职业分类
软件开发工程师
描述
Guide for adding support for new LLM or VLM models in Megatron-Bridge. Covers bridge, provider, recipe, tests, docs, and examples.
原文语言：英语
更新
这个仓库中的 skills
autohandai/community-skills - 第 2 页
SkillsMP 已收集 autohandai/community-skills 中的 1,040 个 Skill。打开任一 Skill 可查看来源和详情。
autohandai/community-skills已展示 40 / 1,040 个已收集 Skill。
skill
职业分类
描述
更新
职业分类
软件开发工程师
描述
Dev environment setup for Megatron Bridge — container-based development, uv package management, lockfile regeneration, adding dependencies, Slurm container usage, and common build pitfalls.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Bump a pinned dependency (TransformerEngine, Megatron-LM, NRX, etc.), regenerate the lockfile, open a PR, and drive it to green by attaching a watchdog to the "CICD NeMo" workflow and quarantining failing functional tests as flaky until the run is green.
原文语言：英语
更新
职业分类
软件开发工程师
描述
CI/CD reference for Megatron Bridge — pipeline structure, commit and PR workflow, CI failure investigation, and common failure patterns.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Code style and quality rules for Megatron Bridge — ruff configuration, naming conventions, type hints, mypy rules, docstrings, copyright headers, logging, and the code review checklist.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Run Megatron-LM (MLM) and Megatron Bridge training with mock or real data. Covers correlation testing, available recipes, and multi-GPU examples.
原文语言：英语
更新
职业分类
网络与计算机系统管理员
描述
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 allocation.
原文语言：英语
更新
职业分类
软件质量保证分析师与测试员
描述
External NeMo-RL end-to-end validation workflow for Megatron-Bridge model/provider changes, including downstream compatibility checks, external RL lifecycle behavior, Megatron policy setup, HF import/export, checkpoint/resume, non-colocated vLLM refit, delta…
原文语言：英语
更新
职业分类
软件质量保证分析师与测试员
描述
Structured framework for verifying numerical parity of HF-to-MCore weight conversions. References existing tools and the add-model-support skill.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Validate and use selective and full activation recompute in Megatron Bridge to reduce GPU memory usage at the cost of extra compute.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Validate and use CPU offloading in Megatron Bridge, including layer-level activation offloading and fractional optimizer state offloading with HybridDeviceOptimizer.
原文语言：英语
更新
职业分类
软件开发工程师
描述
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.
原文语言：英语
更新
职业分类
软件开发工程师
描述
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.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Operational guide for enabling hierarchical context parallelism in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Operational guide for enabling Megatron FSDP in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Techniques for reducing peak GPU memory in Megatron Bridge — expandable segments, parallelism resizing, activation recompute, CPU offloading constraints, and common OOM fixes.
原文语言：英语
更新
职业分类
软件开发工程师
描述
MoE expert-parallel communication overlap in Megatron Bridge. Covers dispatch/combine overlap, flex dispatcher backends, and expert wgrad scheduling.
原文语言：英语
更新
职业分类
软件开发工程师
描述
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.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Representative MoE training playbooks by hardware platform and model family. Summarizes rounded throughput bands, parallelism patterns, and common tuning stacks.
原文语言：英语
更新
职业分类
软件开发工程师
描述
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.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Systematic workflow for MoE training optimization in Megatron Bridge, based on the Megatron-Core MoE paper. Covers the Three Walls framework, parallel folding, recompute strategy, dispatcher choice, and CUDA-graph bring-up.
原文语言：英语
更新
职业分类
软件开发工程师
描述
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.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Operational guide for choosing and combining parallelism strategies in Megatron Bridge, including sizing rules, hardware topology mapping, and combined parallelism configuration.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Validate and use packed sequences and long-context training in Megatron-Bridge, distinguishing offline packed SFT for LLMs from in-batch packing for VLMs, and applying the right CP constraints.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Operational guide for enabling TP, DP, and PP communication overlap in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Recommend and customize Megatron Bridge recipes for a user's model, GPU count, and training goal. Indexes library recipes (pretrain/SFT/PEFT) and performance recipes.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Resiliency features in Megatron Bridge including fault tolerance, straggler detection, in-process restart, preemption, and re-run state machine.
原文语言：英语
更新
职业分类
软件质量保证分析师与测试员
描述
Testing reference for Megatron Bridge — unit and functional test layout, tier semantics (L0/L1/L2/flaky), script conventions, running tests locally, adding/moving/disabling tests, and pytest conventions.
原文语言：英语
更新
职业分类
软件质量保证分析师与测试员
描述
External verl end-to-end validation workflow for Megatron-Bridge model/provider changes. Covers running a small verl Megatron backend job from a Bridge checkout, choosing LoRA/DDP plus optional save/resume and parallelism variants, setting PYTHONPATH so verl…
原文语言：英语
更新
职业分类
软件开发工程师
描述
Container-based dev environment setup and dependency management for Megatron-LM. Covers acquiring and launching the CI container, uv package management, and updating uv.lock.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Bump the NVIDIA PyTorch base image using the nvcr.io/nvidia/pytorch YY.MM py3 tag used by Megatron-LM CI. Covers the two pin sites (GitHub CI in `docker/.ngc_version.dev` and GitLab CI in `.gitlab/stages/01.build.yml`), the post-bump CI loop (re-run…
原文语言：英语
更新
职业分类
软件开发工程师
描述
CI/CD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering internal GitLab CI, and CI failure investigation.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Investigate a failing GitHub Actions run or job and create a GitHub issue for the failure.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Linting and formatting for Megatron-LM. Covers running autoformat.sh, tools (ruff, black, isort, pylint, mypy), and code style rules.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Domain knowledge for the nightly main-to-dev sync workflow. Covers merge strategy, CI architecture, failure investigation, and known issues.
原文语言：英语
更新
职业分类
软件质量保证分析师与测试员
描述
Onboard 1-node GitHub MR functional tests for GB200 from existing mr-scoped 2-node tests.
原文语言：英语
更新
职业分类
软件开发工程师
描述
Research and draft a response to a GitHub issue or question from an external contributor.
原文语言：英语
更新
职业分类
网络与计算机系统管理员
描述
How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions,…
原文语言：英语
更新
职业分类
软件开发工程师
描述
Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.
原文语言：英语
更新
职业分类
软件质量保证分析师与测试员
描述
Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.
原文语言：英语
更新
已展示 40 / 1,040 个已收集 Skill。