Update Colab demo notebook and test scripts

test_scripts/litert_cli.ipynb

@@ -0,0 +1,238 @@
+{
+ "cells": [
+  {
+   "id": "90c3deda",
+   "cell_type": "markdown",
+   "source": [
+    "# LiteRT CLI Demo\n",
+    "\n",
+    "This notebook demonstrates how to use the `litert-cli` tool to convert a PyTorch model, quantize it and run it."
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "19270446",
+   "cell_type": "markdown",
+   "source": [
+    "## \ud83d\udee0\ufe0f 1. Environment Setup & Installation"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "d34739df",
+   "cell_type": "code",
+   "source": [
+    "# Install required packages\n",
+    "!pip install torch torchvision\n",
+    "\n",
+    "# Install litert-cli\n",
+    "!pip install -q -i https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple litert-cli==0.1.1.dev23\n",
+    "\n",
+    "# 'litert compile' depends on Clang, and below make sure your Clang has version `18.x.x` or above\n",
+    "!wget https://apt.llvm.org/llvm.sh\n",
+    "!chmod +x llvm.sh\n",
+    "!sudo ./llvm.sh 18 all"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "1dd1eb34",
+   "cell_type": "markdown",
+   "source": [
+    "## \ud83d\udcdd 2. Prepare PyTorch Model Script"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "5e40f66c",
+   "cell_type": "code",
+   "source": [
+    "%%writefile resnet18.py\n",
+    "import torch\n",
+    "import torchvision\n",
+    "\n",
+    "def get_model() -> torch.nn.Module:\n",
+    "  model = torchvision.models.resnet18(\n",
+    "      weights=torchvision.models.ResNet18_Weights.IMAGENET1K_V1\n",
+    "  )\n",
+    "  model.eval()\n",
+    "  return model\n",
+    "\n",
+    "def get_args() -> tuple[torch.Tensor, ...]:\n",
+    "  return (torch.randn(1, 3, 224, 224),)"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "427dca8a",
+   "cell_type": "markdown",
+   "source": [
+    "## \ud83d\udd04 3. Model Conversion (PyTorch -> LiteRT)"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "578cc653",
+   "cell_type": "code",
+   "source": [
+    "# Convert PyTorch ResNet18 model to LiteRT\n",
+    "!litert convert resnet18.py --output resnet18"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "606efd4f",
+   "cell_type": "markdown",
+   "source": [
+    "## \ud83d\udcc9 4. Model Quantization"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "c589c711",
+   "cell_type": "code",
+   "source": [
+    "# Quantize the ResNet18 model\n",
+    "!litert quantize resnet18/resnet18.tflite --type int8_dynamic --output resnet18/resnet18_int8_dynamic.tflite\n",
+    "!litert quantize resnet18/resnet18.tflite --type int8_weight_only --output resnet18/resnet18_int8_weight_only.tflite"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "b3ae9ddc",
+   "cell_type": "markdown",
+   "source": [
+    "## \ud83d\ude80 5. Run Inference\n",
+    "### \ud83d\udda5\ufe0f 5.1 CPU Inference"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "ea93a1db",
+   "cell_type": "code",
+   "source": [
+    "# Run Inference on Desktop (Colab VM CPU)\n",
+    "!litert run resnet18/resnet18.tflite --desktop --cpu --iterations 1\n",
+    "!litert run resnet18/resnet18_int8_dynamic.tflite --desktop --cpu --iterations 1"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "9d748475",
+   "cell_type": "markdown",
+   "source": [
+    "### \ud83c\udfae 5.2 GPU Inference"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "00b026b2",
+   "cell_type": "code",
+   "source": [
+    "# Run Inference on Desktop (Colab VM GPU)\n",
+    "!litert run resnet18/resnet18.tflite --desktop --gpu --iterations 1\n",
+    "!litert run resnet18/resnet18_int8_dynamic.tflite --desktop --gpu --iterations 1"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "eaa972bc",
+   "cell_type": "markdown",
+   "source": [
+    "## \u2699\ufe0f 6. NPU Offline Compilation (AOT)"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "d0a61181",
+   "cell_type": "code",
+   "source": [
+    "# Compile the model for qualcomm NPU: SM8750\n",
+    "# TIP: Only support running on Linux and it might take a few minutes, given download large SDKs from SOCs.\n",
+    "# TIP: It depends on Clang, and please make sure your Clang has version `18.x.x` or above when running\n",
+    "#   `clang --version`. To update, choose one of below: \n",
+    "#   a) `sudo apt update && sudo apt upgrade -y`\n",
+    "#   b) wget https://apt.llvm.org/llvm.sh\n",
+    "#      chmod +x llvm.sh\n",
+    "#      sudo ./llvm.sh 18 all\n",
+    "#\n",
+    "!litert compile resnet18/resnet18.tflite --target sm8750"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "2da953c9",
+   "cell_type": "markdown",
+   "source": [
+    "## \ud83c\udfc1 7. Benchmark in Google AI Edge Portal"
+   ],
+   "metadata": {}
+  },
+  {
+   "id": "a9a19e48",
+   "cell_type": "code",
+   "source": [
+    "# Please login into Google Cloud and make sure you have joined the EAP for Google AI Edge Portal\n",
+    "# Check: https://ai.google.dev/edge/ai-edge-portal\n",
+    "!pip install gcloud\n",
+    "!gcloud auth login\n",
+    "\n",
+    "# Benchmark on Google AI Edge Portal\n",
+    "# Please specify you own GCP project: --gcp-project <your-own-gcp-project-id>\n",
+    "# Or set a default environment variable: LITERT_GCP_PROJECT\n",
+    "!litert benchmark model.tflite --gcp --device \"pixel 7\" --cpu --gcp-project aep-e2e-test\n",
+    "!litert benchmark model.tflite --gcp --device \"pixel 7\" --gpu --gcp-project aep-e2e-test\n",
+    "!litert benchmark model.tflite --gcp --devices \"pixel 7, sm-s931u1\" --gpu"
+   ],
+   "metadata": {},
+   "execution_count": null,
+   "outputs": []
+  },
+  {
+   "id": "2685b076",
+   "cell_type": "markdown",
+   "source": [
+    "## Advanced Usage\n",
+    "\n",
+    "### Android Deployment\n",
+    "To run or benchmark on Android, you need an ADB connected device. These commands typically look like:\n",
+    "```bash\n",
+    "# Run on Android\n",
+    "!litert run resnet18/resnet18.tflite --android --cpu\n",
+    "\n",
+    "# Benchmark on Android\n",
+    "!litert benchmark resnet18/resnet18.tflite --android\n",
+    "```\n",
+    "These are not executed in this notebook as Colab does not have access to a physical Android device by default."
+   ],
+   "metadata": {}
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": ".venv",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "name": "python",
+   "version": "3.13.12"
+  }
+ },
+ "nbformat_minor": 5,
+ "nbformat": 4
+}

