lint

Summary:

Test Plan:

Reviewers:

Subscribers:

Tasks:

Tags:

tritonbench/operators/ragged_attention/hstu.py

@@ -79,7 +79,11 @@ def __init__(
         self.persistent_kernel = persistent_kernel
 
     def forward(
-        self, qkv: torch.Tensor, seq_offsets: torch.Tensor, timestamps: torch.Tensor, num_targets: torch.Tensor
+        self,
+        qkv: torch.Tensor,
+        seq_offsets: torch.Tensor,
+        timestamps: torch.Tensor,
+        num_targets: torch.Tensor,
     ) -> torch.Tensor:
         NUM_BUCKETS = self.num_buckets
         torch._check(timestamps.size(0) + 1 == seq_offsets.size(0))
@@ -215,7 +219,13 @@ def generate_sparse_seq_len(
 
 
 def get_test_inputs(
-    batch_size, num_heads, max_seq_len, sparsity, target_size, sort_by_length, requires_grad
+    batch_size,
+    num_heads,
+    max_seq_len,
+    sparsity,
+    target_size,
+    sort_by_length,
+    requires_grad,
 ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
     timestamp_deltas: torch.Tensor = torch.randint(
         86400,

tritonbench/operators/ragged_attention/operator.py

@@ -64,7 +64,9 @@ def hstu_triton_ragged_attention(self, qkv, seq_offsets, timestamps, num_targets
 
     # TODO: enable persistent kernels when the OSS backward is ready
     @register_benchmark(enabled=False)
-    def hstu_triton_ragged_attention_persistent(self, qkv, seq_offsets, timestamps, num_targets):
+    def hstu_triton_ragged_attention_persistent(
+        self, qkv, seq_offsets, timestamps, num_targets
+    ):
         attn = RaggedHSTUAttn(
             self.batch_size,
             self.num_heads,
@@ -79,12 +81,26 @@ def hstu_triton_ragged_attention_persistent(self, qkv, seq_offsets, timestamps,
         return lambda: attn(qkv, seq_offsets, timestamps, num_targets)
 
     def get_x_val(self, example_inputs):
-        return (self.batch_size, self.num_heads, self.max_seq_len, self.num_buckets, self.sparsity, self.target_size, self.sort_by_length)
+        return (
+            self.batch_size,
+            self.num_heads,
+            self.max_seq_len,
+            self.num_buckets,
+            self.sparsity,
+            self.target_size,
+            self.sort_by_length,
+        )
 
     def get_input_iter(self):
         for _input_id in range(self._num_inputs):
             inputs = get_test_inputs(
-                self.batch_size, self.num_heads, self.max_seq_len, self.sparsity, self.target_size, self.sort_by_length, self.requires_grad
+                self.batch_size,
+                self.num_heads,
+                self.max_seq_len,
+                self.sparsity,
+                self.target_size,
+                self.sort_by_length,
+                self.requires_grad,
             )
             yield inputs