Welcome to the second edition of Hugging Fellows Highlights! In this page, we will go through what our Fellows accomplished past couple of weeks.
<aside> 🤗 You can learn more about Fellows program here.
</aside>
🧑🎤 Thomas Chaigneau has open-sourced a project called PicAIsso, it’s a generative art app with API endpoint and discord bot. He’s also working on this project called AI Hardware Advisor (AIHA), to predict hardware requirements of a given Hugging Face model.
He also fine-tuned LLaMa-7b and LLaMa-13b on the natural-instructions dataset using PEFT 🌟
🦹 Simon Dürr has built two new Gradio application, one for editing molecules, and he trained and built an app for a new diffusion model for protein sequence and structure.
👨🎨 Johannes Kolbe has led the computer vision study group session on BLIP-2. You can watch the recording here. He built this demo on Spectral Clustering and another demo on vector quantization using scikit-learn.
👨🔬 Aritra has created this tutorial on temporal latent bottleneck.
👩🎤 María organized an ML hackathon in Spanish, see here.
👨🎤 Manu released two new models 👇
👨🔬 Sylvain has built a fun project that generates bedtime stories, app can be found here.
🦸 Chansung has fine-tuned LLaMa on ALPACA, ALPACA-GPT4 and EvolInstruct, see the repositories here.
He added more models to LLM as Chatbot and different styles to PingPong project. He started LLMPool to create an LLM pool in the mixture of locally and remotely hosted models, then call stream_gen and non_stream_gen in a uniform way. He also wrote a blog post on Serving with Serving With TF and GKE: Stable Diffusion.
👨🏻🎤 Bram has trained Dutch T5-UL2 model, you can find models, datasets and checkpoints below 👇🏼