
Guides an agent through multi-node Slurm sbatch jobs and debugging their failures
- Open source
- 4K GitHub stars
- Developer tools


Guides an agent through multi-node Slurm sbatch jobs and debugging their failures

Runs TAO GPU jobs on a remote SLURM cluster over SSH with sbatch and Pyxis containers

Guides validating and configuring CUDA graph capture in Megatron Bridge training runs

Practitioner runbook for training and serving large models, from GPUs to NaN loss spikes

Configures fault tolerance, straggler detection and restarts for Megatron Bridge training

Guides an agent through evidence-gated MoE performance tuning in Megatron Bridge

Guides an agent through reducing peak GPU memory in Megatron Bridge training runs

Sets up the execution context for NVIDIA FastGen commands: paths, config overrides

Guides an agent through GPU DataFrame work with NVIDIA cuDF and dask-cuDF

Lets an agent submit, monitor and cancel TAO GPU training jobs on Kubernetes clusters

Guides integrating PhysicsNeMo ShardTensor domain parallelism into multi-GPU training

Guides writing and debugging GPU-accelerated RDKit cheminformatics code with nvMolKit
The closest alternatives to Mcore run on slurm are Nemo mbridge multi node slurm, Tao run on slurm, Nemo mbridge perf cuda graphs, ML engineering and Nemo mbridge resiliency.
Yes. Nemo mbridge multi node slurm, Tao run on slurm, Nemo mbridge perf cuda graphs have a free or open-source option.