-
Notifications
You must be signed in to change notification settings - Fork 478
Open
Labels
bug-unconfirmedUnconfirmed bugsUnconfirmed bugs
Description
Prerequisites
Before submitting your issue, please ensure the following:
- [v] I am running the latest version of PowerInfer. Development is rapid, and as of now, there are no tagged versions.
- [v] I have carefully read and followed the instructions in the README.md.
- [v] I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
Expected Behavior
No problem reported by UBSAN
Current Behavior
PowerInfer/build/bin/finetune
error loading model: llama_model_loader: failed to load model from
llama_load_model_from_file_with_context: failed to load model
PowerInfer/ggml.c:19835:23: runtime error: member access within null pointer of type 'const struct gguf_context'
Environment and Context
PowerInfer: master
compiler: gcc (Debian 12.2.0-14) 12.2.0
- Physical (or virtual) hardware you are using, e.g. for Linux:
Ubuntu 22.04.4 LTS
Linux 5.15.0-125-generic #135-Ubuntu SMP Fri Sep 27 13:53:58 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux
Failure Information (for bugs)
Please help provide information about the failure / bug.
Steps to Reproduce
Launched PowerInfer/build/bin/finetune
I’ve scanned PowerInfer with UBSAN and it has triggered on PowerInfer/ggml.c line 19835. I'm not sure whether this is an actual issue, or whether code of PowerInfer is too sophisticated for UBSAN to understand. Do you guys have any idea?
Failure Logs
Metadata
Metadata
Assignees
Labels
bug-unconfirmedUnconfirmed bugsUnconfirmed bugs