FreeLLMNVIDIA releases open‑source Attribution tool for diagnosing distributed AI failures
The new NVIDIA Attribution system correlates logs, metrics and scheduler data to pinpoint root causes in multi‑GPU workloads.
NVIDIA announced an open‑source diagnostic suite called NVIDIA Attribution on 2 Nov 2026.
The system consists of an Attribution Service exposing a REST API and a Python‑based Attribution Blueprint that links cluster schedulers and telemetry sources to the service.
Connectors are provided for Slurm, Kueue, static inputs, Loki logs, local files, Kubernetes pod logs and PromQL metrics.
A worked example shows the tool isolating an NVLink signal‑integrity issue as the root cause while preserving downstream NCCL errors as symptoms.
What it means for you
For AI operators, the tool offers a reproducible, workload‑centric view of failures, helping reduce downtime and target hardware repairs. Free‑or‑local model users gain a template for building similar diagnostics without relying on proprietary monitoring stacks.
Sources
- Diagnose Distributed AI Workload Failures with NVIDIA Attribution, NVIDIA Developer (official), 2 Nov 2026
Written automatically from the sources above, then checked by our code: every number and name must appear in them, and no sentence may copy a source. Spotted a mistake? Report it.
Which AI can you use for free today?
Every free AI model, ranked by blind human votes, with the memory and speed it needs on your own GPU.