Optimized fine tuning config for RTX 4070 ti super
#5
by milad13731995 - opened
Hi, i have a RTX 4070 ti super, what params should i change in the config file run_ocr_vl_sft_16k.yaml?
I want to fine tune with Bengali training dataset.
milad13731995 changed discussion title from optimized fine tuning config for RTX 4070 ti super to Optimized fine tuning config for RTX 4070 ti super
Does PaddleOCR-VL-1.6 have official fine tuning support or tutorial?
Because I was fine tuning with hugging face trl, but the model was is hallucinating a lot.
Is there any valid framework to finetune this?