kendex.ai

Marketplaces / wshobson/agents / dgx-spark-ops/spark-training-gotchas

dgx-spark-ops/spark-training-gotchas

Preflight and diagnose the ten known failure modes for ML training on NVIDIA DGX Spark. Use when a training run on DGX Spark fails to start, OOMs below the 128GB limit, slows down mid-run, or before any multi-hour training job on GB10.

skill · safety 100/100 (clean) · @ 367cb6a

Install in kendex: kendex add --skill dgx-spark-ops/spark-training-gotchas after subscribing to wshobson/agents.

This package carries no rendered README.