Llama-3-70b merged model goes open-source! Specifically designed for story generation and role-playing, supports long contexts

Llama 3 70b Arimas Story RP V1.6 4.0bpw H6 Exl2

Developed by kim512

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

Large Language Model

Transformers

#Long-context reasoning #Role-play optimization #Multi-model fusion

Downloads 20

Release Time : 4/25/2025

Model Overview

This is a model created by merging multiple Llama 3 70B variants using mergekit, focusing on story creation and role-play scenarios, with special optimization for long text generation capabilities

Model Features

Multi-model merge

Integrates 8 different Llama 3 70B variants, including excellent models like Smaug and Synthia

Long context support

Specially optimized for long context window performance, supporting up to 8192 tokens context length

Role-play optimization

Specifically optimized for story creation and role-play scenarios

Multiple quantization versions

Provides various EXL2 quantization versions from 3.0 to 8.0 bits/weight to meet different hardware requirements

Model Capabilities

Long text generation

Role-play dialogue

Story creation

Instruction following

Use Cases

Creative writing

Story generation

Generate coherent long-form story content

Can produce multi-chapter stories with rich plots

Entertainment applications

Role-play chat

Engage in immersive dialogue with AI characters

Provides vivid character interaction experiences

🚀 EXL2 quants of ryzen88/Llama-3-70b-Arimas-story-RP-V1.6

This project provides EXL2 quantization versions of the ryzen88/Llama-3-70b-Arimas-story-RP-V1.6 model, offering different bit - per - weight options to meet various performance and storage requirements.

Available Quantization Versions

Creation Details

Created using the defaults from exllamav2 1.4.0 convert.py
3.0bpw to 6.0bpw head bits = 6
8.0bpw head bits = 8
length = 8192
dataset rows = 200
measurement rows = 32
measurement length = 8192

📚 Documentation

Model Introduction

The Llama-3-70b-Arimas-story-RP-V1.6 is a merged pre - trained language model created using mergekit.

Merge Details

Expansion and Experimentation: Greatly expanded the number of models used in this merge and experimented extensively with different ideas. This version seems more convincing than V1.5. It is hoped that the long - context window will remain robust after quantization.
Data Type Change: Switched back from BFloat to Float due to multiple merges.
Failed Experiment: Tried breadcrumbs without the Ties, which turned out very poorly.

Merge Method

This model was merged using the breadcrumbs_ties merge method, with I:\Llama-3-70B-Instruct-Gradient-262k as the base model.

Models Merged

The following models were included in the merge:

\Smaug-Llama-3-70B-Instruct
\Meta-LLama-3-Cat-Smaug-LLama-70b
\Meta-LLama-3-Cat-A-LLama-70b
\Llama-3-70B-Synthia-v3.5
\Llama-3-70B-Instruct-Gradient-524k
\Llama-3-70B-Instruct-Gradient-262k
\Tess-2.0-Llama-3-70B-v0.2
\Llama-3-Lumimaid-70B-v0.1-alt

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: \Llama-3-70B-Instruct-Gradient-262k
    parameters:
      weight: 0.25
      density: 0.90
      gamma: 0.01
  - model: \Meta-LLama-3-Cat-Smaug-LLama-70b
    parameters:
      weight: 0.28
      density: 0.90
      gamma: 0.01
  - model: \Llama-3-Lumimaid-70B-v0.1-alt
    parameters:
      weight: 0.15
      density: 0.90
      gamma: 0.01
  - model: \Tess-2.0-Llama-3-70B-v0.2
    parameters:
      weight: 0.06
      density: 0.90
      gamma: 0.01
  - model: \Smaug-Llama-3-70B-Instruct
    parameters:
      weight: 0.04
      density: 0.90
      gamma: 0.01
  - model: \Llama-3-70B-Synthia-v3.5
    parameters:
      weight: 0.05
      density: 0.90
      gamma: 0.01
  - model: \Llama-3-70B-Instruct-Gradient-524k
    parameters:
      weight: 0.03
      density: 0.90
      gamma: 0.01
  - model: \Meta-LLama-3-Cat-A-LLama-70b
    parameters:
      weight: 0.14
      density: 0.90
      gamma: 0.01
merge_method: breadcrumbs_ties
base_model: I:\Llama-3-70B-Instruct-Gradient-262k
dtype: float16

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