Model Card for Model ID
kAIra-gemma-2-2b is a finetuned version of google/gemma-2-2b-it can be used in French to translate any text into a slang version of it.
Model Details
Model Description
- Developed by: Sébastien Fischman
- Funded by [optional]: None
- Shared by [optional]: Sébastien Fischman
- Model type: [More Information Needed]
- Language(s) (NLP): French
- License: MIT
- Finetuned from model [optional]: google/gemma-2-2b-it
Model Sources [optional]
Uses
This model is a toy example to showcase how to teach an LLM French slang, it can be used to generate funny translation in argot.
Bias, Risks, and Limitations
Since the model is trained to translate in slang, some generated content can be considered offensive or inappropriate.
Recommendations
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
How to Get Started with the Model
Use the code from Kaggle here : https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/ Or use the HF Space: Coming
[More Information Needed]
Training Details
Training Data
https://www.kaggle.com/datasets/optimo/french-kaira-training
Training Procedure
Training code is fully accesible here: https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/
Training Hyperparameters
- Training regime: bf16
[More Information Needed]
Framework versions
- PEFT 0.14.0