Results for “megatron-bridge”

11 skills

nemo-mbridge-perf-cuda-graphs

Validate and use CUDA graph capture in Megatron Bridge, including local full-iteration graphs and Transformer Engine scoped graphs for attention, MLP, and MoE modules.

More results

training-llms-megatron

Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies. Use when training models >1B parameters, need maximum GPU efficiency (47% MFU on H100), or require tensor/pipeline/sequence/context/expert parallelism. Production-ready framework used for Nemotron, LLaMA, DeepSeek.

mcore-run-on-slurm

Launch distributed Megatron-LM training jobs on a SLURM cluster with a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules, container conventions, monitoring, and per-rank failure diagnosis.

nemo-mbridge-multi-node-slurm

Convert single-node PyTorch distributed scripts into multi-node Slurm sbatch jobs and debug common multi-node failures, covering srun-native and torch.distributed approaches, container setup, NCCL timeouts, and interactive allocation.

training-llms-megatron

Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies. Use when training models >1B parameters, need maximum GPU efficiency (47% MFU on H100), or require tensor/pipeline/sequence/context/expert parallelism. Production-ready framework used for Nemotron, LLaMA, DeepSeek.

orchestra-research

training-llms-megatron

Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies for maximum GPU efficiency.

10.4k · bundle

Selects Jetson-compatible containers, vLLM runtime images, and Jetson AI Lab PyPI indexes based on Orin SM 8.7 vs Thor SM 11.0 and JetPack version.

jetson-customize-mgbe

Generates kernel-DT overlay fragments to enable 25G/10G/1G MGBE QSFP interfaces on Jetson Thor, verifying pinmux and integrating with the BSP customization workflow.

Data pipeline specialist for ML experiments. Use when the user asks to talk to Gekko, requests the data engineer, or needs DataLoader optimization.

auditing-foundry-smart-contract-security

Runs a pre-deployment security audit of Solidity smart contracts in a Foundry project, combining static analysis (Slither, Aderyn), symbolic execution (Mythril), and property-based testing to catch reentrancy, access-control, and arithmetic bugs before deploying to an EVM chain.

24.6k · bundle

johnalbertini14-glitch

Orchestrates VPN switching, geo-block diagnostics, and leak-aware task resumption by coordinating shell, curl, WireGuard, Tailscale, DNS, IPinfo, and MoltGuard skills.