optimizing-attention-flash
De la autor
Enables Flash Attention for transformer models using PyTorch native scaled_dot_product_attention (PyTorch 2.2+) or the flash-attn library, including multi-query attention, sliding window attention, and FP8 on H100 (FlashAttention-3). Covers profiling speedup, checking accuracy against a baseline, and troubleshooting install and GPU support errors. Use when training or running transformers on long sequences (over 512 tokens), when standard attention runs out of GPU memory, when attention is the inference bottleneck, when switching a PyTorch model to the flash backend, or when tuning attention on H100 GPUs. Not for CPU inference, V100 GPUs, or sequences under 256 tokens; consider xFormers for other attention variants.
Instalare
npx skills add KalarisLabs/research-agent-skills --skill optimizing-attention-flash -a claude-code
Se instalează în .claude/skills/optimizing-attention-flash/ · Fiecare comandă descarcă skill-ul direct de pe GitHub cu utilitarul open source skills.
Identic octet cu octet cu commit-ul scanat, cu licența depozitului și un fișier SOURCE.txt.
Vrei ca agentul să-ți propună skill-uri în timp ce scrii? Vezi cum folosești skill-urile din IDE.
Verificări de siguranță pe SKILL.md
Niciun avertisment.
De ce 82
aiqbench calculează scorul de încredere din semnale publice. Nu este un audit de securitate, așa că citește un skill înainte să-l instalezi.
Skill-uri similare
Cele mai apropiate de acesta prin ce fac și când sunt folosite.
- pytorchMindrally/skillsPyTorch deep learning development with transformers, diffusion models, and GPU optimization.Altele★ 286
- tensorrt-llmOrchestra-Research/AI-Research-SKILLsOptimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency. Use for production deployment on NVIDIA GPUs (A100/H100), when you need 10-100x faster inference than PyTorch, or for serving models with quantization (FP8/INT4), in-flight batching, and multi-GPU scaling.DevOps★ 13.9K
- V3 Performance Optimizationspencermarx/open-code-reviewAchieve aggressive v3 performance targets: 2.49x-7.47x Flash Attention speedup, 150x-12,500x search improvements, 50-75% memory reduction. Comprehensive benchmarking and optimization suite.Productivitate★ 380
- pytorch-fsdp2Orchestra-Research/AI-Research-SKILLsAdds 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 need DTensor-based sharding with DeviceMesh.Productivitate★ 13.9K
- huggingface-accelerateKalarisLabs/research-agent-skillsWraps existing PyTorch training scripts with HuggingFace Accelerate (Accelerator class, accelerate config, accelerate launch) so the same code runs on CPU, single GPU, multi-GPU, multi-node, TPU, or Apple MPS. Covers device placement, FP16/BF16/FP8 mixed precision, gradient accumulation, distributed checkpointing, and switching between DDP, DeepSpeed ZeRO, FSDP, and Megatron backends. Use when converting a single-GPU script to multi-GPU, enabling mixed precision, configuring DeepSpeed ZeRO or FSDP, or setting up gradient accumulation. Use when one script must run on different hardware. Not for callback-heavy training loops (use PyTorch Lightning) or multi-node orchestration with hyperparameter tuning (use Ray Train).Inginerie★ 61
- triton-sageattentionartokun/comfyui-mcpInstall Triton + SageAttention to accelerate ComfyUI (the sageattn attention_mode and inductor torch.compile used by WanVideoWrapper / many video graphs). Windows-first (triton-windows + woct0rdho prebuilt SageAttention wheels matched to torch/CUDA/python into the RIGHT python), plus Linux (official triton + build) and Mac (N/A → sdpa/MPS). Also covers the SAFE sdpa / no-compile fallback so an example that assumes sageattn + torch.compile still runs when these aren't installed (video-extend TRAP 5). Use when a loader crashes with "No module named 'sageattention'" or reports triton unavailable, when asked to speed up Wan/video workflows, or when deciding whether to install acceleration vs. fall back.Media★ 861
Note și recenzii
—
nicio recenzie
Ai încercat acest skill? Notează-l ca să-i ajuți pe alții. Autentifică-te
Discuție
Intră în discuție. Autentifică-te