Nvidia Patents a System That Builds Task-Specific AI Models from One Unified Network
What if you could train dozens of specialized AI models and then collapse them all into a single network — pulling out…
278 filings · topic
AI training & infrastructure patents
The unglamorous machinery underneath: training methods, inference speedups, model compression, and the scheduling tricks that make AI affordable to run.
Recent filings
278 total
Nvidia Patents a Dynamic Neural Network Workload Scheduler Across Multiple Processors
Running a giant neural network on a single chip is increasingly a bottleneck — Nvidia's new patent describes a system that watches…
Nvidia Patents a Self-Curating Synthetic Data Loop for Stereo Vision Models
Training a depth-sensing AI normally requires mountains of carefully labeled real-world images — a slow, expensive process. Nvidia's new patent describes a…
Nvidia Patents an Unsupervised AI That Finds Network Topology Anomalies
What if your network could spot its own wiring mistakes — without anyone telling it what a mistake looks like? That's the…
Amazon Patents a GPU Resource Sharing System for Multi-Project ML Workloads
Running three ML training jobs at once shouldn't mean one hogs the GPUs while the others wait. Amazon's new patent describes a…
Adobe Patents a Generative AI That Designs Content Around Performance Data
Adobe is patenting an AI model that doesn't just generate creative content — it learns to link that content directly to real-world…
Adobe Patents a System for Auto-Generating Music-AI Training Data
Teaching an AI to connect music and language is only as good as the training data you feed it — and Adobe…
Salesforce Patents a Self-Verifying Reasoning Pipeline for Vision-Language AI
Salesforce is teaching AI models to show their work — and only keep the answers where the reasoning actually checks out. The…
Samsung Patents an ML Training Method for Fixing Blurry Handheld Photos
Taking sharp photos in low light is hard — your hands shake, frames blur, and stacking multiple shots makes the misalignment worse.…
IBM Patents a Neural Network That Predicts Quantum Circuit Outputs
Running a quantum circuit is expensive and error-prone — so IBM is training a neural network to predict what a quantum computer…
Adobe Patents a Parallel Approach to Causal Linear Attention in Neural Networks
Attention mechanisms are the engine inside modern AI — but they're also the bottleneck. Adobe's new patent describes a way to run…
Salesforce Patents a User-Directed Pipeline for Building Multimodal AI Training Data
Teaching an AI to understand images is only as good as the training data you feed it — and most pipelines for…
Get the weekly Patentlyze email
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.