Modelpublic

gvij/gpt-j-code-alpaca-instruct

sourceHugging Faceupdated 3y agoView on Hugging Face

0likes10downloads

Model Card

We finetuned GPT-J 6B on Code-Alpaca-Instruct Dataset (ewof/code-alpaca-instruct-unfiltered) for 5 epochs or ~ 25,000 steps using MonsterAPI no-code LLM finetuner.

This dataset is HuggingFaceH4/CodeAlpaca_20K unfiltered, removing 36 instances of blatant alignment.

The finetuning session got completed in 206 minutes and costed us only $8 for the entire finetuning run!

Hyperparameters & Run details:

—Model Path: EleutherAI/gpt-j-6b
—Dataset: ewof/code-alpaca-instruct-unfiltered
—Learning rate: 0.0003
—Number of epochs: 5
—Data split: Training: 90% / Validation: 10%
—Gradient accumulation steps: 1

Loss metrics: [image]

license: apache-2.0 ---