Merge branch 'main' into exir-flatbuffer-serialize-fastpath_v2

Author: JacobSzwejbka

.ci/scripts/build-qnn-sdk.sh

@@ -18,7 +18,7 @@ build_qnn_backend() {
   export EXECUTORCH_ROOT="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")/../.." && pwd)"
 
   parallelism=$(( $(nproc) - 1 ))
-  bash backends/qualcomm/scripts/build.sh --skip_linux_android --skip_linux_embedded --job_number ${parallelism} --release
+  bash backends/qualcomm/scripts/build.sh --skip_linux_android --job_number ${parallelism} --release
 }
 
 set_up_aot() {

.ci/scripts/export_model_artifact.sh

@@ -22,6 +22,7 @@ Arguments:
                  - mistralai/Voxtral-Mini-4B-Realtime-2602
                  - openai/whisper series (whisper-{small, medium, large, large-v2, large-v3, large-v3-turbo})
                  - google/gemma-3-4b-it
+                 - nvidia/diar_streaming_sortformer_4spk-v2
                  - nvidia/parakeet-tdt
 
   quant_name   Quantization type (optional, default: non-quantized)
@@ -45,6 +46,7 @@ Examples:
   export_model_artifact.sh metal "mistralai/Voxtral-Mini-4B-Realtime-2602" "quantized-int4-metal"
   export_model_artifact.sh metal "mistralai/Voxtral-Mini-4B-Realtime-2602" "non-quantized" "." "vr-streaming"
   export_model_artifact.sh cuda "mistralai/Voxtral-Mini-3B-2507" "quantized-int4-tile-packed"
+  export_model_artifact.sh cuda-windows "nvidia/diar_streaming_sortformer_4spk-v2" "non-quantized" "./output"
   export_model_artifact.sh cuda "google/gemma-3-4b-it" "non-quantized" "./output"
   export_model_artifact.sh cuda "nvidia/parakeet-tdt" "non-quantized" "./output"
   export_model_artifact.sh xnnpack "nvidia/parakeet-tdt" "quantized-8da4w" "./output"
@@ -157,6 +159,14 @@ case "$HF_MODEL" in
     PREPROCESSOR_FEATURE_SIZE=""
     PREPROCESSOR_OUTPUT=""
     ;;
+  nvidia/diar_streaming_sortformer_4spk-v2)
+    MODEL_NAME="sortformer"
+    TASK=""
+    MAX_SEQ_LEN=""
+    EXTRA_PIP=""
+    PREPROCESSOR_FEATURE_SIZE=""
+    PREPROCESSOR_OUTPUT=""
+    ;;
   mistralai/Voxtral-Mini-4B-Realtime-2602)
     MODEL_NAME="voxtral_realtime"
     TASK=""
@@ -167,7 +177,7 @@ case "$HF_MODEL" in
     ;;
   *)
     echo "Error: Unsupported model '$HF_MODEL'"
-    echo "Supported models: mistralai/Voxtral-Mini-3B-2507, mistralai/Voxtral-Mini-4B-Realtime-2602, openai/whisper-{small, medium, large, large-v2, large-v3, large-v3-turbo}, google/gemma-3-4b-it, Qwen/Qwen3-0.6B, nvidia/parakeet-tdt"
+    echo "Supported models: mistralai/Voxtral-Mini-3B-2507, mistralai/Voxtral-Mini-4B-Realtime-2602, openai/whisper-{small, medium, large, large-v2, large-v3, large-v3-turbo}, google/gemma-3-4b-it, Qwen/Qwen3-0.6B, nvidia/diar_streaming_sortformer_4spk-v2, nvidia/parakeet-tdt"
     exit 1
     ;;
 esac
@@ -247,6 +257,42 @@ if [ "$MODEL_NAME" = "parakeet" ]; then
   exit 0
 fi
 
