Strange output in oobabooga
I loaded this within oobabooga using the ExLlama_HF model loader, all seems fine.
Within the hash I ask it to.... "write me qt python code that has a gui to open a file and generate a sha1 hash" as a test
The response I get is this.
here it is
import os
from PyQt5.QtWidgets import QFileDialog
defclassfunctionmethod: def write_me_some_qt_pyqt5_python_code_that_has_a_gui_to_open_a_file_and_generate_a_sha1_hash():()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()
Am I using this wrong?
we need a proper template
It's not an instruct model, there should be no template.
This is a base model, not fine tuned for instructions. You use it by writing enough of what you want to give the right inertia and then the model continues.
Example:
There is an instruction trained one here: https://huggingface.co/TheBloke/CodeLlama-34B-Instruct-GPTQ