test_scripts/models/efficientnet.sh

@@ -115,9 +115,8 @@ fi
 
 
 # --- 5. Compile (AOT Compilation) ---
-# TODO: Add this back when we fix the NPU compile issue.
-# run_case "Compile: EfficientNet FP32 for Qualcomm sm8750 NPU" \
-#    litert compile "$EFFICIENTNET_TFLITE" --target sm8750 --output-dir "$MODEL_DIR/efficientnet"
+run_case "Compile: EfficientNet FP32 for Qualcomm sm8750 NPU" \
+    litert compile "$EFFICIENTNET_TFLITE" --target sm8750 --output-dir "$MODEL_DIR/efficientnet"
 
 # --- 6. Visualize (Model Explorer) ---
 run_case "Visualize: Launch Model Explorer in the background" \

test_scripts/models/resnet.sh

@@ -117,9 +117,8 @@ fi
 
 
 # --- 5. Compile (AOT Compilation) ---
-# TODO: Add this back when we fix the NPU compile issue.
-# run_case "Compile: ResNet18 FP32 for Qualcomm sm8750 NPU" \
-#     litert compile "$RESNET_TFLITE" --target sm8750 --output-dir "$MODEL_DIR/resnet18"
+run_case "Compile: ResNet18 FP32 for Qualcomm sm8750 NPU" \
+     litert compile "$RESNET_TFLITE" --target sm8750 --output-dir "$MODEL_DIR/resnet18"
 
 # --- 6. Visualize (Model Explorer) ---
 run_case "Visualize: Launch Model Explorer in the background" \