AIFuture
Back to news
ProductMarkTechPost·

Meet Nemotron Labs 3 Puzzle 75B A9B: A Compressed Hybrid MoE LLM Delivering 2.03x Server Throughput

Meet Nemotron Labs 3 Puzzle 75B A9B: A Compressed Hybrid MoE LLM Delivering 2.03x Server Throughput

NVIDIA has released Nemotron-Labs-3-Puzzle-75B-A9B, a compressed variant of Nemotron-3-Super. Iterative Puzzle alternates hardware-aware structural compression with short knowledge distillation recovery phases. The model drops from 120.7B total / 12.8B active parameters to 75.3B / 9.3B. On a single 8xB200 node it delivers 2.03x Super's total throughput at 100 tok/s per user. On one H100, 1M-token…

This is a summary curated by AIFuture. Read the complete article at the original source:

Read the full story on MarkTechPost

Build the skills behind the headlines

Data ScienceedX

CS50's Introduction to AI with Python

Harvard's deep dive into the algorithms behind modern AI — search, knowledge, optimization, and machine learning.

Intermediate·Free / Verified
View Course
Data ScienceCoursera

Deep Learning Specialization

Five-course series on neural networks, CNNs, sequence models, and transformers from DeepLearning.AI.

Intermediate·Subscription
View Course
Data ScienceCoursera

Machine Learning Specialization

Andrew Ng's flagship program covering supervised and unsupervised learning, neural networks, and best practices for real-world ML.

Beginner·Subscription
View Course

Never miss what matters in AI

Get the most important AI news and course picks in your inbox.