Karol Horosin sorry it wasn't your tutorial. I was confused. but kindly if you have knowledge about fine-tune gpt3.5 on binary classification task. i have a more than 200 000 samples. i think i will need just part of it.

i training gpt3.5 for binary classification task i would like to know the accuracy of fine-tuned model. i got a csv file which include 4 columns [step train_loss train_accuracy valid_loss valid_mean_token_accuracy] but the accuracy values is fluctuated is that mean my model doesn't work as appropriate ? Karol Horosin

it's a detailed article with all useful information. i fine-tuned gpt3.5 with train and valid sets, i got the result file but i don't know how to analyse it! as the train accuracy column values fluctuate with 0.333 - 0.66 - 0,66-0- 0,66- 1e-05- 0-0.66-0.7-1 . does this mean i need more samples for training ?