Troubleshooting Guide - React Native ExecuTorch

This guide provides solutions to common issues you may encounter when using React Native ExecuTorch.

Installation Issues

Resource Fetcher Not Initialized

Error Message: ResourceFetcherAdapterNotInitialized (Code 186) Cause: The resource fetcher was not initialized before using model hooks. Solution: For Expo projects:

import { initExecutorch } from 'react-native-executorch';
import { ExpoResourceFetcher } from '@react-native-executorch/expo-resource-fetcher';

// Call this before using any hooks, typically in App.tsx or _layout.tsx
initExecutorch({
  resourceFetcher: ExpoResourceFetcher,
});

For bare React Native projects:

import { initExecutorch } from 'react-native-executorch';
import { BareResourceFetcher } from '@react-native-executorch/bare-resource-fetcher';

initExecutorch({
  resourceFetcher: BareResourceFetcher,
});

Native Module Not Found

Error Message: Module not found or NativeModule.RNExecutorch is null Solution for iOS:

cd ios
rm -rf Pods Podfile.lock
pod install
cd ..
npx react-native run-ios

Solution for Android:

cd android
./gradlew clean
cd ..
npx react-native run-android

If issues persist, reset Metro cache:

npx react-native start --reset-cache

New Architecture Not Enabled

Error Message: Architecture-related errors or feature not available Solution for Expo: Update app.json:

{
  "expo": {
    "plugins": [
      [
        "expo-build-properties",
        {
          "ios": { "newArchEnabled": true },
          "android": { "newArchEnabled": true }
        }
      ]
    ]
  }
}

Then rebuild:

npx expo prebuild --clean
npx expo run:ios  # or run:android

Solution for Bare React Native: iOS - Edit ios/Podfile:

:fabric_enabled => true

Android - Edit android/gradle.properties:

newArchEnabled=true

Then clean and rebuild:

cd ios && rm -rf Pods Podfile.lock && pod install && cd ..
cd android && ./gradlew clean && cd ..

Model Loading Issues

Model Not Loading (isReady stays false)

Symptoms: isReady stays false, downloadProgress not increasing Debugging Steps:

const llm = useLLM({ model: LLAMA3_2_1B });

useEffect(() => {
  console.log('Download progress:', llm.downloadProgress);
  console.log('Is ready:', llm.isReady);
  console.log('Error:', llm.error);
}, [llm.downloadProgress, llm.isReady, llm.error]);

Common Causes:

Network Issues: Check internet connectivity
Invalid URL: Verify model URL is accessible
Insufficient Storage: Check available device storage
Resource Fetcher Not Initialized: See above

Solution:

import { ExpoResourceFetcher } from '@react-native-executorch/expo-resource-fetcher';

// Check if model is already downloaded
const models = await ExpoResourceFetcher.listDownloadedModels();
console.log('Downloaded models:', models);

// Try manual download with error handling
try {
  await ExpoResourceFetcher.fetch(
    (progress) => console.log(`Progress: ${progress * 100}%`),
    'https://model-url.pte'
  );
} catch (error) {
  console.error('Download failed:', error);
}

Download Fails or Interrupts

Error Message: ResourceFetcherDownloadFailed (Code 180) or DownloadInterrupted (Code 118) Solution: Implement retry logic

const downloadWithRetry = async (
  url: string,
  maxRetries = 3
): Promise<string[] | null> => {
  for (let i = 0; i < maxRetries; i++) {
    try {
      const paths = await ExpoResourceFetcher.fetch(
        (progress) => console.log(`Attempt ${i + 1}: ${progress * 100}%`),
        url
      );
      return paths;
    } catch (error) {
      console.log(`Attempt ${i + 1} failed`);
      if (i === maxRetries - 1) throw error;
      
      // Wait before retrying (exponential backoff)
      await new Promise(resolve => 
        setTimeout(resolve, 1000 * Math.pow(2, i))
      );
    }
  }
  return null;
};

Invalid Model File

Error Message: InvalidProgram (Code 35) Causes:

Corrupted download
Wrong file format
Model not properly exported

Solution:

Delete and re-download:

await ExpoResourceFetcher.deleteResources('https://model-url.pte');
await ExpoResourceFetcher.fetch(() => {}, 'https://model-url.pte');

Verify file is valid .pte format
Re-export model using correct ExecuTorch export process

