~ / leaderboard / skill / pytorch-fsdp2
Skill · #4965
pytorch-fsdp2
Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you ne
4rigs use it
0%of all rigs
6.9kavg rig context tax
Get it
$ npx degit Yichen-ZJU/lemvo/claude/skills/pytorch-fsdp2 .claude/skills/pytorch-fsdp2 From the most-starred rig that ships it: Yichen-ZJU/lemvo. Same-named skills in different rigs may differ — review before use.
Often used together with
skill-creator · 4ml-paper-writing · 4clip · 4awq-quantization · 4brainstorming-research-ideas · 4creative-thinking-for-research · 4deepspeed · 4evaluating-llms-harness · 4gguf-quantization · 4gptq · 4
Rigs using pytorch-fsdp2 (4)
Yichen-ZJU/lemvo
Lemvo — a research agent that thinks and acts. One research system, two CLI engines: Claude Code and Codex CLI.
Skill Collector 6.8k tok · · Q0
Yichen-ZJU/lemvo-for-claude
Lemvo for Claude Code — choose routes, build methods, learn from experiments, and write discoveries. 115 skills, two research loops, shared project memory.
Skill Collector 7.1k tok · · Q0
Yichen-ZJU/claude-for-research
Autonomous research for Claude Code: literature, ideas, unattended experiments, writing and review. 115 skills, two loops.
Skill Collector 7.0k tok · · Q0
TrainToGPB/agentic-coding-setup
Personal Claude Code setup for LLM engineering — custom agents, skills, and a library organized by context
Skill Collector 6.6k tok · · Q0