IAJw

/

flan-alpaca-base-18378

like 0

Used scraped data as input, ChatGPT result as output and limited types of instruction to fine-tune flan-t5-small.
Used declare-lab repo. Refer to https://github.com/declare-lab/flan-alpaca
Epoch set to 1
Input token max 512
Output token max 512
Trained on Free Google Colab T4
Training time ~ 40 minutes