Skip to content

A Guide to Trelis Resources

Trelis is an Irish company based in Dublin, run by Ronan McGovern.

Trelis provides tutorials and scripts to accelerate the fine-tuning and deployment of AI Models (language, audio and vision).


Free Resources

The Trelis YouTube Channel is the primary source of free content, covering AI model explainers, videos on fine-tuning and on deploying production models. See the Trelis YouTube channel here and/or check out the following playlists:

Trelis One-Click-LLMs is a GitHub repo for deploying language models on Runpod, VastAI or on your own computer. See here.

The Trelis Blog. Roughly weekly updates of new video content. You can subscribe here if you’d like.

Trelis on X. Formerly Twitter, this platform is probably the best source of the latest news on AI models, some of which Trelis re-tweets. All Trelis videos are also posted to X. Follow along here.

Paid Repositories

All paid repositories include life-time access and allow members to post issues related to scripts within. Members gain free access to any future content added after their purchase.

ADVANCED-fine-tuning: The most popular repo (over 200 members), covering a wide range of data preparation and fine-tuning scripts that pair with Trelis YouTube videos. Read more here.

ADVANCED-inference: The second most popular repo (100+ members) providing guidance around deploying APIs for your custom language models. Read more here.

ADVANCED-vision: A specialised repo focused on fine-tuning of vision models (multi-modal text + speech). Read more here.

ADVANCED-transcription: Another specialised repo for fine-tuning voice to text models. Read more here.

Function-calling models and datasets. Models fine-tuned for function calling, and a custom dataset for fine-tuning your own model for structured responses. Read more here.

Support and Consulting

I’ve broken this into a few areas:

  • Questions on Trelis products, post a comment on the relevant product page on Trelis.com .
  • Questions on YouTube videos or content, post on the relevant video (the same goes for videos on X).
  • Questions on HuggingFace models/datasets, create a community discussion in the relevant model/dataset on HuggingFace .
  • Questions on paid repos (post purchase), create a GitHub issue for any questions on the repo content.

For any other queries (i.e. not related to videos or products), there are two further options:

  1. Trelis Discord – accessible via a monthly subscription. This is intentionally paid and structured to ensure that those who subscribe get help in a timely fashion. I also post, weekly, some of my reading notes on new models and techniques. Feel free to just subscribe for a month and then unsubscribe if you only briefly need assistance. Read more here.
  2. Consulting Review Sessions. I have a small number of slots each month for short review sessions (often providing feedback on fine-tuning and inference approaches). Most devs can save the expense of consulting reviews by making use of the free videos, life access to relevant repositories and Discord support. Read more about review sessions here.

Trelis Limited, Blackrock, Dublin.

VAT: IE3808855SH

Company Registration: 705524