Update config.json

config.json

@@ -67,6 +67,10 @@
     "model_type": "llava-qwen2",
     "num_attention_heads": 16,
     "num_hidden_layers": 24,
+    "quantization": {
+      "group_size": 64,
+      "bits": 4
+    },
     "num_key_value_heads": 16,
     "rms_norm_eps": 1e-06,
     "rope_theta": 1000000.0,