support llama2-7b and llama2-13b

examples/cpp/llama/llama_config.ini

@@ -17,6 +17,12 @@ model_dir=/data/models/llama-7b-infer/1-gpu
 ; model_name=llama_65b
 ; model_dir=/data/models/llama-65b-hf-infer/8-gpu
 
+; model_name=llama2_7b
+; model_dir=/data/models/chinese-llama-2-7b-infer/1-gpu
+
+; model_name=llama2_13b
+; model_dir=/data/models/chinese-llama-2-13b-infer/1-gpu
+
 [request]
 beam_width=1 # beam width for beam search
 top_k=1 ; k value for top k sampling
@@ -76,3 +82,27 @@ vocab_size = 32000
 start_id = 0
 end_id = 1
 weight_data_type = fp16
+
+[llama2_7b]
+head_num = 32
+size_per_head = 128
+inter_size = 11008
+num_layer = 32
+rotary_embedding = 128
+layernorm_eps = 1e-05
+vocab_size = 55296
+start_id = 0
+end_id = 1
+weight_data_type = fp16
+
+[llama2_13b]
+head_num = 40
+size_per_head = 128
+inter_size = 13824
+num_layer = 40
+rotary_embedding = 128
+layernorm_eps = 1e-05
+vocab_size = 55296
+start_id = 0
+end_id = 1
+weight_data_type = fp16