From 50febbe2424f39c29b9b30691472c25ac6cd9867 Mon Sep 17 00:00:00 2001
From: Basil Wong <basilwong@meta.com>
Date: Wed, 29 Jan 2025 08:55:26 -0800
Subject: [PATCH] Updating split_table_batched_embeddings_ops_training.py
 (#3613)

Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/690


After this diff stack:

EmbeddingKernelConfig now supports adding embedding_table_int32_index_type and embedding_table_int32_offset_type to the fused_params.

These are used downstream to determine whether the indices and offsets types for split_table_batched_embeddings_ops_training.py

Reviewed By: q10

Differential Revision: D68005929
---
 ...t_table_batched_embeddings_ops_training.py | 30 +++++++++++++++++++
 1 file changed, 30 insertions(+)

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 65a21c99c5..70693def2c 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
@@ -586,6 +586,14 @@ class SplitTableBatchedEmbeddingBagsCodegen(nn.Module):
             are not enabled by default.
             - `use_rowwise_bias_correction` is used in Adam to enable rowwise
                 bias correction computation
+
+        embedding_table_index_type (torch.dtype = torch.int64): The data type of
+            the embedding table index tensor. Options are `torch.int32` and
+            `torch.int64`
+
+        embedding_table_offset_type (torch.dtype = torch.int64): The data type of
+            the embedding table offset tensor. Options are `torch.int32` and
+            `torch.int64`
     """
 
     embedding_specs: List[Tuple[int, int, EmbeddingLocation, ComputeDevice]]
@@ -654,6 +662,8 @@ def __init__(  # noqa C901
         uvm_host_mapped: bool = False,
         extra_optimizer_config: Optional[UserEnabledConfigDefinition] = None,
         tbe_input_multiplexer_config: Optional[TBEInputMultiplexerConfig] = None,
+        embedding_table_index_type: torch.dtype = torch.int64,
+        embedding_table_offset_type: torch.dtype = torch.int64,
     ) -> None:
         super(SplitTableBatchedEmbeddingBagsCodegen, self).__init__()
 
@@ -1343,6 +1353,17 @@ def __init__(  # noqa C901
             FeatureGateName.BOUNDS_CHECK_INDICES_V2
         )
 
+        if embedding_table_index_type not in [torch.int32, torch.int64]:
+            raise ValueError(
+                f"embedding_table_index_type must be torch.int32 or torch.int64, but got {embedding_table_index_type}"
+            )
+        self.embedding_table_index_type: torch.dtype = embedding_table_index_type
+        if embedding_table_offset_type not in [torch.int32, torch.int64]:
+            raise ValueError(
+                f"embedding_table_offset_type must be torch.int32 or torch.int64, but got {embedding_table_offset_type}"
+            )
+        self.embedding_table_offset_type: torch.dtype = embedding_table_offset_type
+
     @torch.jit.ignore
     def log(self, msg: str) -> None:
         """
@@ -3409,6 +3430,15 @@ def prepare_inputs(
             # NOTE: Force offsets to have the same dtype as indices since the
             # kernels assume same dtype.  We might need to revisit the assumption
             # of same dtypes in the future.
+            if self.embedding_table_index_type == torch.int32:
+                self.log(
+                    "Casting indices to int32 based on embedding_table_index_type input."
+                )
+                indices = indices.to(torch.int32)
+            if self.embedding_table_index_type != self.embedding_table_offset_type:
+                self.log(
+                    f"Force casting offsets to {self.embedding_table_index_type} so that it is the same as the indices type."
+                )
             offsets = offsets.to(dtype=indices.dtype)
 
             # Force casting per_sample_weights to float