
Jetson LLM benchmark
SkillBenchmarks LLM and VLM serving speed on Jetson devices with structured JSON results
- Open source
- 4K GitHub stars
- AI models and LLM tooling


Benchmarks LLM and VLM serving speed on Jetson devices with structured JSON results

Guides an agent through iterative profiling and tuning of cuTile GPU kernels in TileGym

Trains, distills, quantizes and runs RT-DETR real-time object detection models

Trains, evaluates and exports PointPillars 3D LiDAR object detection models with TAO

Fine-tunes and runs NVIDIA CLIP models for image-text retrieval and zero-shot

Fine-tunes HuggingFace vision, VLM and LLM models in an NVIDIA GPU container

Guides an agent in converting cuTile GPU kernels to Triton and debugging the result

Guides agents through diagnosing GKE TPU multi-slice training hangs using MXLA analyzer

Guides an agent to swap TileGym cuTile kernels into Hugging Face transformers models

Recommends an LLM serving runtime and memory flags for NVIDIA Jetson devices

Picks Jetson-compatible containers and PyPI indexes for GPU ML on Orin and Thor

Guides an agent to write, debug and validate cuTile GPU kernels in Python
The closest alternatives to Litert google tensor are Jetson LLM benchmark, Tilegym improve cutile kernel perf, Tao train rtdetr, Tao train pointpillars and Tao finetune clip.
Yes. Jetson LLM benchmark, Tilegym improve cutile kernel perf, Tao train rtdetr have a free or open-source option.