i3
Collection
7 items
•
Updated
config = {
"d_model": 1180,
"n_rwkv_layers": 16,
"n_attn_layers": 4,
"n_heads": 10,
"seq_len": 1024,
"batch_size": 4,
"accum_steps": 8,
"lr": 4e-4,
}