+# Sortformer uses a custom export script
+if [ "$MODEL_NAME" = "sortformer" ]; then
+  if [ "$QUANT_NAME" != "non-quantized" ]; then
+    echo "Error: Sortformer currently supports only non-quantized export"
+    exit 1
+  fi
+
+  pip install -r examples/models/sortformer/install_requirements.txt
+
+  SORTFORMER_BACKEND="$DEVICE"
+  if [ "$DEVICE" = "cuda-windows" ]; then
+    SORTFORMER_BACKEND="cuda-windows"
+  elif [ "$DEVICE" = "cuda" ]; then
+    SORTFORMER_BACKEND="cuda"
+  elif [ "$DEVICE" = "xnnpack" ]; then
+    SORTFORMER_BACKEND="xnnpack"
+  else
+    SORTFORMER_BACKEND="portable"
+  fi
+
+  python -m executorch.examples.models.sortformer.export_sortformer \
+      --hf-model "${HF_MODEL}" \
+      --backend "${SORTFORMER_BACKEND}" \
+      --output-dir "${OUTPUT_DIR}"
+
+  test -f "${OUTPUT_DIR}/sortformer.pte"
+  mv "${OUTPUT_DIR}/sortformer.pte" "${OUTPUT_DIR}/model.pte"
+  # CUDA saves named data to separate .ptd file, XNNPACK/portable do not.
+  if [ "$DEVICE" = "cuda" ] || [ "$DEVICE" = "cuda-windows" ]; then
+    test -f "${OUTPUT_DIR}/aoti_cuda_blob.ptd"
+  fi
+  ls -al "${OUTPUT_DIR}"
+  echo "::endgroup::"
+  exit 0
+fi
+
 # Voxtral Realtime uses a custom export script
 if [ "$MODEL_NAME" = "voxtral_realtime" ]; then
   pip install safetensors huggingface_hub
@@ -296,6 +342,9 @@ if [ "$MODEL_NAME" = "voxtral_realtime" ]; then
 
   test -f "${OUTPUT_DIR}/model.pte"
   test -f "${OUTPUT_DIR}/preprocessor.pte"
+  if [ "$DEVICE" = "cuda" ] || [ "$DEVICE" = "cuda-windows" ]; then
+    test -f "${OUTPUT_DIR}/aoti_cuda_blob.ptd"
+  fi
   # Copy tokenizer from downloaded model weights
   cp "$LOCAL_MODEL_DIR/tekken.json" "${OUTPUT_DIR}/tekken.json"
   ls -al "${OUTPUT_DIR}"

.ci/scripts/test_model_e2e.sh

@@ -19,6 +19,7 @@ Arguments:
   hf_model    HuggingFace model ID (required)
               Supported models:
                 - mistralai/Voxtral-Mini-3B-2507
+                - nvidia/diar_streaming_sortformer_4spk-v2
                 - openai/whisper series (whisper-{small, medium, large, large-v2, large-v3, large-v3-turbo})
                 - google/gemma-3-4b-it
                 - Qwen/Qwen3-0.6B
@@ -44,6 +45,7 @@ Arguments:
 Examples:
   test_model_e2e.sh metal "openai/whisper-small" "non-quantized"
   test_model_e2e.sh cuda "mistralai/Voxtral-Mini-3B-2507" "quantized-int4-tile-packed" "./model_output"
+  test_model_e2e.sh cuda "nvidia/diar_streaming_sortformer_4spk-v2" "non-quantized" "./model_output"
   test_model_e2e.sh cuda "nvidia/parakeet-tdt" "non-quantized" "./model_output"
   test_model_e2e.sh xnnpack "nvidia/parakeet-tdt" "quantized-8da4w" "./model_output"
   test_model_e2e.sh metal "mistralai/Voxtral-Mini-4B-Realtime-2602" "non-quantized" "." "vr-streaming"
@@ -176,6 +178,18 @@ case "$HF_MODEL" in
     AUDIO_FILE="test_audio.wav"
     IMAGE_PATH=""
     ;;
+  nvidia/diar_streaming_sortformer_4spk-v2)
+    MODEL_NAME="sortformer"
+    RUNNER_TARGET="sortformer_runner"
+    RUNNER_PATH="sortformer"
+    EXPECTED_OUTPUT="Speaker 1"
+    PREPROCESSOR=""
+    TOKENIZER_URL=""
+    TOKENIZER_FILE=""
+    AUDIO_URL="https://github.com/voxserv/audio_quality_testing_samples/raw/refs/heads/master/testaudio/16000/test01_20s.wav"
+    AUDIO_FILE="poem.wav"
+    IMAGE_PATH=""
+    ;;
   mistralai/Voxtral-Mini-4B-Realtime-2602)
     MODEL_NAME="voxtral_realtime"
     RUNNER_TARGET="voxtral_realtime_runner"
@@ -190,7 +204,7 @@ case "$HF_MODEL" in
     ;;
   *)
     echo "Error: Unsupported model '$HF_MODEL'"
-    echo "Supported models: mistralai/Voxtral-Mini-3B-2507, mistralai/Voxtral-Mini-4B-Realtime-2602, openai/whisper series (whisper-{small, medium, large, large-v2, large-v3, large-v3-turbo}), google/gemma-3-4b-it, Qwen/Qwen3-0.6B, nvidia/parakeet-tdt"
+    echo "Supported models: mistralai/Voxtral-Mini-3B-2507, mistralai/Voxtral-Mini-4B-Realtime-2602, nvidia/diar_streaming_sortformer_4spk-v2, openai/whisper series (whisper-{small, medium, large, large-v2, large-v3, large-v3-turbo}), google/gemma-3-4b-it, Qwen/Qwen3-0.6B, nvidia/parakeet-tdt"
     exit 1
     ;;
 esac
@@ -203,8 +217,8 @@ echo "::endgroup::"
 echo "::group::Prepare $MODEL_NAME Artifacts"
 
