diff --git a/fbgemm_gpu/bench/sparse_ops_benchmark.py b/fbgemm_gpu/bench/sparse_ops_benchmark.py
index 3e19d32615..fdd0519096 100644
--- a/fbgemm_gpu/bench/sparse_ops_benchmark.py
+++ b/fbgemm_gpu/bench/sparse_ops_benchmark.py
@@ -287,13 +287,21 @@ def gen_inverse_index(curr_size: int, final_size: int) -> np.array:
 
     # Benchmark forward
     time_ref, output_ref = benchmark_torch_function(
-        torch.index_select, (input, 0, offset_indices), **bench_kwargs
+        # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+        # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
+        torch.index_select,
+        (input, 0, offset_indices),
+        # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+        # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
+        **bench_kwargs,
     )
 
     input_group = input.split(batch_size, 0)
     time, output_group = benchmark_torch_function(
         torch.ops.fbgemm.group_index_select_dim0,
         (input_group, indices_group),
+        # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+        # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
         **bench_kwargs,
     )
     logging.info(
@@ -306,13 +314,19 @@ def gen_inverse_index(curr_size: int, final_size: int) -> np.array:
     time_ref, _ = benchmark_torch_function(
         functools.partial(output_ref.backward, retain_graph=True),
         (grad,),
+        # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+        # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
         **bench_kwargs,
     )
 
+    # pyre-fixme[6]: For 1st argument expected `Union[List[Tensor],
+    #  typing.Tuple[Tensor, ...]]` but got `Tensor`.
     cat_output = torch.cat(output_group)
     time, _ = benchmark_torch_function(
         functools.partial(cat_output.backward, retain_graph=True),
         (grad,),
+        # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+        # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
         **bench_kwargs,
     )
     logging.info(
@@ -714,6 +728,8 @@ def batch_group_index_select_bwd(
         time_pyt, out_pyt = benchmark_torch_function(
             index_select_fwd_ref,
             (inputs, indices),
+            # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+            # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
             **bench_kwargs,
         )
 
@@ -726,12 +742,16 @@ def batch_group_index_select_bwd(
                 input_rows,
                 input_columns,
             ),
+            # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+            # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
             **bench_kwargs,
         )
 
         time_gis, out_gis = benchmark_torch_function(
             group_index_select_fwd,
             (gis_inputs, indices),
+            # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+            # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
             **bench_kwargs,
         )
 
@@ -746,6 +766,8 @@ def batch_group_index_select_bwd(
         time_bwd_pyt, _ = benchmark_torch_function(
             index_select_bwd_ref,
             (out_pyt, grads),
+            # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+            # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
             **bench_kwargs,
         )
 
@@ -756,6 +778,8 @@ def batch_group_index_select_bwd(
                 concat_grads,
                 optim_batch,
             ),
+            # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+            # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
             **bench_kwargs,
         )
 
@@ -766,6 +790,8 @@ def batch_group_index_select_bwd(
                 concat_grads,
                 optim_group,
             ),
+            # pyre-fixme[6]: For 3rd argument expected `bool` but got `int`.
+            # pyre-fixme[6]: For 3rd argument expected `str` but got `int`.
             **bench_kwargs,
         )
 
diff --git a/fbgemm_gpu/bench/split_table_batched_embeddings_benchmark.py b/fbgemm_gpu/bench/split_table_batched_embeddings_benchmark.py
index c76f4f4b05..a809b5e9b3 100644
--- a/fbgemm_gpu/bench/split_table_batched_embeddings_benchmark.py
+++ b/fbgemm_gpu/bench/split_table_batched_embeddings_benchmark.py
@@ -646,6 +646,9 @@ def run_bench(indices: Tensor, offsets: Tensor, per_sample_weights: Tensor) -> N
 
     time_per_iter = benchmark_requests(
         requests_uvm,
+        # pyre-fixme[6]: For 2nd argument expected `(Tensor, Tensor,
+        #  Optional[Tensor]) -> Tensor` but got `(indices: Tensor, offsets: Tensor,
+        #  per_sample_weights: Tensor) -> None`.
         run_bench,
         flush_gpu_cache_size_mb=flush_gpu_cache_size_mb,
         num_warmups=warmup_runs,
@@ -1934,6 +1937,9 @@ def nbit_uvm(
                 indices,
                 offsets,
             ),
+            # pyre-fixme[6]: For 3rd argument expected `(Tensor, Tensor,
+            #  Optional[Tensor]) -> None` but got `(indices: Any, offsets: Any,
+            #  indices_weights: Any) -> Tensor`.
             lambda indices, offsets, indices_weights: emb_mixed.forward(
                 indices,
                 offsets,
@@ -2421,6 +2427,9 @@ def nbit_cache(  # noqa C901
             indices,
             offsets,
         ),
+        # pyre-fixme[6]: For 3rd argument expected `(Tensor, Tensor,
+        #  Optional[Tensor]) -> None` but got `(indices: Any, offsets: Any,
+        #  indices_weights: Any) -> Tensor`.
         lambda indices, offsets, indices_weights: emb.forward(
             indices,
             offsets,
@@ -3061,6 +3070,7 @@ def device_with_spec(  # noqa C901
             reuse=reuse,
             alpha=alpha,
             weighted=weighted,
+            # pyre-fixme[61]: `sigma_Ls` is undefined, or not always defined.
             sigma_L=sigma_Ls[t] if use_variable_bag_sizes else None,
             zipf_oversample_ratio=3 if Ls[t] > 5 else 5,
         )
diff --git a/fbgemm_gpu/codegen/genscript/jinja_environment.py b/fbgemm_gpu/codegen/genscript/jinja_environment.py
index 7e82b65ae9..d94cac99a2 100644
--- a/fbgemm_gpu/codegen/genscript/jinja_environment.py
+++ b/fbgemm_gpu/codegen/genscript/jinja_environment.py
@@ -404,7 +404,10 @@ def replace_pta_namespace(pta_str_list: List[str]) -> List[str]:
 
