ai-agent

llm-foundry

Name: llm-foundry
Author: mosaicml

LLM training code for Databricks foundation models

4,460 stars+5★ this week586 forksPythonUpdated 3/25/2026100% free · open source

What it does

LLM-Foundry is a Python library for training and fine-tuning large language models on Databricks, allowing founders to build custom AI models for their startups

When to use it

•When you need to train a custom language model for a specific industry or domain
•When you want to fine-tune a pre-trained language model for your startup's use case
•When you require a scalable and efficient way to train large language models on a cloud-based platform like Databricks

Ready-to-paste prompt

python train.py --model-name transformer-xl --dataset-name wiki_text --batch-size 16 --learning-rate 1e-4 --num-workers 4

Heads up: Make sure you have a Databricks account and a cluster set up with the necessary dependencies installed, including the `databricks` library, before attempting to use LLM-Foundry

Saves to your device

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

✅ Light setup

Installs with a command or two; your AI agent can do it for you.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/llm-foundry && curl -fsSL https://workflowstacks.com/api/skills/llm-foundry/claude-skill -o ~/.claude/skills/llm-foundry/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

How llm-foundry works

Codeflow

Free to inspect

Llm-foundry is a large Python project (~36k lines across 145 code files, plus 101 test files). Setup is light: installs like a normal app. Reading the code is optional. Last commit 6 months ago, Apache-2.0 license, has a test suite.

Size

Large codebase

~36k lines · 145 code files · ~5 h to skim

Setup

One-command install

Installs like a normal app. Reading the code is optional.

Runs on

Python · Docker

No API keys detected

Python 98%Shell 2%

Where to start reading

1
README.md
Start here — what it does and how to install it
2
pyproject.toml
Dependencies and the commands it exposes
3
llmfoundry/__init__.py
Inside llmfoundry/ — the main logic begins here

What's in each folder

llmfoundry/Core code — the actual logic119 files

scripts/Helper scripts172 files

mcli/Folder12 files

tests/Tests — proof it works112 files

.github/CI / automation (GitHub Actions)4 files

assets/Images & static assets2 files

.pre-commit/Tool / agent settings1 files

READMEHas testsDocumentedNo CIDocker readyApache-2.0 licenseLast update 6 mo ago

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

✅ Light setup

Installs with a command or two; your AI agent can do it for you.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/llm-foundry && curl -fsSL https://workflowstacks.com/api/skills/llm-foundry/claude-skill -o ~/.claude/skills/llm-foundry/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

Quick Actions

Details

Creator

mosaicml

Language

Python