banana-idk-3m / config.json
Compactbot's picture
Proper architecture config for banana-idk-3m
863358d verified
Raw History Blame Contribute Delete
321 Bytes
{
"architectures": ["BananaIDKForCausalLM"],
"model_type": "banana_idk",
"d_model": 144,
"n_layers": 6,
"n_heads": 2,
"n_kv_heads": 1,
"head_dim": 80,
"ffn_mult": 3,
"vocab_size": 8192,
"tie_word_embeddings": true,
"use_rope": true,
"use_rmsnorm": true,
"use_swiglu": true,
"params": 2715984
}