Honrado de participar da minha primeira banca de avaliação de qualificação de doutorado, do aluno Diego Corrêa Silva na UFBA. A tese é focada em Calibração de Sistemas de Recomendação. Parabéns pelo trabalho Diego, que já está bastante avançado em termos…lnkd.in/dSMUyjee

In #recsys you might want to predict different types of user events (click, like, share, purchase). In this post we introduce multi-task learning (MTL) for ranking models and how to easily build and train such models using #Merlin and #tensorflow. medium.com/nvidia-merlin/…

Glad to be part again of the organizing committee for EvalRS workshop. Our new home is KDD'23. This year we came up with a new format, which includes a hackaton on #recsys evaluation with monetary prizes, as well as the usual paper session. Check it out! lnkd.in/dehy47ZV

How do you know your #recsys works as expected? After the success of #EvalRS22, we are back at SIGKDD 2025! With prizes for papers, students and projects, #EvalRS is a workshop with a huge social component: come for the papers, stay for the pizza! CFP: reclist.io/kdd2023-cup/

#BardByGoogle is impressive!

⏰Still 9 days to submit to #EvalRS2023 on the well-rounded evaluation of #RecSys at #KDD. Covering fairness, bias, robustness, and trustworthiness of #RecSys. Workshop + hackathon! reclist.io/kdd2023-cup/

Here are 8 key considerations for building *production-grade* LLM apps over your data (RAG) 💡 (see 🧵): 1️⃣ Chunks used for retrieval shouldn’t necessarily be the same as chunks used for LLM synthesis (@md_rumpf) 2️⃣ Embeddings should live in a different latent space than what

LLMs were definitely the hot topic at #KDD23. I had the opportunity to attend that conference for the first time a few weeks ago and just published a post with my summary and curated resources. Hope you get some interesting insights from my highlights! gspmoreira.medium.com/highlights-on-…

Excited to have my recent post about KDD 2023 and LLMs republished by Towards Data Science

The open-source models were starting to be too good for the Hugging Face open LLM leaderboards so we added 3 new metrics thanks to EleutherAI to make them harder and more relevant for real-life performance. Re-running evaluation of 2,000+ models wasn't easy and took more than

What if you could use #generativeai to compose a storytelling about moments you have captured with your camera? I just open-sourced Photo StoryTelling, which converts your photos into a blog post using Google Cloud APIs. Warning: Results are impressive. gspmoreira.medium.com/photo-storytel…

I shared the Photo Storytelling example last month using #google Imagen for image captioning and #palm2 for blog post generation. I just tried the new multi-modal prompt from #vertexai with #gemini and it does a decent job for that same task 😆 medium.com/@gspmoreira/ph…

Excited for the strong partnership between NVIDIA and Google ! Doing my first experiments with #Gemma 7B and it looks great!

I was invited to participate of an episode about #LLMs at the CI&T Humans of Digital #podcast, and had a nice chat with Mikaeri Ohana about it. You can check the episode here youtube.com/watch?v=k9lRsv…

Will present tomorrow our paper about #Nemo Retriever Ranking models at ACM CIKM 2024 RAG for Enterprises workshop. Join us if you are at #cikm2024

Our new Omni-Embed-Nemotron cross-modal retrieval model can search text, images, audio, or video using a single bi-encoder architecture that process all modalities in a shared embedding space for simpler, faster retrieval-- all in one open-source architecture 🛠️🌐 Try out the

📢 ViDoRe V3, our new multimodal retrieval benchmark for enterprise use cases, is finally here! It focuses on real-world applied RAG scenarios using high-quality human-verified data. huggingface.co/blog/QuentinJG… 🧵(1/N)