nvidia/skills 已索引

源文件证据覆盖

已存储的包版本在这个仓库中引用的固定源文件的原始计数。它反映已记录了什么，不代表仓库有多完整。

引用了 395 个固定源文件

• 395 已知
• 0 提交中不存在
• 0 文件列表截断
• 0 不受支持
• 0 尚无记录
• 0 读取失败

按仓库、提交和路径去重；同一路径出现在两个提交中，按两个文件身份计数。不是组件组数。

读取失败的提交 0

尚无记录的文件没有对应的提交读取失败记录。

关联的包 394

按仓库的已验证身份关联。数量是历史登记记录，不是当前可用性。

第 201–220 条，共 394 条结果

• nemo-mbridge-multi-node-slurm技能

  Convert single-node scripts to multi-node Slurm sbatch jobs and debug common multi-node failures. Covers srun-native vs uv run torch.distributed approaches, container setup, NCCL timeouts, OOM sizing for MoE models, and interactive allocat…

  392ab8ab2587

• nemo-mbridge-perf-activation-recompute技能

  Validate and use selective and full activation recompute in Megatron Bridge to reduce GPU memory usage at the cost of extra compute. Use for activation memory OOMs or regressions involving recompute_granularity, recompute_num_layers, recom…

  392ab8ab2587

• nemo-mbridge-perf-cpu-offloading技能

  Validate and use CPU offloading in Megatron Bridge, including layer-level activation offloading and fractional optimizer state offloading with HybridDeviceOptimizer.

  392ab8ab2587

• nemo-mbridge-perf-cuda-graphs技能

  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.

  392ab8ab2587

• nemo-mbridge-perf-expert-parallel-overlap技能

  Validate and use MoE expert-parallel communication overlap in Megatron-Bridge, including overlap_moe_expert_parallel_comm, delay_wgrad_compute, and flex dispatcher backends such as DeepEP and HybridEP.

  392ab8ab2587

• nemo-mbridge-perf-hierarchical-context-parallel技能

  Operational guide for enabling hierarchical context parallelism in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.

  392ab8ab2587

• nemo-mbridge-perf-megatron-fsdp技能

  Operational guide for enabling Megatron FSDP in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.

  392ab8ab2587

• nemo-mbridge-perf-memory-tuning技能

  Techniques for reducing peak GPU memory in Megatron Bridge — expandable segments, PEFT + SP input re-gather, parallelism resizing, activation recompute, CPU offloading constraints, and common OOM fixes.

  392ab8ab2587

• nemo-mbridge-perf-moe-comm-overlap技能

  MoE expert-parallel communication overlap in Megatron Bridge. Covers dispatch/combine overlap, flex dispatcher backends, and expert wgrad scheduling.

  392ab8ab2587

• nemo-mbridge-perf-moe-dispatcher-selection技能

  Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.

  392ab8ab2587

• nemo-mbridge-perf-moe-hardware-configs技能

  Representative, point-in-time MoE training playbooks by hardware and model family. Use them as candidate seeds, then revalidate the exact runtime, semantics, topology, and steady-state throughput.

  392ab8ab2587

• nemo-mbridge-perf-moe-long-context技能

  Long-context MoE training guidance for Megatron Bridge. Covers CP sizing, selective recompute, dispatcher choices, and practical patterns from DSV3, Qwen3, and Qwen3-Next long-context experiments.

  392ab8ab2587

• nemo-mbridge-perf-moe-optimization-workflow技能

  Evidence-gated workflow for MoE performance optimization in Megatron Bridge. Covers measurement contracts, the Three Walls framework, parallel folding, profiling, matched A/B tuning, and final validation.

  392ab8ab2587

• nemo-mbridge-perf-moe-vlm-training技能

  Practical guidance for training MoE VLMs in Megatron Bridge. Compares FSDP and 3D-parallel approaches, using rounded lessons from Qwen3-VL, Qwen3-Next, and other multimodal experiments.

  392ab8ab2587

• nemo-mbridge-perf-parallelism-strategies技能

  Operational guide for choosing and combining parallelism strategies in Megatron Bridge, including sizing rules, hardware topology mapping, and combined parallelism configuration.

  392ab8ab2587

• nemo-mbridge-perf-sequence-packing技能

  Validate and use packed sequences and long-context training in Megatron-Bridge, including offline LLM packing, collate-time VLM packing, Energon online packing, and CP constraints.

  392ab8ab2587

• nemo-mbridge-perf-tp-dp-comm-overlap技能

  Operational guide for enabling TP, DP, and PP communication overlap in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.

  392ab8ab2587

• nemo-mbridge-recipe-recommender技能

  Recommend and customize Megatron Bridge library and benchmark recipes for a user's model, GPU count, hardware, sequence length, and pretrain/SFT/PEFT goal. Use when selecting a starting recipe, comparing library and benchmark configs, resi…

  392ab8ab2587

• nemo-mbridge-resiliency技能

  Resiliency features in Megatron Bridge including fault tolerance, straggler detection, in-process restart, preemption, and re-run state machine.

  392ab8ab2587

• nemo-relay-debug-runtime-integration技能

  Use this skill when NeMo Relay is installed or imported but application-side runtime behavior is missing or incorrect, including load failures, inactive scopes, missing events, and plugin or adaptive wiring problems.

  392ab8ab2587

已保存的扫描说明 0

没有保存额外的扫描说明。

记录的别名 0

这些名称在上次记录时指向这个仓库。这是记录下来的解析结果，不是实时的 GitHub 检查；本站不会因此合并包或元数据。

没有记录到指向这个仓库的别名。

不支持的市场条目 0

这个仓库的市场清单里、本站暂不支持的条目。它们是原样保存的来源数据，不会被抓取或执行。

没有记录到不支持的条目。