Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912977281613824 |
|---|---|
| author | Yang, Zhuolin Liu, Zihan Chen, Yang Dai, Wenliang Wang, Boxin Lin, Sheng-Chieh Lee, Chankyu Chen, Yangyi Jiang, Dongfu He, Jiafan Pi, Renjie Lam, Grace Lee, Nayeon Bukharin, Alexander Shoeybi, Mohammad Catanzaro, Bryan Ping, Wei |
| author_facet | Yang, Zhuolin Liu, Zihan Chen, Yang Dai, Wenliang Wang, Boxin Lin, Sheng-Chieh Lee, Chankyu Chen, Yangyi Jiang, Dongfu He, Jiafan Pi, Renjie Lam, Grace Lee, Nayeon Bukharin, Alexander Shoeybi, Mohammad Catanzaro, Bryan Ping, Wei |
| contents | We introduce Nemotron-Cascade 2, an open 30B MoE model with 3B activated parameters that delivers best-in-class reasoning and strong agentic capabilities. Despite its compact size, its mathematical and coding reasoning performance approaches that of frontier open models. It is the second open-weight LLM, after DeepSeekV3.2-Speciale-671B-A37B, to achieve Gold Medal-level performance in the 2025 International Mathematical Olympiad (IMO), the International Olympiad in Informatics (IOI), and the ICPC World Finals, demonstrating remarkably high intelligence density with 20x fewer parameters. In contrast to Nemotron-Cascade 1, the key technical advancements are as follows. After SFT on a meticulously curated dataset, we substantially expand Cascade RL to cover a much broader spectrum of reasoning and agentic domains. Furthermore, we introduce multi-domain on-policy distillation from the strongest intermediate teacher models for each domain throughout the Cascade RL process, allowing us to efficiently recover benchmark regressions and sustain strong performance gains along the way. We release the collection of model checkpoint and training data. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_19220 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation Yang, Zhuolin Liu, Zihan Chen, Yang Dai, Wenliang Wang, Boxin Lin, Sheng-Chieh Lee, Chankyu Chen, Yangyi Jiang, Dongfu He, Jiafan Pi, Renjie Lam, Grace Lee, Nayeon Bukharin, Alexander Shoeybi, Mohammad Catanzaro, Bryan Ping, Wei Computation and Language Artificial Intelligence Machine Learning We introduce Nemotron-Cascade 2, an open 30B MoE model with 3B activated parameters that delivers best-in-class reasoning and strong agentic capabilities. Despite its compact size, its mathematical and coding reasoning performance approaches that of frontier open models. It is the second open-weight LLM, after DeepSeekV3.2-Speciale-671B-A37B, to achieve Gold Medal-level performance in the 2025 International Mathematical Olympiad (IMO), the International Olympiad in Informatics (IOI), and the ICPC World Finals, demonstrating remarkably high intelligence density with 20x fewer parameters. In contrast to Nemotron-Cascade 1, the key technical advancements are as follows. After SFT on a meticulously curated dataset, we substantially expand Cascade RL to cover a much broader spectrum of reasoning and agentic domains. Furthermore, we introduce multi-domain on-policy distillation from the strongest intermediate teacher models for each domain throughout the Cascade RL process, allowing us to efficiently recover benchmark regressions and sustain strong performance gains along the way. We release the collection of model checkpoint and training data. |
| title | Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2603.19220 |