sync main to W4V8

.gitattributes

@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 Llama-3.1-8B-Instruct.mxq filter=lfs diff=lfs merge=lfs -text
 Llama-3.1-8B-Instruct-W8.mxq filter=lfs diff=lfs merge=lfs -text
+Llama-3.1-8B-Instruct-W4V8.mxq filter=lfs diff=lfs merge=lfs -text

Llama-3.1-8B-Instruct-W8.mxq → Llama-3.1-8B-Instruct-W4V8.mxq

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:38bfd2bb5101236b8e7a95210c226c26eefd99763afd9cb0ed1eff32d8088b7a
-size 7551802853
+oid sha256:e55d74b63c342473ac7afe1a175e823071c2cc9708e754c6941eb2c6c63e5ac6
+size 3866355461

config.json

@@ -22,7 +22,7 @@
   "max_position_embeddings": 4096,
   "mlp_bias": false,
   "model_type": "mobilint-llama",
-  "mxq_path": "Llama-3.1-8B-Instruct-W8.mxq",
+  "mxq_path": "Llama-3.1-8B-Instruct-W4V8.mxq",
   "num_attention_heads": 32,
   "num_hidden_layers": 32,
   "num_key_value_heads": 8,