 
-# Download tokenizer files (skip for parakeet and voxtral_realtime which bundle tokenizer in export)
-if [ "$MODEL_NAME" != "parakeet" ] && [ "$MODEL_NAME" != "voxtral_realtime" ]; then
+# Download tokenizer files (skip for models that bundle tokenizer in export or do not use one)
+if [ "$MODEL_NAME" != "parakeet" ] && [ "$MODEL_NAME" != "voxtral_realtime" ] && [ "$MODEL_NAME" != "sortformer" ]; then
   if [ "$TOKENIZER_FILE" != "" ]; then
     curl -L $TOKENIZER_URL/$TOKENIZER_FILE -o $MODEL_DIR/$TOKENIZER_FILE
   else
@@ -296,6 +310,12 @@ EOF
       RUNNER_ARGS="$RUNNER_ARGS --data_path ${MODEL_DIR}/aoti_cuda_blob.ptd"
     fi
     ;;
+  sortformer)
+    RUNNER_ARGS="--model_path ${MODEL_DIR}/model.pte --audio_path ${MODEL_DIR}/$AUDIO_FILE"
+    if [ "$DEVICE" = "cuda" ]; then
+      RUNNER_ARGS="$RUNNER_ARGS --data_path ${MODEL_DIR}/aoti_cuda_blob.ptd"
+    fi
+    ;;
   voxtral_realtime)
     RUNNER_ARGS="--model_path ${MODEL_DIR}/model.pte --tokenizer_path ${MODEL_DIR}/$TOKENIZER_FILE --preprocessor_path ${MODEL_DIR}/$PREPROCESSOR --audio_path ${MODEL_DIR}/$AUDIO_FILE --temperature 0"
     # Add CUDA data path if present

