repowiserepowise

Command-line tool

arcee-ai / mergekit

Tools for merging pretrained large language models.

★ 7,485stars on GitHubPythonLGPL-3.0

This connects Repowise to arcee-ai/mergekit's code. It does not install mergekit.

Repowise maps public repos from their source. Not affiliated with arcee-ai. What is Repowise?

github.com/arcee-ai/mergekitIndexed at 0f1423e

What is mergekit?

mergekit is a toolkit for merging pre-trained language models, written in Python. It uses an out-of-core approach, so merges can run in resource-constrained setups either entirely on CPU or accelerated on a GPU. It supports a range of merge methods, such as linear averaging, SLERP, TIES and DARE, that combine the weights of several models into one. To install it, run git clone https://github.com/arcee-ai/mergekit.git and follow the steps below. Its code health is 8.6 out of 10 (Excellent), ranked #36 of 4,000+ indexed repos. The last commit to mergekit was on Oct 31, 2026, and 51 people have contributed.

As of commit 0f1423e, indexed Nov 21, 2026. What it is comes from a model reading the README; the rest is measured from the source and git history, not from the maintainer.

For agentsmergekit.md

Repo facts: type, license, health, rank, contributors
mergekit facts
TypeCommand-line tool
LicenseLGPL-3.0
Code health8.6 / 10 (Excellent)
Health rank#36 of 4,000+ indexed repos
Last commitOct 31, 2026
Contributors51
Bus factor1
Size193 files, 17,226 lines

Install mergekit

Commands copied from the README

  1. 1Clone the mergekit repository.
    git clone https://github.com/arcee-ai/mergekit.git
  2. 2Clone the mergekit repository and change into its directory.
    cd mergekit
  3. 3Install the package in editable mode so the scripts are available.
    pip install -e .

Releases on GitHub (latest v0.1.5, Oct 20, 2026)

Full install notes on GitHub

Explore mergekit

Ask about this code in your agent

This connects Repowise to mergekit's code. It does not install mergekit. To install mergekit, use the install steps.

Useful if you want to change or fork mergekit: your agent can ask how it is put together and get answers that cite its files.

How to use mergekit

  1. 1Run a merge by passing a YAML configuration file and an output directory to mergekit-yaml.
    mergekit-yaml path/to/your/config.yml ./output-model-directory [--cuda] [--lazy-unpickle] [--allow-crimes] [... other options]

Ask about mergekit

Answers come from mergekit's README and source, with the files they cite. Two free answers a day without an account.

Want this for your own codebase? Map your repo free

About this page

Get a page like this for your own codebase

Repowise built this map of arcee-ai/mergekit from its source and git history. It parses every symbol, draws the dependency graph, scores each file's health from complexity, duplication, coverage and churn, finds hotspots and owners in the history, and writes documentation for every module. Your coding agent reads the same index over MCP, so its answers cite real files.

Free for 2 public repos, with a page like this one. No card. See how Repowise works

Prefer local? pip install repowise && repowise init indexes your own checkout, no account needed. Star Repowise on GitHub

arcee-ai/mergekit FAQ

What is mergekit?

mergekit is a toolkit for merging pre-trained language models. It uses an out-of-core approach so that merges can run on CPU or on a GPU in resource-constrained setups.

How does mergekit work?

mergekit runs a merge from a YAML configuration file and an output path using mergekit-yaml, then writes the merged model to the output directory. Tensors are loaded lazily to keep memory use low.

How do I install mergekit?

Clone the mergekit repository, change into the mergekit directory, and run pip install -e . to install the package in editable mode and make its scripts available. See the install steps

What language is mergekit written in?

mergekit is written in Python, and its packaging and dependencies are declared in pyproject.toml.

How is the mergekit code organized?

The package lives in the mergekit directory, command-line scripts are under mergekit/scripts, and tests are in the tests directory.

When should I use mergekit?

mergekit is for merging pre-trained language models, such as combining specialized models into a single model or extracting LoRA from a fine-tuned model.

What license does mergekit use?

mergekit uses the LGPL-3.0 license.

Code health

8.5out of 10Excellent

On the leaderboards:#4 of 71 Python repos

Key features of mergekit

  • Supports merging of models such as Llama, Mistral, GPT-NeoX and StableLM.
  • Runs on GPU or CPU and loads tensors lazily to keep memory use low.
  • Provides a range of merge methods, including linear, SLERP, TIES, DARE and Mixture of Experts merging.
  • Extracts LoRA from fine-tuned models.
  • Offers evolutionary merge methods and multi-stage merging for complex workflows.

How arcee-ai/mergekit works

The main command-line entry point is mergekit-yaml, which takes a YAML configuration file and an output path and writes the merged model to that path. The mergekit directory holds the package, and mergekit/__init__.py marks it as the package entry point. Additional command-line scripts live in mergekit/scripts, including mergekit/scripts/extract_lora.py for LoRA extraction and mergekit/scripts/evolve.py for evolutionary merges. Merge methods can also be called from Python through mergekit.merge_methods, using merge_tensors for a single tensor or merge_state_dicts for models already in memory. The tests directory contains the project's test files, and pyproject.toml declares the dependencies, including safetensors, huggingface_hub and ray.

The codebase has 193 files and 988 symbols in 2 modules, written in Python. Code health is 8.5 out of 10, rated excellent.