211924836428982,
"n_samples": 1454
},
{
"loss": 0.7141936532896848,
"n_samples": 1454
},
{
"loss": 0.773055208434429,
"n_samples": 1454
},
{
"loss": 1.3679038423947487,
"n_samples": 1454
},
{
"loss": 0.6390884835913552,
"n_samples": 1454
},
{
"loss": 0.6397012937019747,
"n_samples": 1454
},
{
"loss": 0.6198025200685084,
"n_samples": 1454
},
{
"loss": 0.6775723626065943,
"n_samples": 1454
},
{
"loss": 1.147026978791826,
"n_samples": 1454
},
{
"loss": 0.9496700583330703,
"n_samples": 1454
},
{
"loss": 0.7171065121930108,
"n_samples": 1454
},
{
"loss": 0.6501950878224642,
"n_samples": 1454
},
{
"loss": 0.6247917262035354,
"n_samples": 1454
},
{
"loss": 0.6408829362717094,
"n_samples": 1454
},
{
"loss": 0.6341315902246898,
"n_samples": 1454
},
{
"loss": 0.6743946996154943,
"n_samples": 1454
},
{
"loss": 0.6599862504202201,
"n_samples": 1454
}
]
}