Memory Issues

Out of Memory / App Crashes

Error Message: MemoryAllocationFailed (Code 33) or app crashes without error Immediate Solution: Use quantized models

import { 
  LLAMA3_2_1B_SPINQUANT,  // Instead of LLAMA3_2_1B
} from 'react-native-executorch';

const llm = useLLM({ model: LLAMA3_2_1B_SPINQUANT });

Memory Requirements:

Model	iOS Memory	Android Memory
LLAMA3_2_1B	3.1 GB	3.3 GB
LLAMA3_2_1B_SPINQUANT	2.4 GB	1.9 GB
LLAMA3_2_3B	7.3 GB	7.1 GB
LLAMA3_2_3B_SPINQUANT	3.8 GB	3.7 GB

Long-term Solutions:

Increase Android Emulator RAM: Set to 4GB+ in AVD Manager
Enable large heap (Android):

<!-- android/app/src/main/AndroidManifest.xml -->
<application
  android:largeHeap="true"
  ...>
</application>

Use context strategies:

import { SlidingWindowContextStrategy } from 'react-native-executorch';

llm.configure({
  chatConfig: {
    contextStrategy: new SlidingWindowContextStrategy({
      maxTokens: 2048,  // Limit context size
    }),
  },
});

Unload models when not needed:

import { LLMModule } from 'react-native-executorch';

const llm = new LLMModule();
await llm.load(/* ... */);

// Use model
await llm.generate(messages);

// Free memory when done
llm.delete();

See Memory Management for comprehensive strategies.

Memory Warnings (iOS)

Solution: Handle memory warnings

import { AppState } from 'react-native';
import { useEffect, useRef } from 'react';
import { LLMModule } from 'react-native-executorch';

function App() {
  const llmRef = useRef<LLMModule | null>(null);

  useEffect(() => {
    const subscription = AppState.addEventListener('memoryWarning', () => {
      console.warn('Memory warning - unloading model');
      if (llmRef.current) {
        llmRef.current.delete();
        llmRef.current = null;
      }
    });

    return () => subscription.remove();
  }, []);

  return /* Your app */;
}

Generation Issues

Model Not Generating (hangs)

Symptoms: generate() never completes, isGenerating stays true Solution: Implement timeout and interrupt

const generateWithTimeout = async (
  llm: LLMType,
  messages: Message[],
  timeoutMs = 30000
) => {
  const timeout = setTimeout(() => {
    console.warn('Generation timeout - interrupting');
    llm.interrupt();
  }, timeoutMs);

  try {
    await llm.generate(messages);
    clearTimeout(timeout);
    return llm.response;
  } catch (error) {
    clearTimeout(timeout);
    throw error;
  }
};

Empty or Unexpected Responses

Causes:

Invalid input format
Model not configured correctly
Temperature/sampling issues

Solution: Validate inputs and configuration

const llm = useLLM({ model: LLAMA3_2_1B });

// Configure properly
llm.configure({
  generationConfig: {
    temperature: 0.7,  // Reasonable value
    topP: 0.9,
    maxTokens: 512,
  },
});

// Ensure valid message format
const messages: Message[] = [
  { role: 'system', content: 'You are a helpful assistant' },
  { role: 'user', content: 'Hello!' },
];

if (messages.length === 0) {
  console.error('Empty message array');
  return;
}

try {
  await llm.generate(messages);
  
  if (llm.response.length === 0) {
    console.warn('Empty response generated');
  }
} catch (error) {
  console.error('Generation error:', error);
}

ModelGenerating Error

Error Message: ModelGenerating (Code 104) Cause: Trying to run inference while model is already generating Solution: Check isGenerating before calling

const handleGenerate = async () => {
  if (llm.isGenerating) {
    console.warn('Model is already generating');
    return;
  }

  await llm.generate(messages);
};

Input/Output Issues

FileReadFailed Error

Error Message: FileReadFailed (Code 114) Causes:

Invalid file path
File doesn’t exist
Unsupported image format
Permissions issue

Solution for Images:

import { useClassification } from 'react-native-executorch';

const classifier = useClassification({ model: EFFICIENTNET_V2_S });

