Bo

Deep Learning at NVIDIA

[NVIDIA AI @NVIDIAAI]

Today we released Nemotron 3 Embed 8B and it reached #1 overall on RTEB 🏆

RTEB benchmarks retrieval accuracy across real-world tasks. Better retrieval gives agents more relevant context, helping improve response accuracy.

[NVIDIA AI Developer @NVIDIAAIDev]

Nemotron just landed on the ViDoRe V3 leaderboard — testing RAG capabilities across visuals, documents, and reasoning tasks.

Nemotron-colembed-8B-V2 pushes ahead with sharper visual comprehension for complex charts, better multilingual reading, and improved accuracy through model merging.

Model: https://t.co/lL2ZzBmujR
HF Blog: https://t.co/NVPVQ9Utja

[JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱 @JFPuget]

My team a NVIDIA is hiring, see the job req at https://t.co/KH1HX2I4CT

This says US, but we can hire in any country where NVIDIA has a legal entity, see the list at the bottom of page: https://t.co/hBAJVdVWn2

[NVIDIA AI Developer @NVIDIAAIDev]

Congratulations to our NVIDIA Kaggle Grandmasters on their winning solution in the ARC Prize 2025 competition 🎉

The team fine-tuned a compact model that out-reasoned massive systems — proving synthetic data + adaptive reinforcement learning can punch far above their weight.

🔗 Details: https://t.co/0xeMlXvtu6

[António Loison @antonio_loison]

📢 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. https://t.co/Fs6W2gNQJc
🧵(1/N)

[merve @mervenoyann]

people are sleeping on this release

NVIDIA blessed us with a new family of Nemotron RAG models 🔥

it comes with text retrievers, multimodal retrievers as well as layout detectors with commercially permissive license 👏

[Bo @boliu0]

Incredibly proud of @ybabakhin @radekosmulski @ak_ronay and the entire team for this huge achievement! Topping the MTEB leaderboard as an open model is a massive and well-deserved accomplishment. Congratulations!

[NVIDIA AI Developer @NVIDIAAIDev]

The Llama-Embed-Nemotron-8B text embedding model now ranks as the top open and portable model on the @huggingface MMTEB leaderboard. 🏆 https://t.co/4joyNRV0Ln

Why it’s leading the race:

✅ Strong performance for retrieval, reranking, classification, and bi‑text mining
✅ Effectively handles retrieval in low-resource languages and cross-lingual setups
✅ Trained on 16M query‑document pairs (8M public + 8M synthetic)
✅ Combines model merging and refined hard‑negative mining for better accuracy

Leaderboard 🏆 https://t.co/4joyNRV0Ln
Download 📥 https://t.co/nqdMAOb5Ii

[NVIDIA AI Developer @NVIDIAAIDev]

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 code yourself ➡️ https://t.co/X1JkRIc4Xp

[JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱 @JFPuget]

NVIDIA is hiring! You'd work with Kaggle GMs like Bo Liu and Yauhen Babakin. Kaggle GMs welcomed, but this is not a mandatory qualification.

Job post details: https://t.co/qf8lhbpS1Z

[NVIDIA AI Developer @NVIDIAAIDev]

👀 Just released - 4 new classifier models for curating high-quality data with NVIDIA NeMo Curator ✨

Read blog ➡️ https://t.co/owiYcXRv0e

Categorize your data into predefined groups or classes, annotate it with metadata, and optimize your #dataprocessing pipelines. #LLM

[NVIDIA AI Developer @NVIDIAAIDev]

Now on @huggingface: Classify docs into 26 domains with our advanced text classification model. Accelerated using intelligent batching + RAPIDS for faster inference. Trained on 1M Common Crawl samples and 500k Wikipedia articles. Download ➡️ https://t.co/jauvrfFnMG ✨

[Bo @boliu0]

@Giba1 First human in history to get 60 Kaggle competition gold medals! 🫡

[Bo @boliu0]

Thanks for hosting me on Chai Time Data Science Show Sanyam. I got so excited that I brought a two-liter Chai to the show!

[Sanyam Bhutani @bhutanisanyam1]

I'm super excited to be back at the Chai office and host more interviews 🙏

Today, @boliu0 from the @NVIDIAAI Grandmaster team will be teaching us about their 4th pos Happywhale competition solution

https://t.co/jPVIg8K18L

[NVIDIA AI @NVIDIAAI]

Have you ever wondered what happens to Kaggle winning solutions after the competitions? Join us at #GTC and find out how Kaggle Grandmasters of NVIDIA (KGMON) team reuses their competition-winning models and techniques for real-world problems:https://t.co/xUq0pbIuSP #GTC22

[ACM RecSys @ACMRecSys]

Here are the winners of the general leaderboard of the RecSys Challenge!

The winners decided to donate their prize to the Women in RecSys Community 😍 #RecSys2021

Thank you @walteranelli for being the host tonight of the RecSys Challenge Awards!

[NVIDIA AI @NVIDIAAI]

#NVIDIA's @Kaggle Grandmaster team provides an insight on what it involves being part of Kaggle competition and how to become Kaggle Grandmasters: https://t.co/GUhpSYuInH

[Jarrel Seah @jarrelscy]

#RANZCRASM21 Tune in to AI Session 4 to hear Kaggle Grandmaster Bo Liu @boliu0 talk about his incredible first-place winning approach to the inaugural RANZCR CLiP Challenge!

[Jennifer Tang @DrJenTang]

#RANZCRASM21 AI Session 4 will include RANZCR 2021 CLiP Kaggle Challenge with international speakers including Kaggle Challenge winners Bo Liu (@boliu0) and Hiroshi Yoshihara (@analokmaus), Maggie Demkin (@MDemkin) from Kaggle and A/Prof George Shih from https://t.co/Dgi5QWPdoW

[Bo @boliu0]

@bhutanisanyam1 @wandb Congrats Sanyam!