---
base_model:
- DataPilot/ArrowPro-7B-RobinHood
- Aratako/AntlerStar-RP
- ohwi/japanese-stablelm-instruct-gamma-7b-dpo-uf-v1
- stabilityai/japanese-stablelm-base-gamma-7b
- Aratako/Ninja-v1-RP-expressive
- Aratako/ArrowPro-7B-RobinHood-toxic
- Elizezen/Phos-7B-RP
- DataPilot/ArrowPro-7B-KillerWhale
- umiyuki/Umievo-itr012-Gleipnir-7B
- DataPilot/ArrowPro-7B-KUJIRA
library_name: transformers
tags:
- mergekit
- merge
language:
  - ja
pipeline_tag: text-generation
license: apache-2.0
---
# RP-7b-instruct

🚨 **This model is tuning to RP and knowledge is likely unstable.**

This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).


## Output example

```
[INST] <<SYS>>
あなたは日本語を話す優秀なアシスタントです。回答には必ず日本語で答えてください。また考える過程も出力してください。
<</SYS>>

まどマギで一番好きなキャラは？ [/INST]まどマギで一番好きなキャラクターは、鹿目まどかです。彼女は物語の中心人物であり、優しくて芯の強い性格が魅力的です。また、彼女が魔法少女としての運命を受け入れ、他の魔法少女たちを救うために奔走する姿には感動します。彼女の最期の選択も、多くの人々の心に残る名シーンとなっています。
```

## Test environment

This model was tested using [text-generation-webui](https://github.com/oobabooga/text-generation-webui/tree/main). I use preset `min_p` and `Null preset` with temperature=0.3 for Generation.


## Usage

This format must be adhered to strictly, as deviations may result in less optimal outputs from the model.

The template used to construct a prompt for the Instruct model is specified as follows:

```
<s>[INST] <<SYS>>\n{SYSTEM_PROMPT}\n<</SYS>>\n\n{USER_MESSAGE_1} [/INST] {BOT_MESSAGE_1}</s>[INST] {USER_MESSAGE_2} [/INST] 
```


Please be aware that ``<s>`` and ``</s>`` are special tokens used for the beginning of string (BOS) and end of string (EOS), respectively, while [INST] and [/INST] are considered regular strings.

For the "{SYSTEM_PROMPT}" part, We recommend using "あなたは日本語を話す優秀なアシスタントです。回答には必ず日本語で答えてください。また考える過程も出力してください。"

For the "{USER_MESSAGE_1}" part, We recommend using {instruction}\n{input}

In other words, We recommend the following:

``` 
<s>[INST] <<SYS>>\nあなたは日本語を話す優秀なアシスタントです。回答には必ず日本語で答えてください。また考える過程も出力してください。\n<</SYS>>\n\n{instruction1}\n{input1} [/INST] {BOT_MESSAGE_1}</s>[INST] {instruction2}\n{input2} [/INST] 
```

### Use the instruct model

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

model_name = "nitky/RP-7b-instruct"
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_name)

device = "cuda"

messages = [
    {"role": "system", "content": "あなたは日本語を話す優秀なアシスタントです。回答には必ず日本語で答えてください。また考える過程も出力してください。"},
    {"role": "user", "content": "まどマギで一番好きなキャラは？"}
]

encodeds = tokenizer.apply_chat_template(messages, return_tensors="pt")

model_inputs = encodeds.to(device)
model.to(device)

generated_ids = model.generate(model_inputs, max_new_tokens=256, do_sample=True, temperature=0.3)
decoded = tokenizer.batch_decode(generated_ids)
print(decoded[0])
```

## Merge Details
### Merge Method

This model was merged using the [Model Stock](https://arxiv.org/abs/2403.19522) merge method using [stabilityai/japanese-stablelm-base-gamma-7b](https://huggingface.co/stabilityai/japanese-stablelm-base-gamma-7b) as a base.

### Models Merged

The following models were included in the merge:
* [DataPilot/ArrowPro-7B-RobinHood](https://huggingface.co/DataPilot/ArrowPro-7B-RobinHood)
* [Aratako/AntlerStar-RP](https://huggingface.co/Aratako/AntlerStar-RP)
* [ohwi/japanese-stablelm-instruct-gamma-7b-dpo-uf-v1](https://huggingface.co/ohwi/japanese-stablelm-instruct-gamma-7b-dpo-uf-v1)
* [Aratako/Ninja-v1-RP-expressive](https://huggingface.co/Aratako/Ninja-v1-RP-expressive)
* [Aratako/ArrowPro-7B-RobinHood-toxic](https://huggingface.co/Aratako/ArrowPro-7B-RobinHood-toxic)
* [Elizezen/Phos-7B-RP](https://huggingface.co/Elizezen/Phos-7B-RP)
* [DataPilot/ArrowPro-7B-KillerWhale](https://huggingface.co/DataPilot/ArrowPro-7B-KillerWhale)
* [umiyuki/Umievo-itr012-Gleipnir-7B](https://huggingface.co/umiyuki/Umievo-itr012-Gleipnir-7B)
* [DataPilot/ArrowPro-7B-KUJIRA](https://huggingface.co/DataPilot/ArrowPro-7B-KUJIRA)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
merge_method: model_stock
base_model: stabilityai/japanese-stablelm-base-gamma-7b
models:
  - model: Aratako/AntlerStar-RP
  - model: Aratako/ArrowPro-7B-RobinHood-toxic
  - model: Aratako/Ninja-v1-RP-expressive
  - model: DataPilot/ArrowPro-7B-KUJIRA
  - model: DataPilot/ArrowPro-7B-RobinHood
  - model: DataPilot/ArrowPro-7B-KillerWhale
  - model: Elizezen/Phos-7B-RP
  - model: ohwi/japanese-stablelm-instruct-gamma-7b-dpo-uf-v1
  - model: umiyuki/Umievo-itr012-Gleipnir-7B
dtype: bfloat16
tokenizer_source: model:stabilityai/japanese-stablelm-base-gamma-7b
name: RP-7b-instruct
```