Back to News Feed
Nvidia Developer Blog
November 19, 2026
neutral
Reducing Rollout Bottlenecks in Reinforcement Learning with NVIDIA Dynamo
NVIDIAAI / LLMSemiconductorsNVIDIA / GPU
<img width="768" height="432" src="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-768x432.png" class="webfeedsFeaturedVisual wp-post-image" alt="" style="display: block; margin-bottom: 5px; clear:both;max-width: 100%;" link_thumbnail="" decoding="async" srcset="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-768x432.png 768w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-179x101.png 179w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-300x169.png 300w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-625x352.png 625w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-645x363.png 645w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-660x370.png 660w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-500x281.png 500w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-160x90.png 160w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-362x204.png 362w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-196x110.png 196w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-1024x576.png 1024w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-960x540.png 960w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17.webp 1536w" sizes="(max-width: 768px) 100vw, 768px" title="image3" />Reinforcement learning (RL) uses rewards to train models to solve problems and complete tasks more successfully. Inference drives this process. Workers generate...<img width="768" height="432" src="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-768x432.png" class="webfeedsFeaturedVisual wp-post-image" alt="" style="display: block; margin-bottom: 5px; clear:both;max-width: 100%;" link_thumbnail="" decoding="async" loading="lazy" srcset="https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-768x432.png 768w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-179x101.png 179w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-300x169.png 300w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-625x352.png 625w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-645x363.png 645w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-660x370.png 660w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-500x281.png 500w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-160x90.png 160w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-362x204.png 362w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-196x110.png 196w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-1024x576.png 1024w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17-960x540.png 960w, https://developer-blogs.nvidia.com/wp-content/uploads/2026/11/image3-17.webp 1536w" sizes="auto, (max-width: 768px) 100vw, 768px" title="image3" /><p>Reinforcement learning (RL) uses rewards to train models to solve problems and complete tasks more successfully. Inference drives this process. Workers generate rollouts: sequences of model outputs, actions, and environment interactions. The RL framework scores the results, updates model weights, and sends those weights back to the workers for the next round. As RL training scales…</p>
<p><a href="https://developer.nvidia.com/blog/reducing-rollout-bottlenecks-in-reinforcement-learning-with-nvidia-dynamo/" rel="nofollow" data-wpel-link="internal" target="_self">Source</a></p>
Read original article ↗