fix typos

Signed-off-by: Kyle Sayers <[email protected]>

Author: kylesayrs

src/llmcompressor/observers/helpers.py

@@ -22,6 +22,23 @@ def flatten_for_calibration(
     args: QuantizationArgs,
     g_idx: Optional[torch.Tensor] = None,
 ) -> torch.Tensor:
+    """
+    Reshapes the value according to the quantization strategy for the purposes of
+    scale/zp calibration. The value after flattening has the following shape:
+
+    `(num_observations, *qparam_shape, group_size)`
+
+    The first dim is the number of observations (usually the batch size times number of
+    tokens), the middle dims are the dimension of the scales, and the last dim is the
+    number of elements being quantized per group.
+
+    :param value: value being flattened
+    :param base_name: weight, input, output, q/k/v. Used to characterize the value as
+        being a weight, activation, or attention state
+    :param args: quantization args for determining how the value is flattened
+    :param g_idx: optional gidx for weight activation ordering
+    :return: value which has been reshaped for calibration
+    """
     if base_name == "weight":
         return _flatten_weight(value, args, g_idx)
     elif base_name in ("input", "output"):

src/llmcompressor/observers/min_max.py

@@ -23,7 +23,7 @@ def __init__(
         module: Optional[torch.nn.Module] = None,
         **observer_kwargs,
     ):
-        super().__init__(module, base_name, args)
+        super().__init__(base_name, args, module, **observer_kwargs)
 
         observer_kwargs = self.args.observer_kwargs
         self.averaging_constant = observer_kwargs.get("averaging_constant", 0.01)

src/llmcompressor/observers/mse.py

@@ -23,7 +23,7 @@ def __init__(
         module: Optional[torch.nn.Module] = None,
         **observer_kwargs,
     ):
-        super().__init__(module, base_name, args)
+        super().__init__(base_name, args, module, **observer_kwargs)
 
         observer_kwargs = self.args.observer_kwargs
         self.maxshrink = observer_kwargs.get("maxshrink", 0.20)
@@ -51,8 +51,7 @@ def get_min_max(self, observed: torch.Tensor) -> Tuple[torch.Tensor, torch.Tenso
         )
         min_val = torch.ones_like(absolute_min_val)
         max_val = torch.zeros_like(absolute_max_val)
-
-        global_scale = getattr(self.parent(), f"{self.base_name}_global_scale", None)
+        global_scale = self._get_module_param("global_scale")
 
         # Early stopping params
         no_improve_count = 0