// Validate image URI
const classifyImage = async (imageUri: string) => {
  if (!imageUri) {
    console.error('No image URI provided');
    return;
  }

  if (!imageUri.startsWith('file://')) {
    console.error('Invalid image URI:', imageUri);
    return;
  }

  try {
    const result = await classifier.classify({ image: imageUri });
    return result;
  } catch (error) {
    console.error('Classification failed:', error);
  }
};

WrongDimensions Error

Error Message: WrongDimensions (Code 116) Cause: Input tensor shape doesn’t match model’s expected shape Solution: Verify input dimensions

import { Tensor } from 'react-native-executorch';

// Check model requirements
// Example: ResNet expects [1, 3, 224, 224]

const input = Tensor.from({
  data: new Float32Array(1 * 3 * 224 * 224),
  shape: [1, 3, 224, 224],  // Match model's expected shape
  dtype: 'float32',
});

const output = await module.forward([input]);

TokenizerError

Error Message: TokenizerError (Code 167) Causes:

Tokenizer file missing or corrupted
Tokenizer config missing
Incompatible tokenizer

Solution:

import { LLAMA3_2_1B } from 'react-native-executorch';

// Ensure all three sources are provided
const llm = useLLM({
  model: {
    modelSource: LLAMA3_2_1B.modelSource,
    tokenizerSource: LLAMA3_2_1B.tokenizerSource,
    tokenizerConfigSource: LLAMA3_2_1B.tokenizerConfigSource,
  },
});

// Check for errors
useEffect(() => {
  if (llm.error?.code === RnExecutorchErrorCode.TokenizerError) {
    console.error('Tokenizer failed to load');
    // Re-download tokenizer files
  }
}, [llm.error]);

Platform-Specific Issues

iOS Simulator Issues

Problem: Slow performance or crashes on simulator Solution: Test on real devices. Simulators don’t reflect actual performance. For basic testing on simulator:

import { Platform } from 'react-native';

if (Platform.OS === 'ios' && __DEV__) {
  // Use smaller model for simulator testing
  const model = LLAMA3_2_1B_SPINQUANT; // Smaller footprint
}

Android Emulator Performance

Problem: Slow inference or OOM on emulator Solution: Increase emulator resources

Open Android Studio
Tools → AVD Manager
Edit your virtual device
Show Advanced Settings
Set RAM to 4096 MB or higher
Set VM heap to 512 MB
Enable hardware acceleration

iOS Build Errors

Problem: CocoaPods errors or build failures Solution:

cd ios

# Clean everything
rm -rf Pods Podfile.lock
rm -rf ~/Library/Developer/Xcode/DerivedData

# Reinstall
pod deintegrate
pod install

cd ..
npx react-native run-ios

Android Build Errors

Problem: Gradle errors or dependency conflicts Solution:

cd android

# Clean build
./gradlew clean
rm -rf .gradle

# Clear cache
rm -rf ~/.gradle/caches

cd ..
npx react-native run-android

Performance Issues

Slow Inference Speed

Solutions:

Use quantized models:

const llm = useLLM({ model: LLAMA3_2_1B_SPINQUANT });

Reduce generation length:

llm.configure({
  generationConfig: {
    maxTokens: 256,  // Shorter responses = faster
  },
});

Limit context:

import { SlidingWindowContextStrategy } from 'react-native-executorch';

llm.configure({
  chatConfig: {
    contextStrategy: new SlidingWindowContextStrategy({
      maxTokens: 1024,  // Smaller context
    }),
  },
});

Use appropriate backends: Ensure model was exported with XNNPACK or Core ML

See Performance Optimization for more strategies.

Slow Downloads

Solution: Pre-download models and cache them

import { ExpoResourceFetcher } from '@react-native-executorch/expo-resource-fetcher';

// Download during app idle time
const preloadModels = async () => {
  try {
    await ExpoResourceFetcher.fetch(
      (progress) => console.log(`Preload: ${progress * 100}%`),
      LLAMA3_2_1B.modelSource,
      LLAMA3_2_1B.tokenizerSource,
      LLAMA3_2_1B.tokenizerConfigSource
    );
    console.log('Models preloaded');
  } catch (error) {
    console.error('Preload failed:', error);
  }
};

