#Error with Model Loading in Ollama

3 messages · Page 1 of 1 (latest)

amber kiln
#

I'm encountering a problem when training with the model unsloth/Meta-Llama-3.1-8B-bnb-4bit on a Linux system. Despite updating all packages and rebuilding the trained model multiple times, the error persists. Below is the error message I'm receiving:

scss
Copy code
(llama3.1) Ubuntu@0009-kci-prxmx10105:~$ ollama run roo5150/lisa
Error: llama runner process has terminated: error loading model: done_getting_tensors: wrong number of tensors; expected 292, got 291
llama_load_model_from_file: exception loading model
Environment Details:

OS: Ubuntu
Model: unsloth/Meta-Llama-3.1-8B-bnb-4bit
All packages are up to date
Is there a known bug or issue with this model or version? Has anyone else encountered this issue? Any help or guidance would be greatly appreciated.

mental spoke
#

Emmm, i also met the same problem, and this problem appeared in LMStudio too. Maybe llama.cpp not support yet

amber kiln
#

sigh