llama-7b/bert4torch_config.json

{
	"model": "llama",
	"bos_token_id": 1,
	"eos_token_id": 2,
	"hidden_size": 4096,
	"intermediate_size": 11008, 
	"num_attention_heads": 32,
	"num_hidden_layers": 32,
	"layer_norm_eps": 1e-06,
	"hidden_act": "silu",
	"vocab_size": 32000,
	"segment_vocab_size": 0,
	"skip_init": true,
	"rope_rank": "updown",
    "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "max_length": 2048, "end_id": 2}
}