Debugging Best Practices

Enable Comprehensive Logging

import { useLLM } from 'react-native-executorch';
import { useEffect } from 'react';

function DebugComponent() {
  const llm = useLLM({ model: LLAMA3_2_1B });

  useEffect(() => {
    console.log('=== LLM State ===');
    console.log('isReady:', llm.isReady);
    console.log('isGenerating:', llm.isGenerating);
    console.log('downloadProgress:', llm.downloadProgress);
    console.log('error:', llm.error);
    console.log('response length:', llm.response?.length);
    console.log('================');
  }, [
    llm.isReady,
    llm.isGenerating,
    llm.downloadProgress,
    llm.error,
    llm.response,
  ]);

  return /* Your UI */;
}

Check Model Files

import { ExpoResourceFetcher } from '@react-native-executorch/expo-resource-fetcher';

const debugStorage = async () => {
  const files = await ExpoResourceFetcher.listDownloadedFiles();
  const models = await ExpoResourceFetcher.listDownloadedModels();
  
  console.log('All files:', files);
  console.log('Model files:', models);
  
  for (const model of models) {
    console.log('Model path:', model);
  }
};

Profile Performance

const profileGeneration = async () => {
  const startTime = Date.now();
  
  await llm.generate(messages);
  
  const endTime = Date.now();
  const duration = (endTime - startTime) / 1000;
  
  console.log('=== Performance ===');
  console.log('Duration:', duration, 'seconds');
  console.log('Prompt tokens:', llm.getPromptTokenCount());
  console.log('Generated tokens:', llm.getGeneratedTokenCount());
  console.log('Total tokens:', llm.getTotalTokenCount());
  console.log('Speed:', (llm.getGeneratedTokenCount() / duration).toFixed(2), 'tokens/sec');
  console.log('==================');
};

Getting Help

If you’re still experiencing issues:

Check Documentation:
Search Existing Issues: GitHub Issues
Ask the Community: GitHub Discussions
Report a Bug: Include:
- React Native ExecuTorch version
- React Native version
- Platform (iOS/Android) and version
- Minimal reproduction code
- Error messages and logs
- Device specs (RAM, OS version)

Quick Checklist

When encountering issues, verify:

Resource fetcher is initialized
New Architecture is enabled
Model is downloaded (check downloadProgress)
Model is loaded (check isReady)
No errors (check error state)
Sufficient device memory for model
Input format matches model requirements
Testing on real device, not just emulator
Using latest version of React Native ExecuTorch

Next Steps

Review Debugging Guide for advanced techniques
Learn about Memory Management
Optimize with Performance Guide

Getting Started

Core Concepts

Large Language Models

Computer Vision

Speech & Audio

Text Embeddings

Advanced

Guides

​Installation Issues

​Resource Fetcher Not Initialized

​Native Module Not Found

​New Architecture Not Enabled

​Model Loading Issues

​Model Not Loading (isReady stays false)

​Download Fails or Interrupts

​Invalid Model File

​Memory Issues

​Out of Memory / App Crashes

​Memory Warnings (iOS)

​Generation Issues

​Model Not Generating (hangs)

​Empty or Unexpected Responses

​ModelGenerating Error

​Input/Output Issues

​FileReadFailed Error

​WrongDimensions Error

​TokenizerError

​Platform-Specific Issues

​iOS Simulator Issues

​Android Emulator Performance

​iOS Build Errors

​Android Build Errors

​Performance Issues

​Slow Inference Speed

​Slow Downloads

​Debugging Best Practices

​Enable Comprehensive Logging

​Check Model Files

​Profile Performance

​Getting Help

​Quick Checklist

​Next Steps

Build docs developers (and LLMs) love

Installation Issues

Resource Fetcher Not Initialized

Native Module Not Found

New Architecture Not Enabled

Model Loading Issues

Model Not Loading (isReady stays false)

Download Fails or Interrupts

Invalid Model File

Memory Issues

Out of Memory / App Crashes

Memory Warnings (iOS)

Generation Issues

Model Not Generating (hangs)

Empty or Unexpected Responses

ModelGenerating Error

Input/Output Issues

FileReadFailed Error

WrongDimensions Error

TokenizerError

Platform-Specific Issues

iOS Simulator Issues

Android Emulator Performance

iOS Build Errors

Android Build Errors

Performance Issues

Slow Inference Speed

Slow Downloads

Debugging Best Practices

Enable Comprehensive Logging

Check Model Files

Profile Performance

Getting Help

Quick Checklist

Next Steps