Qwen3-0.6B-Looped / loop_config.json
coolpoodle's picture
Initial upload
2d58dce verified
raw
history blame contribute delete
274 Bytes
{
"base_model": "/content/Qwen3-0.6B",
"loop_window_size": 64,
"num_layers": 28,
"num_heads": 16,
"head_dim": 128,
"final_val_loss": 3.6202090362707775,
"final_val_ppl": 37.34537124633789,
"training_epochs": 3,
"training_time_minutes": 38.990576179822284
}