.ci/scripts/test_model_e2e_windows.ps1

@@ -64,8 +64,30 @@ switch ($HfModel) {
         $audioUrl = "https://dldata-public.s3.us-east-2.amazonaws.com/2086-149220-0033.wav"
         $audioFile = "test_audio.wav"
     }
+    "nvidia/diar_streaming_sortformer_4spk-v2" {
+        $runnerTarget = "sortformer_runner"
+        $runnerPath = "sortformer"
+        $runnerPreset = "sortformer-cuda"
+        $expectedOutput = "Speaker 1"
+        $preprocessor = ""
+        $tokenizerUrl = ""
+        $tokenizerFile = ""
+        $audioUrl = "https://github.com/voxserv/audio_quality_testing_samples/raw/refs/heads/master/testaudio/16000/test01_20s.wav"
+        $audioFile = "poem.wav"
+    }
+    "mistralai/Voxtral-Mini-4B-Realtime-2602" {
+        $runnerTarget = "voxtral_realtime_runner"
+        $runnerPath = "voxtral_realtime"
+        $runnerPreset = "voxtral-realtime-cuda"
+        $expectedOutput = "Loading audio from"
+        $preprocessor = "preprocessor.pte"
+        $tokenizerUrl = ""
+        $tokenizerFile = "tekken.json"
+        $audioUrl = "https://github.com/voxserv/audio_quality_testing_samples/raw/refs/heads/master/testaudio/16000/test01_20s.wav"
+        $audioFile = "poem.wav"
+    }
     default {
-        throw "Unsupported model '$HfModel'. Supported: mistralai/Voxtral-Mini-3B-2507, nvidia/parakeet-tdt"
+        throw "Unsupported model '$HfModel'. Supported: mistralai/Voxtral-Mini-3B-2507, mistralai/Voxtral-Mini-4B-Realtime-2602, nvidia/diar_streaming_sortformer_4spk-v2, nvidia/parakeet-tdt"
     }
 }
 
@@ -171,6 +193,21 @@ try {
                 "--data_path", $cudaBlob
             )
         }
+        "nvidia/diar_streaming_sortformer_4spk-v2" {
+            $runnerArgs = @(
+                "--model_path", $modelPte,
+                "--audio_path", (Join-Path -Path $resolvedModelDir -ChildPath $audioFile),
+                "--data_path", $cudaBlob
+            )
+        }
+        "mistralai/Voxtral-Mini-4B-Realtime-2602" {
+            $runnerArgs += @(
+                "--temperature", "0",
+                "--tokenizer_path", (Join-Path -Path $resolvedModelDir -ChildPath $tokenizerFile),
+                "--audio_path", (Join-Path -Path $resolvedModelDir -ChildPath $audioFile),
+                "--preprocessor_path", (Join-Path -Path $resolvedModelDir -ChildPath $preprocessor)
+            )
+        }
     }
 
     $stdoutFile = Join-Path -Path $env:TEMP -ChildPath ("et_runner_stdout_{0}.log" -f ([Guid]::NewGuid().ToString("N")))
@@ -202,10 +239,15 @@ try {
         Write-Warning "Runner exited with code $exitCode (may be benign)"
     }
 
-    if ($expectedOutput -ne "" -and $stdout -notmatch [Regex]::Escape($expectedOutput)) {
-        throw "Expected output '$expectedOutput' not found in runner output"
+    if ($expectedOutput -ne "") {
+        if ($stdout -notmatch [Regex]::Escape($expectedOutput)) {
+            throw "Expected output '$expectedOutput' not found in runner output"
+        }
+        Write-Host "Success: '$expectedOutput' found in output"
+    }
+    else {
+        Write-Host "Success: runner completed"
     }
-    Write-Host "Success: '$expectedOutput' found in output"
     Write-Host "::endgroup::"
 }
 finally {

.ci/scripts/unittest-linux-cmake.sh

@@ -14,14 +14,11 @@ if ! python -c "import tosa_serializer" >/dev/null 2>&1; then
   TOSA_SERIALIZATION_DIR="./examples/arm/arm-scratch/tosa-tools/serialization"
   if [[ ! -d "${TOSA_SERIALIZATION_DIR}" ]]; then
     TOSA_TOOLS_DIR="$(mktemp -d /tmp/tosa-tools.XXXXXX)"
-    git clone --depth 1 --branch v2025.11.0 \
+    git clone --depth 1 --branch v2025.11.2 \
       https://git.gitlab.arm.com/tosa/tosa-tools.git "${TOSA_TOOLS_DIR}"
     TOSA_SERIALIZATION_DIR="${TOSA_TOOLS_DIR}/serialization"
   fi
 
-  # Workaround to allow TOSA serializer to build for v2025.11.0
-  python -m pip install pybind11==2.10.4
-
   CMAKE_POLICY_VERSION_MINIMUM=3.5 BUILD_PYBIND=1 \
     python -m pip install --no-dependencies \
     "${TOSA_SERIALIZATION_DIR}"

.ci/scripts/zephyr-utils.sh

@@ -6,14 +6,18 @@
 # This source code is licensed under the BSD-style license found in the
 # LICENSE file in the root directory of this source tree.
 
