Any plans to release the evaluation data or evaluate AraGPT2 1.5B ?
the evaluation dataset can serve as a good benchmark for the evaluation of future Arabic LLMs.
Also are you planning to evaluate AraGPT2 1.5B? was the grover modeling file causing issues?
may i ask u please if u already installed and ran the model how to do so, cause am facing issues and dontknow how to handle?
@wissamantoun
@MazenSiraj why ? what was the issue or the error?
did you try the code snippet provided, I tried it and worked perfectly.
Can you share your problem in different post so we can help
Hello @sherif1 , it was the offload folder path issue and I managed to handle it earlier.
can you tell me how did you run the model because every time I run the sample code it takes so long to run and downloads the model all over again.
after downloading it for the first time and storing it say in a variable called model and tokenizer , save the mode and tokenizer using
model.save_pretrained(<ur path>)
tokenizer.save_pretrained(<ur path>)
The next time, don't use AutoTokenizer.from_pretrained('inception-mbzuai/jais-13b-chat')
Use ur local path like this AutoTokenizer.from_pretrained(<ur path>)