Quantization made by Richard Erkhov.

[Github](https://github.com/RichardErkhov)

[Discord](https://discord.gg/pvy7H8DZMG)

[Request more models](https://github.com/RichardErkhov/quant_request)


gemma-2-baku-2b-it - bnb 4bits
- Model creator: https://huggingface.co/rinna/
- Original model: https://huggingface.co/rinna/gemma-2-baku-2b-it/




Original model description:
---
thumbnail: https://github.com/rinnakk/japanese-pretrained-models/blob/master/rinna.png
license: gemma
language:
- ja
- en
tags:
- gemma2
- conversational
base_model:
- google/gemma-2-2b
- google/gemma-2-2b-it
- rinna/gemma-2-baku-2b
base_model_relation: merge
pipeline_tag: text-generation
library_name: transformers
---


# `Gemma 2 Baku 2B Instruct (rinna/gemma-2-baku-2b-it)`

![rinna-icon](./rinna.png)

# Overview

The model is an instruction-tuned variant of [rinna/gemma-2-baku-2b](https://huggingface.co/rinna/gemma-2-baku-2b), utilizing Chat Vector and Odds Ratio Preference Optimization (ORPO) for fine-tuning. It adheres to the gemma-2 chat format.

| Size | Continual Pre-Training | Instruction-Tuning |
| :-   | :-                     | :-                 |
| 2B   | Gemma 2 Baku 2B [[HF]](https://huggingface.co/rinna/gemma-2-baku-2b) | Gemma 2 Baku 2B Instruct [[HF]](https://huggingface.co/rinna/gemma-2-baku-2b-it) |

* **Model architecture**

    A 26-layer, 2304-hidden-size transformer-based language model. Please refer to the [Gemma 2 Model Card](https://www.kaggle.com/models/google/gemma-2/) for detailed information on the model's architecture.

* **Training**

    **Model merging.** The base model was endowed with instruction-following capabilities through a chat vector addition process. The chat vector was derived by subtracting the parameter vectors of [google/gemma-2-2b](https://huggingface.co/google/gemma-2-2b) from [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it), as follows. 

    ~~~~text
    rinna/gemma-2-baku-2b + 1.0 * (google/gemma-2-2b-it - google/gemma-2-2b)
    ~~~~

    During this process, the embedding layer was excluded during the subtraction and addition of parameter vectors.
    
    **ORPO** was applied using a subset of the following dataset to further refine the performance of the merged model.

    - rinna's internal dataset
  
* **Contributors**

    - [Xinqi Chen](https://huggingface.co/Keely0419)
    - [Toshiaki Wakatsuki](https://huggingface.co/t-w)
    - [Kei Sawada](https://huggingface.co/keisawada)

---

# Benchmarking

Please refer to [rinna's LM benchmark page](https://rinnakk.github.io/research/benchmarks/lm/index.html).

---

# How to use the model

~~~~python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "rinna/gemma-2-baku-2b-it"
dtype = torch.bfloat16

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="cuda",
    torch_dtype=dtype,
    attn_implementation="eager",
)

chat = [
    { "role": "user", "content": "西田幾多郎とはどんな人物ですか？" },
]
prompt = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)

input_ids = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt").to(model.device)
outputs = model.generate(
    input_ids,
    max_new_tokens=512,
)

response = tokenizer.decode(outputs[0][input_ids.shape[-1]:], skip_special_tokens=True)
print(response)
~~~~

It is recommended to use eager attention when conducting batch inference under bfloat16 precision. 
Currently, Gemma 2 yields NaN values for input sequences with padding when the default attention mechanism (torch.scaled_dot_product_attention) is employed in conjunction with bfloat16.

---

# Tokenization
The model uses the original [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it) tokenizer.

---

# How to cite
```bibtex
@misc{rinna-gemma-2-baku-2b-it,
    title = {rinna/gemma-2-baku-2b-it},
    author = {Chen, Xinqi and Wakatsuki, Toshiaki and Sawada, Kei},
    url = {https://huggingface.co/rinna/gemma-2-baku-2b-it}
}

@inproceedings{sawada2024release,
    title = {Release of Pre-Trained Models for the {J}apanese Language},
    author = {Sawada, Kei and Zhao, Tianyu and Shing, Makoto and Mitsui, Kentaro and Kaga, Akio and Hono, Yukiya and Wakatsuki, Toshiaki and Mitsuda, Koh},
    booktitle = {Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)},
    month = {5},
    year = {2024},
    pages = {13898--13905},
    url = {https://aclanthology.org/2024.lrec-main.1213},
    note = {\url{https://arxiv.org/abs/2404.01657}}
}
```
---

# References
```bibtex
@article{gemma-2-2024,
    title = {Gemma 2},
    url = {https://www.kaggle.com/models/google/gemma-2},
    publisher = {Kaggle},
    author = {Gemma Team},
    year = {2024}
}

@article{huang2023chat,
    title = {Chat Vector: A Simple Approach to Equip LLMs with Instruction Following and Model Alignment in New Languages},
    author = {Huang, Shih-Cheng and Li, Pin-Zu and Hsu, Yu-Chi and Chen, Kuang-Ming and Lin, Yu Tung and Hsiao, Shih-Kai and Tzong-Han Tsai, Richard and Lee, Hung-yi},
    year = {2023},
    url = {https://arxiv.org/abs/2310.04799}
}

@article{hong2024orpo,
  title = {ORPO: Monolithic Preference Optimization without Reference Model},
  author = {Hong, Jiwoo and Lee, Noah and Thorne, James},
  year = {2024},
  url = {https://arxiv.org/abs/2403.07691}
}
```
---

# License
[Gemma Terms of Use](https://ai.google.dev/gemma/terms)