 
 def replace_placeholder_types(
-    arg_str_list: List[str], type_combo: Optional[Dict[str, TensorType]]
+    # pyre-fixme[11]: Annotation `TensorType` is not defined as a type.
+    arg_str_list: List[str],
+    # pyre-fixme[11]: Annotation `TensorType` is not defined as a type.
+    type_combo: Optional[Dict[str, TensorType]],
 ) -> List[str]:
     """
     Replace the placeholder types with the primitive types
diff --git a/fbgemm_gpu/codegen/genscript/optimizer_args.py b/fbgemm_gpu/codegen/genscript/optimizer_args.py
index 3f7185c663..bd10fa8d6f 100644
--- a/fbgemm_gpu/codegen/genscript/optimizer_args.py
+++ b/fbgemm_gpu/codegen/genscript/optimizer_args.py
@@ -166,7 +166,11 @@ def schema_sym_int_arg_no_default(name: str) -> str:
 
 
 def make_kernel_arg(
-    ty: ArgType, name: str, default: Union[int, float, None], pass_by_ref: bool = False
+    # pyre-fixme[11]: Annotation `ArgType` is not defined as a type.
+    ty: ArgType,
+    name: str,
+    default: Union[int, float, None],
+    pass_by_ref: bool = False,
 ) -> str:
     return {
         ArgType.TENSOR: lambda x: acc_cache_tensor_arg(x, pass_by_ref=pass_by_ref),
@@ -318,6 +322,7 @@ class OptimizerArgs:
     split_variables: List[str]
     split_ref_kernel_args: List[str]
     placeholder_tensor_names: List[str]
+    # pyre-fixme[11]: Annotation `TensorType` is not defined as a type.
     placeholder_type_combos: Union[List[Dict[str, TensorType]], List[None]]
 
     @staticmethod
@@ -345,6 +350,7 @@ def create(
         else:
             ph_combos = [None]
 
+        # pyre-fixme[28]: Unexpected keyword argument `placeholder_type_combos`.
         return OptimizerArgs(
             # GPU kernel args
             split_kernel_args=[
@@ -434,6 +440,7 @@ def create_optim_args(
         split_arg_spec = []
         for s in arg_spec:
             if s.ty in (ArgType.FLOAT, ArgType.INT, ArgType.SYM_INT):
+                # pyre-fixme[19]: Expected 1 positional argument.
                 split_arg_spec.append(OptimItem(s.ty, s.name, s.default))
             else:
                 assert s.ty in (ArgType.TENSOR, ArgType.PLACEHOLDER_TENSOR)
@@ -446,8 +453,11 @@ def extend_for_cpu(spec: OptimItem) -> List[OptimItem]:
         name = spec.name
         default = spec.default
         return [
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.TENSOR, f"{name}_host", default),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.INT_TENSOR, f"{name}_placements", default),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.LONG_TENSOR, f"{name}_offsets", default),
         ]
 
@@ -458,9 +468,13 @@ def extend_for_cuda(spec: OptimItem) -> List[OptimItem]:
         ty = spec.ty
         ph_tys = spec.ph_tys
         return [
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ty, f"{name}_dev", default, ph_tys),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ty, f"{name}_uvm", default, ph_tys),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.INT_TENSOR, f"{name}_placements", default),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.LONG_TENSOR, f"{name}_offsets", default),
         ]
 
