Tom Hunn's picture
1 18

Tom Hunn

thunnai
ยท

AI & ML interests

I've been working in generative AI since it was still called "Deep Learning." Most of my experience is in the audio space, but I've also worked on projects involving llms. I've had hands-on experience across the full AI product lifecycle, including: - AI Research & Model Training - MLOps & Productionizing Models - Product Strategy โ€” exploring whatโ€™s possible with emerging tech

Recent Activity

View all activity

Organizations

None yet

thunnai's activity

reacted to MoritzLaurer's post with ๐Ÿ”ฅ 3 days ago
view post
Post
1091
I've been building a small library for working with prompt templates on the HF hub: pip install prompt-templates. Motivation:

The community currently shares prompt templates in a wide variety of formats: in datasets, in model cards, as strings in .py files, as .txt/.yaml/.json/.jinja2 files etc. This makes sharing and working with prompt templates unnecessarily complicated.

Prompt templates are currently the main hyperparameter that people tune when building complex LLM systems or agents. If we don't have a common standard for sharing them, we cannot systematically test and improve our systems. After comparing different community approaches, I think that working with modular .yaml or .json files is the best approach.

The prompt-templates library :
- proposes a standard for sharing prompts (entirely locally or on the HF hub)
- provides some utilities that are interoperable with the broader ecosystem

Try it:
# !pip install prompt-templates
from prompt_templates import PromptTemplateLoader 
prompt_template = PromptTemplateLoader.from_hub(repo_id="MoritzLaurer/closed_system_prompts", filename="claude-3-5-artifacts-leak-210624.yaml")


The library is in early stages, feedback is welcome!

More details in the docs: https://github.com/MoritzLaurer/prompt_templates/
replied to freddyaboulton's post 4 days ago
view reply

Awesome work on this! Another big win for rapid prototyping with Gradio!

reacted to freddyaboulton's post with ๐Ÿš€๐Ÿ”ฅ 4 days ago
reacted to TuringsSolutions's post with ๐Ÿ‘€ 23 days ago
view post
Post
849
I created something called 'Hyperbolic Embeddings'. I literally just embed the tokens into Hyperbolic Space instead of Euclidean space. At first, this did not get me the gains I was expecting. I was a sad panda. Then I thought about it, a Hyperbolic Embedding needs a Hyperbolic Optimizer. So, instead of Adam, I used Riemannian Adam (RAdam). "Ladies and Gentlemen, We Got 'Em!"
  • 27 replies
ยท
reacted to cdminix's post with ๐Ÿ”ฅ 25 days ago
view post
Post
982
As part of some ongoing work, I'm releasing the currently biggest collection of docker containers for state-of-the-art voice cloning TTS systems.
https://github.com/ttsds/datasets

Alongside there is also a nice overview of all systems (see below)