chore: add pre-commit (#1569)

server/exllamav2_kernels/exllamav2_kernels/cuda/quant/qdq_6.cuh

@@ -40,5 +40,3 @@ __forceinline__ __device__ void dequant_6bit_16
 #endif
 
 #endif
-
-

server/text_generation_server/models/custom_modeling/flash_llama_modeling.py

@@ -251,9 +251,9 @@ class LlamaMLP(nn.Module):
             if "gelu" not in act
             else lambda x: torch.nn.functional.gelu(
                 x,
-                approximate="tanh"
-                if act in ["gelu_fast", "gelu_pytorch_tanh"]
-                else "none",
+                approximate=(
+                    "tanh" if act in ["gelu_fast", "gelu_pytorch_tanh"] else "none"
+                ),
             )
         )
         # Fuse gate and up proj

server/text_generation_server/models/custom_modeling/flash_mistral_modeling.py

@@ -255,9 +255,9 @@ class MistralMLP(nn.Module):
             if "gelu" not in act
             else lambda x: torch.nn.functional.gelu(
                 x,
-                approximate="tanh"
-                if act in ["gelu_fast", "gelu_pytorch_tanh"]
-                else "none",
+                approximate=(
+                    "tanh" if act in ["gelu_fast", "gelu_pytorch_tanh"] else "none"
+                ),
             )
         )
         # Fuse gate and up proj

server/text_generation_server/models/custom_modeling/flash_mixtral_modeling.py

@@ -344,9 +344,9 @@ class BlockSparseMoE(nn.Module):
         if "gelu" in act:
             self.act = lambda x: torch.nn.functional.gelu(
                 x,
-                approximate="tanh"
-                if act in ["gelu_fast", "gelu_pytorch_tanh"]
-                else "none",
+                approximate=(
+                    "tanh" if act in ["gelu_fast", "gelu_pytorch_tanh"] else "none"
+                ),
             )
         elif "silu" in act:
             self.act = torch.nn.functional.silu
@@ -600,9 +600,9 @@ class DenseMoE(nn.Module):
         if "gelu" in act:
             self.act = lambda x: torch.nn.functional.gelu(
                 x,
-                approximate="tanh"
-                if act in ["gelu_fast", "gelu_pytorch_tanh"]
-                else "none",
+                approximate=(
+                    "tanh" if act in ["gelu_fast", "gelu_pytorch_tanh"] else "none"
+                ),
             )
         elif "silu" in act:
             self.act = torch.nn.functional.silu

server/text_generation_server/models/custom_modeling/flash_neox_modeling.py

@@ -187,9 +187,9 @@ class FlashMLP(nn.Module):
             if "gelu" not in act
             else lambda x: torch.nn.functional.gelu(
                 x,
-                approximate="tanh"
-                if act in ["gelu_fast", "gelu_pytorch_tanh"]
-                else "none",
+                approximate=(
+                    "tanh" if act in ["gelu_fast", "gelu_pytorch_tanh"] else "none"
+                ),
             )
         )
 

server/text_generation_server/models/custom_modeling/flash_phi_modeling.py

@@ -225,9 +225,9 @@ class PhiMLP(nn.Module):
             if "gelu" not in act
             else lambda x: torch.nn.functional.gelu(
                 x,
-                approximate="tanh"
-                if act in ["gelu_fast", "gelu_pytorch_tanh"]
-                else "none",
+                approximate=(
+                    "tanh" if act in ["gelu_fast", "gelu_pytorch_tanh"] else "none"
+                ),
             )
         )