{
"architectures": [
"LlamaForCausalLM"
],
"hidden_size": 2048,
"intermediate_size": 5632,
"num_attention_heads": 32,
"num_hidden_layers": 22,
"num_key_value_heads": 4,
"vocab_size": 32000,
"rms_norm_eps": 1e-05,
"rope_theta": 10000.0,
"tie_word_embeddings": true,
"model_type": "llama"
}