Enforce range setting only used with per-channel (pytorch#13889)

Differential Revision: D81537565

Author: rohansjoshi

examples/qualcomm/oss_scripts/llama/range_setting_pt2e.py

@@ -210,12 +210,13 @@ def make_custom_quantizer(
         per_channel_linear=True,
         act_observer=MinMaxObserver,
     )
-    if range_setting in ("mse_weight_only", "mse_with_act_loss", "na"):
-        if range_setting == "na":
-            observer = PerChannelMinMaxObserver
-        elif range_setting == "mse_weight_only":
+    if range_setting in ("mse_weight_only", "mse_with_act_loss"):
+        assert (
+            quant_dtype != QuantDtype.use_16a4w_block
+        ), "Range setting only supported for per-channel quantization"
+        if range_setting == "mse_weight_only":
             observer = PerChannelMSEObserver.with_args(
-                **{"steps": 200, "use_mse": True}
+                **{"steps": 1600, "use_mse": True}
             )
         else:
             observer = PerChannelFixedQParamsObserver.with_args(**{"eps": 2**-12})