CBT / config.json
Neithanvarvas's picture
[UPDATE] Restructure release: safetensors weights, cbt package, chat demo, clean model card
dde517d verified
Raw History Blame Contribute Delete
323 Bytes
{
"model_type": "cbt",
"architecture": "CBT-18.95M",
"training_step": 23927,
"vocab_size": 111,
"block_size": 1024,
"n_layer": 6,
"n_head": 8,
"n_embd": 512,
"n_recur": 3,
"dropout": 0.0,
"bias": false,
"rope_theta": 10000.0,
"max_coords": 4096,
"k_rand_shift": 45,
"pos_emb_type": "rope_2d"
}