emre570's picture
Update README.md
0cdc944 verified
metadata
base_model: google/gemma-2-2b
library_name: peft
license: apache-2.0
datasets:
  - cenfis/alpaca-turkish-combined
language:
  - en
  - tr

Model Description

This model is originally created for Kaggle's competition here.

You can also see the model from Kaggle here.

Trained with ~80k Turkish dataset with 3 epochs. Took around 19 hours with 2x RTX 4090 GPUs.

You can use the model from PEFT, Transformers and Kaggle.

Important Notes

  • Use the model with a CUDA supported GPU since it's fine-tuned with bitsandbytes.

Fine-tuned by emre570.