PakMosaic-Small / training_summary.json
mahwizzzz's picture
Add PakMosaic-Small research-preview weights, training summary, and diagnostic evals
9ae456b verified
Raw History Blame Contribute Delete
272 Bytes
{
"step": 2904769,
"tokens_seen": 2000000068,
"token_budget": 2000000000,
"last_loss": 3.107429265975952,
"last_masked_accuracy": 0.4583333432674408,
"mean_loss_last_500": 1.9579948586970568,
"mean_masked_accuracy_last_500": 0.6402293934822082,
"n": 500
}