oops fix double indent on quantize def

export.py

@@ -37,30 +37,30 @@ def serialize_int8(file, tensor):
     file.write(b)
 
 def quantize_q80(w, group_size):
-        """
+    """
-        takes a tensor and returns the Q8_0 quantized version
+    takes a tensor and returns the Q8_0 quantized version
-        i.e. symmetric quantization into int8, range [-127,127]
+    i.e. symmetric quantization into int8, range [-127,127]
-        """
+    """
-        assert w.numel() % group_size == 0
+    assert w.numel() % group_size == 0
-        ori_shape = w.shape
+    ori_shape = w.shape
-        w = w.float() # convert to float32
+    w = w.float() # convert to float32
-        w = w.reshape(-1, group_size)
+    w = w.reshape(-1, group_size)
-        # find the max in each group
+    # find the max in each group
-        wmax = torch.abs(w).max(dim=1).values
+    wmax = torch.abs(w).max(dim=1).values
-        # calculate the scaling factor such that float = quant * scale
+    # calculate the scaling factor such that float = quant * scale
-        scale = wmax / 127.0
+    scale = wmax / 127.0
-        # scale into range [-127, 127]
+    # scale into range [-127, 127]
-        quant = w / scale[:,None]
+    quant = w / scale[:,None]
-        # round to nearest integer
+    # round to nearest integer
-        int8val = torch.round(quant).to(torch.int8)
+    int8val = torch.round(quant).to(torch.int8)
-        # dequantize by rescaling
+    # dequantize by rescaling
-        fp32val = (int8val.float() * scale[:,None]).view(-1)
+    fp32val = (int8val.float() * scale[:,None]).view(-1)
-        fp32valr = fp32val.reshape(-1, group_size)
+    fp32valr = fp32val.reshape(-1, group_size)
-        # calculate the max error in each group
+    # calculate the max error in each group
-        err = torch.abs(fp32valr - w).max(dim=1).values
+    err = torch.abs(fp32valr - w).max(dim=1).values
-        # find the max error across all groups
+    # find the max error across all groups
-        maxerr = err.max().item()
+    maxerr = err.max().item()
-        return int8val, scale, maxerr
+    return int8val, scale, maxerr
 
 # -----------------------------------------------------------------------------
 # legacy