Files
tinybrain-100m-instruct/tinybrain_sft_meta.json

10 lines
212 B
JSON
Raw Permalink Normal View History

{
"step": 522,
"val_loss": 2.2578155994415283,
"base_model": "exnivo/tinybrain-100m-base",
"dataset": "exnivo/tinybrain-instruct",
"batch_size": 100,
"block_size": 1024,
"epochs": 3,
"lr": 7e-05
}