|
--- |
|
library_name: peft |
|
base_model: NousResearch/Llama-2-7b-chat-hf |
|
datasets: |
|
- richardcsuwandi/oasst-javanese |
|
language: |
|
- jv |
|
--- |
|
|
|
# Description |
|
|
|
This model is a fine-tuned adaptation of [Llama-2-7b-chat-hf](https://huggingface.co/NousResearch/Llama-2-7b-chat-hf), specifically tailored for the Javanese language. It has been developed to function as a chatbot, offering fluent and accurate responses in Javanese. The fine-tuning process utilized a dataset translated into Javanese, aiming to improve the representation of the Javanese language within language models and chatbot technologies. |
|
|
|
## Training |
|
|
|
The model was fine-tuned on a dataset translated into Javanese using the [NLLB](https://ai.meta.com/research/no-language-left-behind/) model. This dataset includes texts from both [OASST1](https://huggingface.co/datasets/OpenAssistant/oasst1) and [OASST2](https://huggingface.co/datasets/OpenAssistant/oasst2), covering a wide range of conversational scenarios. The training process employed [PEFT](https://github.com/huggingface/peft) and [TRL](https://github.com/huggingface/trl) to facilitate efficient and rapid fine-tuning. |
|
|
|
## Usage |
|
|
|
This model is intended for deployment as a conversational chatbot in Javanese. It is suitable for a variety of applications that require natural language understanding in Javanese. Users can interact with the model using the standard Hugging Face Transformers pipeline for text generation. |
|
|
|
```python |
|
from peft import PeftModel, PeftConfig |
|
from transformers import AutoModelForCausalLM |
|
|
|
config = PeftConfig.from_pretrained("richardcsuwandi/llama2-javanese") |
|
base_model = AutoModelForCausalLM.from_pretrained("NousResearch/Llama-2-7b-chat-hf") |
|
model = PeftModel.from_pretrained(base_model, "richardcsuwandi/llama2-javanese") |
|
```` |
