Minor lint fixes (#236)

generate.py

@@ -435,7 +435,9 @@ def callback(x):
         import contextlib
 
         generator_args.encoded_prompt = encoded
-        if (i != generator_args.num_samples - 1 or not profile) or (use_tp and rank != 0):
+        if (i != generator_args.num_samples - 1 or not profile) or (
+            use_tp and rank != 0
+        ):
             prof = contextlib.nullcontext()
         else:
             torch.profiler._utils._init_for_cuda_graphs()
@@ -495,7 +497,7 @@ def main(args):
     speculative_builder_args = BuilderArgs.from_speculative_args(args)
     tokenizer_args = TokenizerArgs.from_args(args)
     generator_args = GeneratorArgs.from_args(args)
-    
+
     _main(
         builder_args,
         speculative_builder_args,

quantize.py

@@ -515,11 +515,7 @@ def forward(self, input: torch.Tensor) -> torch.Tensor:
 
 
 def replace_embedding_weight_only_grouped_int8_per_channel(
-        module,
-        device,
-        bitwidth: int = 8,
-        groupsize: Optional[int] = None,
-        packed=False
+    module, device, bitwidth: int = 8, groupsize: Optional[int] = None, packed=False
 ):
     for name, child in module.named_children():
         # print(f"name: {name}")