Add quantized model precision

src/display/utils.py

@@ -99,9 +99,14 @@ class WeightType(Enum):
 class Precision(Enum):
     float16 = ModelDetails("float16")
     bfloat16 = ModelDetails("bfloat16")
-    qt_8bit = ModelDetails("8bit")
-    qt_4bit = ModelDetails("4bit")
-    qt_GPTQ = ModelDetails("GPTQ")
+
+    qt_gptq_3bit = ModelDetails("GPTQ-3bit")
+    qt_gptq_4bit = ModelDetails("GPTQ-4bit")
+    qt_gptq_8bit = ModelDetails("GPTQ-8bit")
+    qt_awq_3bit = ModelDetails("AWQ-3bit")
+    qt_awq_4bit = ModelDetails("AWQ-4bit")
+    qt_awq_8bit = ModelDetails("AWQ-8bit")
+
     Unknown = ModelDetails("?")
 
     def from_str(precision):
@@ -109,12 +114,18 @@ class Precision(Enum):
             return Precision.float16
         if precision in ["torch.bfloat16", "bfloat16"]:
             return Precision.bfloat16
-        if precision in ["8bit"]:
-            return Precision.qt_8bit
-        if precision in ["4bit"]:
-            return Precision.qt_4bit
-        if precision in ["GPTQ", "None"]:
-            return Precision.qt_GPTQ
+        if precision in ["gptq-3bit"]:
+            return Precision.qt_gptq_3bit
+        if precision in ["gptq-4bit"]:
+            return Precision.qt_gptq_4bit
+        if precision in ["gptq-8bit"]:
+            return Precision.qt_gptq_8bit
+        if precision in ["awq-3bit"]:
+            return Precision.qt_awq_3bit
+        if precision in ["awq-4bit"]:
+            return Precision.qt_awq_4bit
+        if precision in ["awq-8bit"]:
+            return Precision.qt_awq_8bit
         return Precision.Unknown