Skip to content

Add model response output to GitHub job summary and support multiple model presets #77

Add model response output to GitHub job summary and support multiple model presets

Add model response output to GitHub job summary and support multiple model presets #77

Workflow file for this run

# 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.
name: LlamaDemo Android
on:
pull_request:
branches: [main]
paths:
- 'llm/android/**'
- '.github/workflows/llm-android.yml'
workflow_dispatch:
inputs:
model_preset:
description: 'Model preset to use'
required: true
type: choice
options:
- stories
- llama
- qwen3
- custom
default: 'stories'
custom_pte_url:
description: 'Custom URL for model .pte file (only used when model_preset is custom)'
required: false
type: string
custom_tokenizer_url:
description: 'Custom URL for tokenizer file (only used when model_preset is custom)'
required: false
type: string
permissions:
contents: read
jobs:
instrumentation-test:
runs-on: 8-core-ubuntu
env:
API_LEVEL: 34
ARCH: x86_64
EMULATOR_OPTIONS: -no-window -gpu swiftshader_indirect -noaudio -no-boot-anim -camera-back none
RAM_SIZE: 16384
name: Instrumentation Test LlamaDemo
steps:
- name: Checkout repository
uses: actions/checkout@v4
- name: Write job summary
run: |
echo "## Test Configuration" >> $GITHUB_STEP_SUMMARY
echo "| Parameter | Value |" >> $GITHUB_STEP_SUMMARY
echo "|-----------|-------|" >> $GITHUB_STEP_SUMMARY
echo "| Model Preset | \`${{ inputs.model_preset || 'stories' }}\` |" >> $GITHUB_STEP_SUMMARY
if [ "${{ inputs.model_preset }}" = "custom" ]; then
echo "| Custom PTE URL | \`${{ inputs.custom_pte_url }}\` |" >> $GITHUB_STEP_SUMMARY
echo "| Custom Tokenizer URL | \`${{ inputs.custom_tokenizer_url }}\` |" >> $GITHUB_STEP_SUMMARY
fi
- name: Enable KVM group perms
run: |
echo 'KERNEL=="kvm", GROUP="kvm", MODE="0666", OPTIONS+="static_node=kvm"' | sudo tee /etc/udev/rules.d/99-kvm4all.rules
sudo udevadm control --reload-rules
sudo udevadm trigger --name-match=kvm
- name: Set up JDK 17
uses: actions/setup-java@v4
with:
java-version: '17'
distribution: 'temurin'
- name: Setup Gradle
uses: gradle/actions/setup-gradle@v4
- name: AVD cache
uses: actions/cache@v4
id: avd-cache
with:
path: |
~/.android/avd/*
~/.android/adb*
key: avd-${{ env.API_LEVEL }}-${{ env.ARCH }}-ram${{ env.RAM_SIZE }}
- name: Create AVD and generate snapshot for caching
if: steps.avd-cache.outputs.cache-hit != 'true'
uses: reactivecircus/android-emulator-runner@v2
with:
api-level: ${{ env.API_LEVEL }}
arch: ${{ env.ARCH }}
force-avd-creation: true
emulator-options: ${{ env.EMULATOR_OPTIONS }}
disable-animations: false
working-directory: llm/android/LlamaDemo
script: echo "Generated AVD snapshot for caching."
- name: Configure AVD RAM
run: |
AVD_DIR="$HOME/.android/avd"
for config in "$AVD_DIR"/*.avd/config.ini; do
if [ -f "$config" ]; then
echo "Updating RAM in $config"
sed -i 's/hw.ramSize=.*/hw.ramSize=${{ env.RAM_SIZE }}/' "$config" || true
grep -q "hw.ramSize" "$config" || echo "hw.ramSize=${{ env.RAM_SIZE }}" >> "$config"
fi
done
- name: Run instrumentation tests
uses: reactivecircus/android-emulator-runner@v2
env:
MODEL_PRESET: ${{ inputs.model_preset || 'stories' }}
CUSTOM_PTE_URL: ${{ inputs.custom_pte_url }}
CUSTOM_TOKENIZER_URL: ${{ inputs.custom_tokenizer_url }}
with:
api-level: ${{ env.API_LEVEL }}
arch: ${{ env.ARCH }}
force-avd-creation: false
emulator-options: -no-snapshot-save ${{ env.EMULATOR_OPTIONS }}
disable-animations: true
working-directory: llm/android/LlamaDemo
script: |
adb shell rm -rf /data/local/tmp/llama
adb shell mkdir -p /data/local/tmp/llama
echo "=== Model directory after cleanup ==="
adb shell ls -la /data/local/tmp/llama/ || echo "Directory empty or not found"
adb logcat -c && adb logcat > /tmp/logcat.txt &
LOGCAT_PID=$!
if [ "$MODEL_PRESET" = "custom" ]; then GRADLE_ARGS="-PmodelPreset=$MODEL_PRESET -PcustomPteUrl=$CUSTOM_PTE_URL -PcustomTokenizerUrl=$CUSTOM_TOKENIZER_URL"; else GRADLE_ARGS="-PmodelPreset=$MODEL_PRESET"; fi
echo "=== Running Gradle with preset: $MODEL_PRESET ==="
./gradlew connectedCheck $GRADLE_ARGS; TEST_EXIT_CODE=$?
echo "=== Model directory after Gradle ==="
adb shell ls -la /data/local/tmp/llama/
kill $LOGCAT_PID || true
# Show which model was used
echo "=== Model configuration used by test ==="
grep "UIWorkflowTest.*Using model" /tmp/logcat.txt || echo "Model config not found in logcat"
# Extract response from logcat
echo "=== Searching for LLAMA_RESPONSE in logcat ==="
grep "LLAMA_RESPONSE" /tmp/logcat.txt || echo "No LLAMA_RESPONSE found in logcat"
grep "LLAMA_RESPONSE" /tmp/logcat.txt | sed 's/.*LLAMA_RESPONSE: //' | grep -v "BEGIN_RESPONSE\|END_RESPONSE" > /tmp/response.txt || true
echo "=== Response file contents ==="
cat /tmp/response.txt || echo "Response file empty or not created"
exit $TEST_EXIT_CODE
- name: Add model response to summary
if: always()
run: |
if [ -f /tmp/response.txt ]; then
echo "" >> $GITHUB_STEP_SUMMARY
echo "## Model Response" >> $GITHUB_STEP_SUMMARY
echo '```' >> $GITHUB_STEP_SUMMARY
cat /tmp/response.txt >> $GITHUB_STEP_SUMMARY
echo '```' >> $GITHUB_STEP_SUMMARY
fi
- name: Upload logcat
if: always()
uses: actions/upload-artifact@v4
with:
name: logcat
path: /tmp/logcat.txt
retention-days: 7