-# Run instruction from zephyr/README.md
+# Run instruction from a README.md file
 #
-# This contains some CI helper runtime functions to dig out and run commands
-# from zephyr/README.md
-# It also try to verify that zephyr/executorch.yaml is in sync and the snippets
-# in the README are sane with various regexps.
+# This contains CI helper runtime functions to dig out and run commands
+# from a README file.
 #
-# Main functions is run_command_block_from_zephyr_readme it will dig out the code between
+# Generic helper:
+# - run_command_block_from_readme
+#
+# Zephyr-specific validator:
+# - verify_zephyr_readme (checks zephyr/executorch.yaml snippet sync)
+#
+# Main function is run_command_block_from_readme. It digs out the code between
 # the two ''' after a blockheader text and run it
 #
 # .e.g. from this README.md snippet
@@ -26,7 +30,7 @@
 #
 # The blockheader is 'Install requirements' and the code block is 'pip install something ... \npip install something_else'
 # so if we run
-#   run_command_block_from_zephyr_readme 'Install requirements'
+#   run_command_block_from_readme path/to/README.md 'Install requirements'
 # it will run them one by one
 
 
@@ -56,10 +60,22 @@ _zephyr_utils_root_dir () {
     (cd "${script_dir}/../.." && pwd)
 }
 
-_zephyr_utils_readme_path () {
+_utils_path_from_root () {
+  local path="$1"
   local root_dir
+
+  if [[ -z "${path}" || "${path}" == "." ]]; then
+    echo "ERROR: Path argument must be a non-empty file path" >&2
+    return 1
+  fi
+
+  if [[ "${path}" = /* ]]; then
+    printf '%s\n' "${path}"
+    return 0
+  fi
+
   root_dir="$(_zephyr_utils_root_dir)"
-  printf '%s/zephyr/README.md\n' "${root_dir}"
+  printf '%s/%s\n' "${root_dir}" "${path}"
 }
 
 _zephyr_utils_ensure_file () {
@@ -117,28 +133,37 @@ _zephyr_utils_run_simple_commands () {
   return 0
 }
 
-run_command_block_from_zephyr_readme () {
-  local blockheader="$1"
+run_command_block_from_readme () {
+  local readme_path="$1"
+  local blockheader="$2"
+  local block
+
+  if [[ -z "${readme_path}" || -z "${blockheader}" ]]; then
+    echo "ERROR: Usage: run_command_block_from_readme <readme_path> <block_header_regex>" >&2
+    return 1
+  fi
+
+  readme_path="$(_utils_path_from_root "${readme_path}")" || return 1
 
-  echo "Run block '${blockheader}' from zephyr/README.md"
+  echo "Run block '${blockheader}' from ${readme_path}"
 
-  readme_path="$(_zephyr_utils_readme_path)"
   _zephyr_utils_ensure_file "${readme_path}" "README.md" || return 1
 
   block="$(_zephyr_utils_extract_block "${readme_path}" "${blockheader}")"
-
-  if [[ -z "${block}" ]]; then
-    echo "ERROR: Failed to locate ${blockheader} block in ${readme_path}" >&2
-    return 1
+  if [[ -n "${block}" ]]; then
+    _zephyr_utils_run_simple_commands "${block}" "${blockheader}"
+    return $?
   fi
-  _zephyr_utils_run_simple_commands "${block}" "${blockheader}"
+
+  echo "ERROR: Failed to locate ${blockheader} block in ${readme_path}" >&2
+  return 1
 }
 
 # Check that zephyr/executorch.yaml match zephyr/README.md
 verify_zephyr_readme () {
   local readme_path manifest_path snippet
 
-  readme_path="$(_zephyr_utils_readme_path)"
+  readme_path="$(_utils_path_from_root "zephyr/README.md")" || return 1
   manifest_path="$(_zephyr_utils_root_dir)/zephyr/executorch.yaml"
 
   _zephyr_utils_ensure_file "${readme_path}" "README" || return 1