OSError: rtf_checkpoints/not-best-disc-model does not appear to have a file named config.json. about realtabformer

[limhasic]

me too shit

[limhasic]

Dear my Ottoman friend,

i found something, its work on based before

!pip install transformers==4.24.0

cuda - pytorch env is very important to RUN

Before not matching enviroment -> we encounter error like

1. RuntimeError: NCCL Error 2: unhandled system error (run with NCCL_DEBUG=INFO for details)
2. old driver

but not finished,

we encounter error : out of memory

I thought I was running out of memory, so I changed the environment 3 times and in the end I used it almost like a super computer, but it said I was out of memory. Oh, fuck, the computer is great, why is that happening?

The answer was in

" output_max_length=None, "

I just accepted unlimited tokens, so 80GB or whatever was blown. Plus, I had to do it separately for multi-GPU settings, so I went crazy.
Is this a reasonably even number, like 1024 or 2048? This seems to be part of the rule regarding tokens, but anyway, it works because I set it to 1024 on the A100 80GB.

So now I'm going to test this with the HMA model and generate AIRBNB data.

have a nice day friend