Re-creating the model from S3

Can someone share the config for the model that is stored in "s3://dlai-generative-ai/models/flan-dialogue-summary-checkpoint/ " from Lab_2_fine_tune_generative_ai_model notebook?

I have been trying different configs but need help fine-tuning the T5 model to perform similarly to the provided model. I’ve tried to increase the max step to 200 and change the epochs to 50, but the model performance could be better than the original mode.

It’d be beneficial to get all the configs that were used to improve the mentioned models in this notebook (Lab_2_fine_tune_generative_ai_model).


Hi Sohrab. Thanks for the feedback. I’ll ask if this information is available and let you know.

Hi. Sorry for the late follow up. I asked this a while back but unfortunately, the training info was lost. Our partner remembers though that the training ran for 5-6 hours on ml.g5.4xlarge to get that performance.