|
--- |
|
base_model: unsloth/gemma-2-9b-bnb-4bit |
|
language: |
|
- en |
|
license: apache-2.0 |
|
tags: |
|
- text-generation-inference |
|
- transformers |
|
- unsloth |
|
- gemma2 |
|
- trl |
|
--- |
|
|
|
# gemma-2-9b-tagalog-chat: |
|
* Tagalog model fine-tuned on [this dataset](https://huggingface.co/datasets/922-Narra/tagaloguanaco_cleaned_03152024) |
|
* Base: [gemma 2 9b](https://huggingface.co/google/gemma-2-9b-it) |
|
* [GGUF](https://huggingface.co/922-Narra/gemma-2-9b-tagalog-chat-gguf) |
|
|
|
### USAGE |
|
This model is meant to be a chat model that has better Tagalog capabilities. |
|
|
|
Best results with "Human" and "Assistant" and prompt with Tagalog. Example: |
|
|
|
"Ito ay isang chat log sa pagitan ng AI Assistant na nagta-Tagalog at isang Pilipino. Magsimula ng chat:\nHuman: Hello po?\nAssistant:" |
|
|
|
### HYPERPARAMS |
|
* Trained for 1 epoch |
|
* rank: 32 |
|
* lora alpha: 32 |
|
* lora dropout: 0 |
|
* lr: 2e-4 |
|
* batch size: 2 |
|
* grad steps: 4 |
|
|
|
### WARNINGS AND DISCLAIMERS |
|
There is always the chance that the model may hallucinate. |
|
|
|
At times, it may also still switch to English or Taglish (most instances of this occurring when tested for coding). |
|
|
|
Finally, this model is not guaranteed to output aligned or safe outputs nor is it meant for production use - use at your own risk! |
