com.facebook.jni.CppException: [enforce fail at inline_container.cc:222] . file not found: archive/constants.pkl

Mobile
1

Hello Everyone. I was able to load the .pt file in python. But when I tried to load the same model in Java Android I was facing this issue.
Android Java Version:
‘org.pytorch:pytorch_android:1.6.0’
‘org.pytorch:pytorch_android_torchvision:1.7.0’
Python Version:
pytorch:1.6

 Caused by: com.facebook.jni.CppException: [enforce fail at inline_container.cc:222] . file not found: archive/constants.pkl
(no backtrace available)
    at org.pytorch.NativePeer.initHybrid(Native Method)
    at org.pytorch.NativePeer.<init>(NativePeer.java:23)
    at org.pytorch.Module.load(Module.java:24)
    at org.pytorch.Module.load(Module.java:34)
    at org.pytorch.helloworld.MainActivity.onCreate(MainActivity.java:39)
    at android.app.Activity.performCreate(Activity.java:7136)
    at android.app.Activity.performCreate(Activity.java:7127)
    at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271)

So please let me know, do I need to make any changes in Java Code while loading the model because I was able to load the sample model provided by the PyTorch Android team(version 1.4.0).

Is it related to versioning of pytorch? And Is pytorch java and pytorch versioning similar?

2

The version of gradle ‘org.pytorch:pytorch_android:1.6.0’ should be the same for pytorch binaries that you used to generate pt file.

Did you use 1.6.0 version for pt file?

3

I’m facing the exact same issue. However, I suspect mine is related to quantized model. I’m using PyTorch 1.7.1 to generate the model:

import torchvision, torch
model = torchvision.models.quantization.mobilenet_v2(pretrained=True, quantize=True)
torch.save(model.state_dict(), “model.pt”)

In Android app, I use:

dependencies {
implementation ‘androidx.appcompat:appcompat:1.1.0’
implementation ‘org.pytorch:pytorch_android:1.7.1’
implementation ‘org.pytorch:pytorch_android_torchvision:1.7.1’
}

I got:

Process: org.pytorch.helloworld, PID: 8644
java.lang.RuntimeException: Unable to start activity ComponentInfo{org.pytorch.helloworld/org.pytorch.helloworld.MainActivity}: com.facebook.jni.CppException: [enforce fail at inline_container.cc:222] . file not found: archive/constants.pkl
(no backtrace available)

Is there something to be done to run quantized model?
I also see the same error w/ Android benchmark binary and iOS benchmark app.

@IvanKobzarev any advice on this?

4

Never mind. I figured out my mistake.

5

Facing the same issue, could you share what fixed it for you?

Thanks!

Update: I’ve used PyTorch 1.6 for both save and load but the error remains.