metadata

language:
  - en
  - fr
  - de
  - es
  - it
  - pt
  - ru
  - zh
  - ja
license: cc-by-nc-4.0
library_name: transformers
tags:
  - nsfw
  - Visual novel
  - roleplay
  - mergekit
  - merge
base_model:
  - mistralai/Mistral-Nemo-Instruct-2407
  - NeverSleep/Lumimaid-v0.2-12B
  - Epiculous/Violet_Twilight-v0.1
  - Sao10K/MN-12B-Lyra-v4
  - anthracite-org/magnum-v2-12b
pipeline_tag: text-generation
model-index:
  - name: ChatWaifu_v1.4
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: IFEval (0-Shot)
          type: HuggingFaceH4/ifeval
          args:
            num_few_shot: 0
        metrics:
          - type: inst_level_strict_acc and prompt_level_strict_acc
            value: 56.91
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=spow12/ChatWaifu_v1.4
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: BBH (3-Shot)
          type: BBH
          args:
            num_few_shot: 3
        metrics:
          - type: acc_norm
            value: 31.63
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=spow12/ChatWaifu_v1.4
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MATH Lvl 5 (4-Shot)
          type: hendrycks/competition_math
          args:
            num_few_shot: 4
        metrics:
          - type: exact_match
            value: 7.85
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=spow12/ChatWaifu_v1.4
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GPQA (0-shot)
          type: Idavidrein/gpqa
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 7.61
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=spow12/ChatWaifu_v1.4
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MuSR (0-shot)
          type: TAUR-Lab/MuSR
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 20.03
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=spow12/ChatWaifu_v1.4
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU-PRO (5-shot)
          type: TIGER-Lab/MMLU-Pro
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 27.5
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=spow12/ChatWaifu_v1.4
          name: Open LLM Leaderboard

Model Card for Model ID

Merged model using mergekit

This model aimed to act like visual novel character.

Merge Format

models:
    - model: spow12/ChatWaifu_modify_data
    - model: anthracite-org/magnum-v2-12b
    - model: Sao10K/MN-12B-Lyra-v4
    - model: Gryphe/Pantheon-RP-1.6-12b-Nemo
    - model: mistralai/Mistral-Nemo-Instruct-2407
    - model: NeverSleep/Lumimaid-v0.2-12B
    - model: Epiculous/Violet_Twilight-v0.1
merge_method: model_stock
base_model: spow12/ChatWaifu_modify_data
dtype: bfloat16

you have to resize chatwaifu and lucimaid's embedding size(131073 to 131072).

WaifuModel Collections

Unified demo

WaifuAssistant

Update

2024.09.10 Update Ver 1.4
- Modify data format and applying flitering.
- Merge with model stock
2024.08.29 Update Ver 1.3.1
- Merge Ver1.2, mistralai/Mistral-Nemo-Instruct-2407 and NeverSleep/Lumimaid-v0.2-12B, Epiculous/Violet_Twilight-v0.1
- Adjust merge weight.
2024.08.16 Update Ver 1.3
- Merge Ver1.2, mistralai/Mistral-Nemo-Instruct-2407 and NeverSleep/Lumimaid-v0.2-12B,
2024.08.08 Update Ver 1.2.1
- Merge Ver1.2 and mistralai/Mistral-Nemo-Instruct-2407
2024.08.07 Update Ver 1.2
- Add Preference Learning in training pipeline
2024.07.29 Update Ver 1.1
- Add dataset format -> generate novel, fill masked sentences
- Remove system role and integrate at user message.
- Remove 『』 in conversation.
2024.06.20 Upload other chara's sample chat history.
2024.06.13 Upload Model

Model Details

Model Description

Developed by: spow12(yw_nam)
Shared by : spow12(yw_nam)
Model type: CausalLM
Language(s) (NLP): japanese
Finetuned from model : NeverSleep/Lumimaid-v0.2-12B

Currently, chatbot has below personality.

character	visual_novel
ムラサメ	Senren＊Banka
茉子	Senren＊Banka
芳乃	Senren＊Banka
レナ	Senren＊Banka
千咲	Senren＊Banka
芦花	Senren＊Banka
愛衣	Café Stella and the Reaper's Butterflies
栞那	Café Stella and the Reaper's Butterflies
ナツメ	Café Stella and the Reaper's Butterflies
希	Café Stella and the Reaper's Butterflies
涼音	Café Stella and the Reaper's Butterflies
あやせ	Riddle Joker
七海	Riddle Joker
羽月	Riddle Joker
茉優	Riddle Joker
小春	Riddle Joker

Feature

Fluent Chat performance
Reduce repetition problem when generate with many turn(over 20~30)
Zero Shot character persona using description of character.
128k context window
Memory ability that does not forget even after long-context generation

Demo

You can use Demo in google colab.

Check Here

Future Work.

Now, i'm quite satisfying the model chat performance.

So, i'm going to focus for integrating the vision modality to model so that our waifu can do more general tasks.

Bias, Risks, and Limitations

This model trained by japanese dataset included visual novel which contain nsfw content.

So, The model may generate NSFW content.

Use & Credit

This model is currently available for non-commercial & Research purpose only. Also, since I'm not detailed in licensing, I hope you use it responsibly.

By sharing this model, I hope to contribute to the research efforts of our community (the open-source community and anime persons).

This repository can use Visual novel-based RAG, but i will not distribute it yet because i'm not sure if it is permissible to release the data publicly.

Citation

@misc {ChatWaifu_v1.4,
    author       = { YoungWoo Nam },
    title        = { ChatWaifu_v1.4 },
    year         = 2024,
    url          = { https://huggingface.co/spow12/ChatWaifu_v1.4 },
    publisher    = { Hugging Face }
}

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	25.25
IFEval (0-Shot)	56.91
BBH (3-Shot)	31.63
MATH Lvl 5 (4-Shot)	7.85
GPQA (0-shot)	7.61
MuSR (0-shot)	20.03
MMLU-PRO (5-shot)	27.50