Some renaming

server/text_generation_server/layers/marlin.py

@@ -59,7 +59,7 @@ def _get_perms() -> Tuple[List[int], List[int]]:
 _scale_perm, _scale_perm_single = _get_perms()
 
 
-def permute_scales(scales: torch.Tensor, in_features: int, group_size: int):
+def permute_scales(scales: torch.Tensor):
     out_features = scales.shape[1]
     if scales.shape[0] == 1:
         scales = scales.reshape((-1, len(_scale_perm_single)))[:, _scale_perm_single]
@@ -95,8 +95,8 @@ def repack_gptq_for_marlin(
     g_idx: torch.Tensor,
     bits: int,
     desc_act: bool,
-    group_size: int,
-    is_sym: bool,
+    groupsize: int,
+    sym: bool,
     sharded_infeatures: bool,
 ) -> GPTQMarlinWeight:
     """Convert GPTQ weights to a layout that's compatible with GPTQ-Marlin kernels."""
@@ -109,12 +109,12 @@ def repack_gptq_for_marlin(
             f"Repacking {bits}-bit GPTQ weights as Marlin is not supported, must be one of: {supported_bits}"
         )
 
-    if group_size not in GPTQ_MARLIN_GROUP_SIZES:
+    if groupsize not in GPTQ_MARLIN_GROUP_SIZES:
         supported_sizes = ", ".join(str(b) for b in GPTQ_MARLIN_GROUP_SIZES)
         raise RuntimeError(
-            f"Repacking GPTQ weights with group size {group_size} as Marlin is not supported, must be one of: {supported_sizes}"
+            f"Repacking GPTQ weights with group size {groupsize} as Marlin is not supported, must be one of: {supported_sizes}"
         )
-    if not is_sym:
+    if not sym:
         raise RuntimeError(
             "Repacking GPTQ weights with asymmetric quantization as Marlin is not supported."
         )
@@ -123,12 +123,12 @@ def repack_gptq_for_marlin(
     in_features = qweight.shape[0] * weights_per_int
     out_features = qweight.shape[1]
 
-    if in_features % group_size != 0:
+    if in_features % groupsize != 0:
         raise ValueError(
-            f"Number of input features ({in_features}) not divisible by group size ({group_size})"
+            f"Number of input features ({in_features}) not divisible by group size ({groupsize})"
         )
 
-    if desc_act and group_size != -1:
+    if desc_act and groupsize != -1:
         perm = torch.argsort(g_idx).to(torch.int)
         g_idx = g_idx[perm]
     else:
@@ -139,7 +139,7 @@ def repack_gptq_for_marlin(
         qweight, perm, in_features, out_features, bits
     )
 
-    scales = permute_scales(scales, in_features, group_size)
+    scales = permute_scales(scales)
 
     is_full_k = not (desc_act and sharded_infeatures)
 
@@ -249,11 +249,11 @@ def __init__(self, *, weight: MarlinWeight, bias: Optional[torch.Tensor]):
             out_features % 256 == 0
         ), f"Number of output features ({out_features}) not divisable by 256"
 
-        group_size = -1 if weight.s.shape[0] == 1 else in_features // weight.s.shape[0]
-        assert group_size in {
+        groupsize = -1 if weight.s.shape[0] == 1 else in_features // weight.s.shape[0]
+        assert groupsize in {
             -1,
             128,
-        }, f"Group size must be -1 or 128, was {group_size}"
+        }, f"Group size must be -1 or 128, was {groupsize}"
 
         self.register_buffer("B", weight.B)
         self.register_buffer("s", weight.s)

server/text_generation_server/utils/weights.py

@@ -251,8 +251,8 @@ def get_weights_col_packed(
                     g_idx=g_idx,
                     bits=gptq_params.bits,
                     desc_act=gptq_params.desc_act,
-                    group_size=gptq_params.groupsize,
-                    is_sym=gptq_params.sym,
+                    groupsize=gptq_params.groupsize,
+                    sym=gptq_params.sym,
                     sharded_infeatures=False,
                 )
 
@@ -416,8 +416,8 @@ def get_multi_weights_col(self, prefixes: List[str], quantize: str, dim: int):
                     g_idx=g_idx,
                     bits=gptq_params.bits,
                     desc_act=gptq_params.desc_act,
-                    group_size=gptq_params.groupsize,
-                    is_sym=gptq_params.sym,
+                    groupsize=gptq_params.groupsize,
+                    sym=gptq_params.sym,
                     sharded_infeatures=False,
                 )
             else:
@@ -638,8 +638,8 @@ def get_multi_weights_row(self, prefix: str, quantize: str):
                     g_idx=g_idx,
                     bits=gptq_params.bits,
                     desc_act=gptq_params.desc_act,
-                    group_size=gptq_params.groupsize,
-                    is_sym=gptq_params.sym,
+                    groupsize=gptq_params.groupsize,
+                    sym=gptq_params.sym,
                     sharded_infeatures=sharded_in_features,
                 )
             else:
@@ -652,7 +652,7 @@ def get_multi_weights_row(self, prefix: str, quantize: str):
 
                 num_groups = self._get_slice(f"{prefix}.s").get_shape()[0]
                 if num_groups == 1:
-                    # The number of groups is 1 when group_size == -1. share
+                    # The number of groups is 1 when groupsize == -1. share
                     # scales between all shards in this case.
                     s = self.get_tensor(f"{prefix}.s")
                 else: