[Android demo app] Load failed: 34

ExecuTorch
1

I exported the stories110M model to ExecuTorch using a Colab notebook on ExecuTorch v0.2.0’s PyPI pip package. I then downloaded the exported model to my device and tried to open it in ExecuTorchLlamaDemo v0.2.0, which I downloaded from GitHub Actions. The specific steps that I took are documented at Executorch LLM tutorial · GitHub.

Unfortunately, I’ve encountered the error “Load failed: 34” after trying to open it in GitHub Actions.

Screenshot showing the error "Load failed: 34" in front of a message saying "Total RAM: 7400 MB. Available RAM: 2177 MB. Model path: /data/local/tmp/llama/xnnpack_llama2.pte Tokenizer path: /data/local/tmp/llama/tokenizer.bin Model loaded time: 20 ms"
Screenshot showing the error "Load failed: 34" in front of a message saying "Total RAM: 7400 MB. Available RAM: 2177 MB. Model path: /data/local/tmp/llama/xnnpack_llama2.pte Tokenizer path: /data/local/tmp/llama/tokenizer.bin Model loaded time: 20 ms"1080×2340 110 KB

Does anyone know what is the cause of this issue or how I can fix it?

2

Sorry about the slow response @ethanc8. If this is still happening, could you please create a bug report with the details?

34 is 0x22, so sounds like there’s an AccessFailed error when loading the model. Maybe a file is missing?

3

Hi @ethanc8 Could you please try with the latest code?

4

Sorry, I didn’t see this. I will try testing with the latest code sometime soon.