# Multi-model Fusion

GodSlayer-12B-ABYSS is a 12B-parameter model obtained by merging multiple pre-trained language models through mergekit, designed to reduce positivity bias while maintaining stability and coherence, and to enhance realism and diversified response capabilities.

A merged model based on Llama 3 70B, optimized for story creation and role-playing, supporting long context windows

This is a specialized version of the Llama-3-70B model for roleplay and story generation, optimized through multi-model merging and EXL2 quantization

A merged model based on Llama-3-70B, specializing in story generation and role-play (RP) tasks, combining multiple high-quality models via the breadcrumbs_ties method

This is a fusion model based on the Llama-3 architecture, specifically designed for mental health therapy and psychotherapist dialogue scenarios, integrating multiple fine-tuned models in the mental health domain.

This is a multi-model fusion result based on the Qwen2.5-14B architecture, utilizing the Model Stock fusion method, combining 22 different 14B-parameter scale models from various sources.

This is a pre-trained language model merged using mergekit and the DARE TIES method, based on multiple Gemma-2-9B variant models.

A pre-trained language model merged using the DARE TIES method, based on L-BASE-V1 and fusing multiple MERGE models

Llama-3-Chinese-8B-Instruct-v3 is a Chinese instruction model fine-tuned from multiple hybrid models, suitable for dialogue, Q&A, and similar scenarios.

Smart-Lemon-Cookie-7B is a pre-trained language model merged using mergekit, combining the strengths of multiple models, suitable for text generation and role-playing tasks.

ECE-TW3-JRGL-V5 is a new model obtained by merging the MoMo-72B-lora-1.8.7-DPO and alpaca-dragon-72b-v1 models through mergekit, integrating the advantages of multiple models.

Neuro 4 Math-7b is a specialized mathematical model created by merging multiple 7B-parameter models using the LazyMerge toolkit, focusing on math-related tasks.

A 7B-parameter large language model focused on role-playing, featuring 8192 context length with emphasis on creativity and reduced clichés

A MoE (Mixture of Experts) Mistral model focused on role-playing, combining the strengths of role-playing and story creation

Astrea-RP-v1-3B is a 3B-parameter large language model that integrates multiple models through task arithmetic, focusing on text generation tasks.

Nethena-20B is a 20B-parameter large language model developed by NeverSleep, suitable for role-playing, emotional interaction, and general purposes.

Athena v4 is an experimental large language model suitable for role-playing, emotional role-playing, and general scenarios. Uses Alpaca format prompt templates.

A specialized model for roleplay and creative work, fine-tuned from multiple LLaMA2 70B models, combining features from Nous-Hermes-Llama2-70b, Xwin-LM-7B-V0.1, and Mythospice-70b

A speech recognition model fine-tuned on Taiwan Mandarin (zh-tw) based on facebook/wav2vec2-large-xlsr-53, supporting 16kHz sampling rate audio input