@@ -471,10 +485,15 @@ def extend_for_any(spec: OptimItem) -> List[OptimItem]:
         ty = spec.ty
         ph_tys = spec.ph_tys
         return [
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.TENSOR, f"{name}_host", default),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ty, f"{name}_dev", default, ph_tys),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ty, f"{name}_uvm", default, ph_tys),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.INT_TENSOR, f"{name}_placements", default),
+            # pyre-fixme[19]: Expected 1 positional argument.
             OptimItem(ArgType.LONG_TENSOR, f"{name}_offsets", default),
         ]
 
diff --git a/fbgemm_gpu/fbgemm_gpu/quantize_comm.py b/fbgemm_gpu/fbgemm_gpu/quantize_comm.py
index fb40bad8a0..ca2ebd02d6 100644
--- a/fbgemm_gpu/fbgemm_gpu/quantize_comm.py
+++ b/fbgemm_gpu/fbgemm_gpu/quantize_comm.py
@@ -50,7 +50,11 @@
 
 
 def none_throws(
-    optional: Optional[TypeVar("_T")], message: str = "Unexpected `None`"
+    # pyre-fixme[31]: Expression `typing.Optional[typing.TypeVar("_T")]` is not a
+    #  valid type.
+    optional: Optional[TypeVar("_T")],
+    message: str = "Unexpected `None`",
+    # pyre-fixme[31]: Expression `typing.TypeVar("_T")` is not a valid type.
 ) -> TypeVar("_T"):
     if optional is None:
         raise AssertionError(message)
diff --git a/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_inference.py b/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_inference.py
index a9af3f0fb1..f8b691be4e 100644
--- a/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_inference.py
+++ b/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_inference.py
@@ -162,10 +162,15 @@ class IntNBitTableBatchedEmbeddingBagsCodegen(nn.Module):
 
     embedding_specs: List[Tuple[str, int, int, SparseType, EmbeddingLocation]]
     record_cache_metrics: RecordCacheMetrics
+    # pyre-fixme[13]: Attribute `cache_miss_counter` is never initialized.
     cache_miss_counter: torch.Tensor
+    # pyre-fixme[13]: Attribute `uvm_cache_stats` is never initialized.
     uvm_cache_stats: torch.Tensor
+    # pyre-fixme[13]: Attribute `local_uvm_cache_stats` is never initialized.
     local_uvm_cache_stats: torch.Tensor
+    # pyre-fixme[13]: Attribute `weights_offsets` is never initialized.
     weights_offsets: torch.Tensor
+    # pyre-fixme[13]: Attribute `weights_placements` is never initialized.
     weights_placements: torch.Tensor
 
     def __init__(  # noqa C901
diff --git a/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_training.py b/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_training.py
index 604509b575..3c824a4222 100644
--- a/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_training.py
+++ b/fbgemm_gpu/fbgemm_gpu/split_table_batched_embeddings_ops_training.py
@@ -343,9 +343,12 @@ class SplitTableBatchedEmbeddingBagsCodegen(nn.Module):
     lxu_cache_locations_empty: Tensor
     timesteps_prefetched: List[int]
     record_cache_metrics: RecordCacheMetrics
+    # pyre-fixme[13]: Attribute `uvm_cache_stats` is never initialized.
     uvm_cache_stats: torch.Tensor
+    # pyre-fixme[13]: Attribute `local_uvm_cache_stats` is never initialized.
     local_uvm_cache_stats: torch.Tensor
     uuid: str
+    # pyre-fixme[13]: Attribute `last_uvm_cache_print_state` is never initialized.
     last_uvm_cache_print_state: torch.Tensor
     _vbe_B_offsets: Optional[torch.Tensor]
     _vbe_max_B: int
diff --git a/fbgemm_gpu/fbgemm_gpu/tbe/ssd/training.py b/fbgemm_gpu/fbgemm_gpu/tbe/ssd/training.py
index 9595c39397..04fac4a62b 100644
--- a/fbgemm_gpu/fbgemm_gpu/tbe/ssd/training.py
+++ b/fbgemm_gpu/fbgemm_gpu/tbe/ssd/training.py
@@ -1431,6 +1431,7 @@ def forward(
         offsets: Tensor,
         per_sample_weights: Optional[Tensor] = None,
         feature_requires_grad: Optional[Tensor] = None,
+        # pyre-fixme[7]: Expected `Tensor` but got implicit return value of `None`.
     ) -> Tensor:
         indices, offsets, per_sample_weights = self.prepare_inputs(
             indices, offsets, per_sample_weights
diff --git a/fbgemm_gpu/fbgemm_gpu/tbe/utils/quantize.py b/fbgemm_gpu/fbgemm_gpu/tbe/utils/quantize.py
index daff7a7132..99d8a28011 100644
--- a/fbgemm_gpu/fbgemm_gpu/tbe/utils/quantize.py
+++ b/fbgemm_gpu/fbgemm_gpu/tbe/utils/quantize.py
@@ -79,6 +79,7 @@ def dequantize_embs(
     weight_ty: SparseType,
     use_cpu: bool,
     fp8_config: Optional[FP8QuantizationConfig] = None,
+    # pyre-fixme[7]: Expected `Tensor` but got implicit return value of `None`.
 ) -> torch.Tensor:
     print(f"weight_ty: {weight_ty}")
     assert (