Add Vulkan variants (fp16, int8)

Exported with the conv1d-as-conv2d lowering, the slice-based STFT magnitude,
hoisted cross-attention K/V and a head-major KV layout. Requires the ExecuTorch
1.4.1 Vulkan runtime from the v0.10.0-libs-1.4.1 artifacts (gelu clamp, reduce
barrier and Adreno local-work-group fixes).

.gitattributes

@@ -38,3 +38,5 @@ xnnpack/whisper_tiny_xnnpack_fp32.pte filter=lfs diff=lfs merge=lfs -text
 coreml/whisper_tiny_coreml_fp16.pte filter=lfs diff=lfs merge=lfs -text
 mlx/whisper_tiny_mlx_bf16.pte filter=lfs diff=lfs merge=lfs -text
 mlx/whisper_tiny_mlx_int8.pte filter=lfs diff=lfs merge=lfs -text
+vulkan/whisper_tiny_vulkan_fp16.pte filter=lfs diff=lfs merge=lfs -text
+vulkan/whisper_tiny_vulkan_int8.pte filter=lfs diff=lfs merge=lfs -text

vulkan/config.json

@@ -0,0 +1,137 @@
+{
+  "$schema": "https://huggingface.co/software-mansion/react-native-executorch-spec/resolve/main/config.schema.json",
+  "model": "whisper_tiny",
+  "family": "whisper",
+  "capabilities": [
+    "speech-to-text"
+  ],
+  "backend": "vulkan",
+  "license": "apache-2.0",
+  "variants": [
+    {
+      "file": "whisper_tiny_vulkan_fp16.pte",
+      "precision": "fp16",
+      "quantized": false,
+      "default": true,
+      "methods": {
+        "decode": {
+          "inputs": [
+            {
+              "shape": [
+                1,
+                1
+              ],
+              "dtype": "int64"
+            },
+            {
+              "shape": [
+                1
+              ],
+              "dtype": "int64"
+            },
+            {
+              "shape": [
+                1,
+                1500,
+                3072
+              ],
+              "dtype": "float32"
+            }
+          ],
+          "outputs": [
+            {
+              "shape": [
+                1,
+                1,
+                51865
+              ],
+              "dtype": "float32"
+            }
+          ]
+        },
+        "encode": {
+          "inputs": [
+            {
+              "shape": [
+                -1
+              ],
+              "dtype": "float32"
+            }
+          ],
+          "outputs": [
+            {
+              "shape": [
+                1,
+                1500,
+                3072
+              ],
+              "dtype": "float32"
+            }
+          ]
+        }
+      }
+    },
+    {
+      "file": "whisper_tiny_vulkan_int8.pte",
+      "precision": "int8",
+      "quantized": true,
+      "methods": {
+        "decode": {
+          "inputs": [
+            {
+              "shape": [
+                1,
+                1
+              ],
+              "dtype": "int64"
+            },
+            {
+              "shape": [
+                1
+              ],
+              "dtype": "int64"
+            },
+            {
+              "shape": [
+                1,
+                1500,
+                3072
+              ],
+              "dtype": "float32"
+            }
+          ],
+          "outputs": [
+            {
+              "shape": [
+                1,
+                1,
+                51865
+              ],
+              "dtype": "float32"
+            }
+          ]
+        },
+        "encode": {
+          "inputs": [
+            {
+              "shape": [
+                -1
+              ],
+              "dtype": "float32"
+            }
+          ],
+          "outputs": [
+            {
+              "shape": [
+                1,
+                1500,
+                3072
+              ],
+              "dtype": "float32"
+            }
+          ]
+        }
+      }
+    }
+  ]
+}

vulkan/whisper_tiny_vulkan_fp16.pte

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7d07de41bcbe76a4b732dd0c848dee023c9c9f7079806713478633cc7d3560
+size 213753984

vulkan/whisper_tiny_vulkan_int8.pte

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9014d2ceb0c6a5bafe94086274556b11575d37840b2ae9d4d9aec31b90821de
+size 154213348