Update README.md

README.md

@@ -1,42 +1,53 @@
----
-base_model: []
-library_name: transformers
-tags:
-- mergekit
-- merge
-
----
-# Ninja-RP-MS
-
-This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
-
-## Merge Details
-### Merge Method
-
-This model was merged using the [Model Stock](https://arxiv.org/abs/2403.19522) merge method using ./Ninja-v1-NSFW-RP as a base.
-
-### Models Merged
-
-The following models were included in the merge:
-* ./Ninja-v1-NSFW-RP-SiliconMaid
-* ./Ninja-v1-NSFW-RP-LoyalMacaroniMaid
-* ./Ninja-v1-NSFW-RP-WestLake
-* ./Ninja-v1-NSFW-RP-Kunoichi
-
-### Configuration
-
-The following YAML configuration was used to produce this model:
-
-```yaml
-models:
-  - model: ./Ninja-v1-NSFW-RP
-  - model: ./Ninja-v1-NSFW-RP-Kunoichi
-  - model: ./Ninja-v1-NSFW-RP-SiliconMaid
-  - model: ./Ninja-v1-NSFW-RP-WestLake
-  - model: ./Ninja-v1-NSFW-RP-LoyalMacaroniMaid
-merge_method: model_stock
-base_model: ./Ninja-v1-NSFW-RP
-dtype: bfloat16
-tokenizer_source: union
-
-```
+---
+license: apache-2.0
+datasets:
+- Aratako/Rosebleu-1on1-Dialogues-RP
+- Aratako/LimaRP-augmented-ja-karakuri
+- grimulkan/LimaRP-augmented
+- Aratako/Bluemoon_Top50MB_Sorted_Fixed_ja
+- SicariusSicariiStuff/Bluemoon_Top50MB_Sorted_Fixed
+- OmniAICreator/Japanese-Roleplay
+language:
+- ja
+library_name: transformers
+tags:
+- roleplay
+base_model:
+- Aratako/Ninja-v1-RP-WIP
+---
+
+# Ninja-v1-RP
+
+## 概要
+This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
+
+[Aratako/Ninja-v1-RP-WIP](https://huggingface.co/Aratako/Ninja-v1-RP-WIP)をベースに、Task Vectorの加算・Model Stockによるマージを行い指示追従能力と表現力を強化したロールプレイ用モデルです。
+
+## マージの詳細
+まず、[Aratako/Ninja-v1-RP-WIP](https://huggingface.co/Aratako/Ninja-v1-RP-WIP)に対し、以下4モデルのTask Vectorを0.8倍して加算し、4種類、Task Vector加算モデルを作成しました。
+
+- [senseable/WestLake-7B-v2](https://huggingface.co/senseable/WestLake-7B-v2)
+- [SanjiWatsuki/Kunoichi-DPO-v2-7B](https://huggingface.co/SanjiWatsuki/Kunoichi-DPO-v2-7B)
+- [SanjiWatsuki/Silicon-Maid-7B](https://huggingface.co/SanjiWatsuki/Silicon-Maid-7B)
+- [SanjiWatsuki/Loyal-Macaroni-Maid-7B](https://huggingface.co/SanjiWatsuki/Loyal-Macaroni-Maid-7B)
+
+各モデルのTask Vectorの加算の式は以下の通りです。
+```
+new_model = Ninja-v1-RP-WIP + 0.8 * (target_model - Mistral-7B-v0.1)
+```
+
+次に、このTask Vector加算によってできた4モデルと元のモデルを、Model Stockという手法を用い以下のようなconfigを使ってmergekitでマージし、このモデルを作成しました。
+
+```yaml
+models:
+  - model: ./Ninja-v1-RP-WIP
+  - model: ./Ninja-v1-RP-WIP-Kunoichi
+  - model: ./Ninja-v1-RP-WIP-SiliconMaid
+  - model: ./Ninja-v1-RP-WIP-WestLake
+  - model: ./Ninja-v1-RP-WIP-LoyalMacaroniMaid
+merge_method: model_stock
+base_model: ./Ninja-v1-RP-WIP
+dtype: bfloat16
+tokenizer_source: union
+
+```