Twilight-SCE-12B-v2 Open-Source Large Language Model - Multi-Model Fusion Intelligent Dialogue Supporting ChatML Markup

Twilight SCE 12B V2

Developed by yamatazen

This is a 12B-parameter large language model based on the Mistral architecture, employing the SCE method to merge multiple pre-trained models and incorporating ChatML token support.

Large Language Model

Transformers

Supports Multiple Languages#Multilingual Chat #SCE Fusion Model #ChatML Support

Downloads 66

Release Time : 4/21/2025

Model Overview

This model is created by merging multiple 12B-parameter pre-trained language models, with special support for ChatML tokens, making it suitable for chat and text generation tasks.

Model Features

Multi-model Merging

Utilizes the SCE method to merge 5 different 12B-parameter models, combining their strengths

ChatML Support

Incorporates ChatML tokens, making it more suitable for building dialogue systems

Multilingual Capabilities

Supports processing in both English and Japanese

Model Capabilities

Text generation

Dialogue systems

Multilingual processing

Use Cases

Dialogue systems

Intelligent Chatbot

Build chatbots capable of multi-turn conversations

Content creation

Creative Writing Assistance

Assist writers with creative writing and content generation

🚀 Twilight-SCE-12B-v2

This is a Mistral model with ChatML tokens added to the tokenizer, created by merging pre - trained language models.

image/png

🚀 Quick Start

This is a Mistral model with ChatML tokens added to the tokenizer. It's a merge of pre - trained language models created using mergekit.

✨ Features

Multi - model Merge: Combines multiple pre - trained language models to potentially enhance performance.
ChatML Support: Has ChatML tokens added to the tokenizer, suitable for chat - based applications.

📚 Documentation

Merge Details

Merge Method

This model was merged using the SCE merge method using [yamatazen/Shisa - v2 - Mistral - Nemo - 12B - Abliterated](https://huggingface.co/yamatazen/Shisa - v2 - Mistral - Nemo - 12B - Abliterated) as a base.

Models Merged

The following models were included in the merge:

[Elizezen/Himeyuri - v0.1 - 12B](https://huggingface.co/Elizezen/Himeyuri - v0.1 - 12B)
[nbeerbower/mistral - nemo - gutenberg - 12B - v4](https://huggingface.co/nbeerbower/mistral - nemo - gutenberg - 12B - v4)
[yamatazen/Himeyuri - Magnum - 12B](https://huggingface.co/yamatazen/Himeyuri - Magnum - 12B)
[PocketDoc/Dans - PersonalityEngine - V1.1.0 - 12b](https://huggingface.co/PocketDoc/Dans - PersonalityEngine - V1.1.0 - 12b)
[inflatebot/MN - 12B - Mag - Mell - R1](https://huggingface.co/inflatebot/MN - 12B - Mag - Mell - R1)

Configuration

The following YAML configuration was used to produce this model:

base_model: yamatazen/Shisa-v2-Mistral-Nemo-12B-Abliterated
models:
  - model: Elizezen/Himeyuri-v0.1-12B
    parameters:
      weight: 1.0
  - model: yamatazen/Himeyuri-Magnum-12B
    parameters:
      weight: 0.6
  - model: nbeerbower/mistral-nemo-gutenberg-12B-v4
    parameters:
      weight: 0.5
  - model: inflatebot/MN-12B-Mag-Mell-R1
    parameters:
      weight: 0.3
  - model: PocketDoc/Dans-PersonalityEngine-V1.1.0-12b
    parameters:
      weight: 0.3
merge_method: sce
dtype: bfloat16
parameters:
  normalize: true
  select_topk: 0.5
tokenizer:
  source: union

Property	Details
Base Models	Elizezen/Himeyuri - v0.1 - 12B, nbeerbower/mistral - nemo - gutenberg - 12B - v4, yamatazen/Himeyuri - Magnum - 12B, PocketDoc/Dans - PersonalityEngine - V1.1.0 - 12b, inflatebot/MN - 12B - Mag - Mell - R1, yamatazen/Shisa - v2 - Mistral - Nemo - 12B - Abliterated
Library Name	transformers
Tags	mergekit, merge, chatml
Language	en, ja

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご