|
--- |
|
language: |
|
- zh |
|
license: gemma |
|
tags: |
|
- text-generation-inference |
|
- transformers |
|
- unsloth |
|
- gemma2 |
|
- trl |
|
- sft |
|
- synthetic data |
|
base_model: unsloth/gemma-2-9b-it-bnb-4bit |
|
--- |
|
|
|
This model is a fine-tune of [gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it), trained for 3 epochs on a synthetic dataset created from the book *Liaozhai Zhiyi*. The stories in the book were translated by an LLM to modern Simplified Chinese and paired with generated writing prompts. The untranslated version of the stories can be found in this dataset: [CoolSpring/liaozhai-zhiyi](https://huggingface.co/datasets/CoolSpring/liaozhai-zhiyi). |
|
|
|
*Liaozhai Zhiyi*, also known as *[Strange Tales from a Chinese Studio](https://en.wikipedia.org/wiki/Strange_Tales_from_a_Chinese_Studio)*, is a collection of approximately 500 stories written in the traditional Chinese [Zhiguai](https://en.wikipedia.org/wiki/Zhiguai_xiaoshuo) and [Chuanqi](https://en.wikipedia.org/wiki/Chuanqi_(short_story)) styles by Pu Songling. The aim of this fine-tuning attempt is to explore incorporating these characteristics into the storytelling capabilities of a specific model. |
|
|
|
**Disclaimer:** Users should be aware that due to the historical nature of the training materials, it may generate biased content that reflects the cultural norms and perspectives of the author's era (late 17th to early 18th century China). These outputs may not align with contemporary values and should be interpreted with appropriate historical context. |
|
|
|
**Q4_K_M GGUF:** [CoolSpring/gemma-2-9b-it-liaozhai-Q4_K_M-GGUF](https://huggingface.co/CoolSpring/gemma-2-9b-it-liaozhai-Q4_K_M-GGUF) |
|
|
|
**Prompt Template - gemma** |
|
|
|
``` |
|
<bos><start_of_turn>user |
|
{input}<end_of_turn> |
|
<start_of_turn>model |
|
{output}<end_of_turn> |
|
``` |
|
|
|
Users must adhere to [Gemma Terms of Use](https://ai.google.dev/gemma/terms) when using this model. |
|
|
|
## Unsloth Metadata |
|
|
|
- **Developed by:** CoolSpring |
|
- **License:** gemma |
|
- **Finetuned from model :** unsloth/gemma-2-9b-it-bnb-4bit |
|
|
|
This gemma2 model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library. |
|
|
|
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth) |
