Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing.
Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing.
提供基于 PyTorch 的分布式 LLM 预训练,使用 torchtitan 实现 4D 并行(FSDP2、TP、PP、CP)。在大规模预训练 Llama 3.1、DeepSeek V3 或自定义模型时,适用于从 8 到 512+ 个 GPU,支持 Float8、torch.compile 和分布式检查点。
Category: developer (开发工具) · Author: davila7 · Version: @main · License: MIT
Tags: Model Architecture, Distributed Training, TorchTitan, FSDP2, Tensor Parallel, Pipeline Parallel, Context Parallel, Float8, Llama, Pretraining
该 Skill 暂无文档文件。
npx skills add davila7/distributed-llm-pretraining-torchtitan下载完整 Skill 目录,包含 SKILL.md 及所有相关文件
Search for places (restaurants, cafes, etc.) via Google Places API proxy on localhost.
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
Create or update AgentSkills. Use when designing, structuring, or packaging skills with scripts, references, and assets.
Start voice calls via the OpenClaw voice-call plugin.
Notion API for creating and managing pages, databases, and blocks.
Gemini CLI for one-shot Q&A, summaries, and generation.
Category:developer
Tags:Model Architecture, Distributed Training, TorchTitan, FSDP2, Tensor Parallel, Pipeline Parallel, Context Parallel, Float8, Llama, Pretraining