[ET-VK] Enable custom rotary embedding module replacement (#6424)

Pull Request resolved: #6393

## Context

Implements a module replacement source transform to use the rotary embedding custom op introduced in the previous diff in the Llama model.
ghstack-source-id: 249175727
@exported-using-ghexport

Differential Revision: [D64697590](https://our.internmc.facebook.com/intern/diff/D64697590/)

Co-authored-by: Stephen Jia <[email protected]>

Author: kirklandsign

backends/vulkan/_passes/custom_ops_defs.py

@@ -183,3 +183,53 @@ def linear_weight_int4_impl(
 )
 lib.impl(name, linear_weight_int4_impl, "CompositeExplicitAutograd")
 linear_weight_int4_op = getattr(getattr(torch.ops, namespace), name)
+
+######################
+## apply_rotary_emb ##
+######################
+
+
+# Note that this implementation is copied from executorch.examples.models.llama.rope
+# but it is copied here to avoid introducing a dependency on the llama code.
+def apply_rotary_emb_impl(
+    xq: torch.Tensor, xk: torch.Tensor, freqs_cos: torch.Tensor, freqs_sin: torch.Tensor
+):
+    def reshape_for_broadcast(freqs_cis: torch.Tensor, x: torch.Tensor):
+        ndim = x.ndim
+        freqs_cis_ndim = freqs_cis.ndim
+        if freqs_cis_ndim == 3:
+            # freqs_cis: (seq_len, n_heads, head_dim // 2)
+            assert freqs_cis.shape == (x.shape[-3], x.shape[-2], x.shape[-1])
+            shape = [
+                d if (i == ndim - 3 or i == ndim - 2 or i == ndim - 1) else 1
+                for i, d in enumerate(x.shape)
+            ]
+        else:
+            # freqs_cis: (seq_len, head_dim // 2)
+            assert freqs_cis.shape == (x.shape[1], x.shape[-1])
+            shape = [d if i == 1 or i == ndim - 1 else 1 for i, d in enumerate(x.shape)]
+        return freqs_cis.view(shape)
+
+    xq_r, xq_i = xq.float().reshape(xq.shape[:-1] + (-1, 2)).unbind(-1)
+    xk_r, xk_i = xk.float().reshape(xk.shape[:-1] + (-1, 2)).unbind(-1)
+
+    freqs_cos = reshape_for_broadcast(freqs_cos, xq_r)
+    freqs_sin = reshape_for_broadcast(freqs_sin, xq_r)
+
+    xq_out_r = xq_r * freqs_cos - xq_i * freqs_sin
+    xq_out_i = xq_r * freqs_sin + xq_i * freqs_cos
+    xk_out_r = xk_r * freqs_cos - xk_i * freqs_sin
+    xk_out_i = xk_r * freqs_sin + xk_i * freqs_cos
+
+    xq_out = torch.stack([xq_out_r, xq_out_i], dim=-1).flatten(3)
+    xk_out = torch.stack([xk_out_r, xk_out_i], dim=-1).flatten(3)
+
+    return xq_out.type_as(xq), xk_out.type_as(xk)
+
+
+name = "apply_rotary_emb"
+lib.define(
+    f"{name}(Tensor xq, Tensor xk, Tensor freqs_cos, Tensor freqs_sin) -> (Tensor, Tensor)"
+)
+lib.impl(name, apply_rotary_emb_impl, "CompositeExplicitAutograd")
+apply_rotary_emb_op = getattr(getattr(torch.ops, namespace), name)

backends/vulkan/partitioner/supported_ops.py

@@ -94,8 +94,9 @@ def __contains__(self, op):
     # Convolution
     exir_ops.edge.aten.convolution.default,
     exir_ops.edge.et_vk.conv_with_clamp.default,
-    # Custom ops
+    # Llama ops
     "llama::sdpa_with_kv_cache",
+    exir_ops.edge.et_vk.apply_rotary_emb.default,
 ]
 
 NO_DYNAMIC_SHAPE = [

examples/models/llama/TARGETS

@@ -91,6 +91,7 @@ runtime.python_library(
         "source_transformation/rope.py",
         "source_transformation/sdpa.py",
         "source_transformation/spin_quant.py",
+        "source_transformation/vulkan_rope.py",
     ],
     _is_external_target = True,
     base_module = "executorch.examples.models.llama",

examples/models/llama/export_llama_lib.py

@@ -69,6 +69,7 @@
     replace_sdpa_with_flex_sdpa,
     replace_sdpa_with_simple_sdpa,
 )
+from .source_transformation.vulkan_rope import replace_with_vulkan_rotary_emb
 
 IS_FBCODE = True  #  os.environ.get("FBCODE_PLATFORM", False)
 FORMAT = "[%(levelname)s %(asctime)s %(filename)s:%(lineno)s] %(message)s"
@@ -943,4 +944,7 @@ def _get_source_transforms(  # noqa
                 transforms.append(replace_sdpa_with_simple_sdpa)
             transforms.append(replace_kv_cache_with_coreml_kv_cache)
 
+    if args.vulkan:
+        transforms.append(replace_with_vulkan_rotary_emb)
+
     return transforms

examples/models/llama/source_transformation/vulkan_rope.py

@@ -0,0 +1,39 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+import torch
+
+from executorch.backends.vulkan._passes.custom_ops_defs import (  # noqa
+    apply_rotary_emb_op,
+)
+
+from executorch.examples.models.llama.rope import RotaryEmbedding
+
+
+class VkRotaryEmbedding(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    def forward(
+        self,
+        xq: torch.Tensor,
+        xk: torch.Tensor,
+        freqs_cos: torch.Tensor,
+        freqs_sin: torch.Tensor,
+    ):
+        xq_out, xk_out = torch.ops.et_vk.apply_rotary_emb(xq, xk, freqs_cos, freqs_sin)
+        return xq_out, xk_out
+
+
+def replace_with_vulkan_rotary_emb(module: torch.nn.Module):
+    for name, child in module.named_children():
+        if isinstance(child, RotaryEmbedding):
+            new_module = VkRotaryEmbedding()
+            setattr(module, name, new_module)
+        else:
+            replace_with_vulkan_rotary_emb(child)
+
+    return module