matilda-mini / config.json
prometheus04's picture
Add trained checkpoint: 3B tokens, loss=3.16, MFU=31.5%
5442313 verified
Raw
History Blame Contribute Delete
488 Bytes
{
"model_type": "matilda",
"architectures": [
"MatildaForCausalLM"
],
"auto_map": {
"AutoConfig": "configuration_matilda.MatildaConfig",
"AutoModelForCausalLM": "modeling_matilda.MatildaForCausalLM"
},
"vocab_size": 50257,
"max_seq_len": 1024,
"d_model": 768,
"n_layers": 12,
"n_heads": 12,
"n_kv_heads": 4,
"qk_norm": true,
"rope_theta": 10000.0,
"tie_weights": true,
"bos_token_id": 50256,
"eos_token_id": 50256,
"torch_dtype": "bfloat16"
}