Ludwig 10k Giveaway Overview
In celebration of our 10,000th star on Github, we're giving away a premium Ludwig swag pack to 5 avid Ludwig users. The swag pack includes a premium limited edition Ludwig hoodie, sticker set, and Yeti mug. To participate in the contest, share a cool use case or project that you developed with Ludwig.
Submit your entry on or before January 7, 2024 by filling out this short form. You are welcome to work in teams (the maximum team size is 4 participants). We have set up the channel “ludwig-10k-contest” in Ludwig Slack – you are welcome to look for project partners in that channel or invite friends/colleagues, etc.
Competition Participation Rules
- Focus on fine-tuning and/or use of LLMs. Suggested topics:
Generating domain-specific code or language (e.g. generating JSON) | Creating marketing content like a press release or customer email | Summarizing content like legal documents or contracts |
Learning tools such as a tutor fine-tuned on a topic like math, physics, etc. | Monitoring system logs and identifying trends or issues | Automating paper-heavy processes like medical claims approvals |
Classification like determining intent and urgency from support call transcript | Determining patient condition from unstructured medical records | ...and any other cool use case that you can think up! |
- You must use Google Colab notebooks for your project. Since we will need to be able to run them upon your submission, please make sure that we have the read access rights and can clearly identify your CPU/GPU requirements (ideally, just a free Colab account).
- We will select 5 of our favorite projects and the winners will receive a premium swag box of Ludwig goodies in the mail. We will also author a blog post highlighting the winners and their projects. We may even have some opportunities for you to share your work at upcoming meetups.
- Winners will be announced on December 14th, 2023.
Installation and Reference Materials
The following links will help you install Ludwig in your development environment and get up to speed with using it for a number of practical scenarios. We will be monitoring the “ludwig-10k-contest” channel in Ludwig Slack to ensure that you get the assistance you need.
Ludwig Introductory Materials
- Main Ludwig site: https://ludwig.ai/
- Ludwig Repository: https://github.com/ludwig-ai/ludwig
- Example Use Cases (expand the “Use Cases” arrow on the left): https://ludwig.ai/latest/examples/
Installation
Since you will be working in Google Colab notebooks, please execute:
!pip uninstall -y tensorflow --quiet
!pip install ludwig --quiet
!pip install ludwig[llm] --quietin a cell of your notebook to install Ludwig.
Some of the guides for different use cases may recommend additional libraries to be installed.
If for some reason your project requires modifying the Ludwig Open Source library itself, then please see the Contributor Guide in the OSS repository (the one on ludwig.ai is outdated): https://github.com/ludwig-ai/ludwig/blob/master/CONTRIBUTING.md.
We recommend that for working with Ludwig on your local machine, you use the command pip install -e ."[full]" as it will install all libraries that you would potentially use and be able to run tests.
How-To Guides
Fine-tuning LLMs (some of these resources contain ready-to-go Google Colab notebooks):
- DeepLearning.AI Webinar: Efficient Fine-Tuning for Llama-v2-7b on a Single GPU with Ludwig
- Ludwig 0.8 Overview Blog: Efficient Fine-tuning
- Ludwig Tutorial Blog: Fine-tuning Mistral 7b
- Ludwig Use Case Blog: Named Entity Recognition for Molecular Biology
Google Colab notebooks containing practical examples:
You may also find lots of other useful tutorials and guides at the following links:
- Main Ludwig site: https://ludwig.ai/
- Ludwig LLM Documentation: https://ludwig.ai/latest/configuration/large_language_model/
- Ludwig Repository: https://github.com/ludwig-ai/ludwig
- Predibase Blog: https://predibase.com/blog
Help us celebrate all of the awesome work that you're doing by sharing your progress in the “ludwig-10k-contest” channel as you are working through it!
