Agent Skills: optimization-flash-attention
Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with long sequences (>512 tokens), encountering GPU memory issues with attention, or need faster inference. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.
[OptimizationFlashAttentionAttentionOptimizationMemoryEfficiencySpeedOptimizationLongContextPyTorchSDPAH100FP8Transformers]
ID: davila7/claude-code-templates/optimization-flash-attention
33,1943,891
Install this agent skill to your local
Skill Files
The files that make up optimization-flash-attention.
More agent skills
- docker-expertDocker containerization expert with deep knowledge of multi-stage builds, image optimization, container security, Docker Compose orchestration, and production deployment patterns. Use PROACTIVELY for Dockerfile optimization, container issues, image size problems, security hardening, networking, and orchestration challenges.
- copywritingWhen the user wants to write, rewrite, or improve marketing copy for any page — including homepage, landing pages, pricing pages, feature pages, about pages, or product pages. Also use when the user says "write copy for," "improve this copy," "rewrite this page," "marketing copy," "headline help," or "CTA copy." For email copy, see email-sequence. For popup copy, see popup-cro.
- cto-advisorTechnical leadership guidance for engineering teams, architecture decisions, and technology strategy. Includes tech debt analyzer, team scaling calculator, engineering metrics frameworks, technology evaluation tools, and ADR templates. Use when assessing technical debt, scaling engineering teams, evaluating technologies, making architecture decisions, establishing engineering metrics, or when user mentions CTO, tech debt, technical debt, team scaling, architecture decisions, technology evaluation, engineering metrics, DORA metrics, or technology strategy.
- code-reviewerYou MUST use this when asked to review code: a pull request, a branch, a commit, staged or uncommitted changes, or \"check this before I merge\". Scopes the review to the diff, reads the surrounding code, hunts for real bugs (correctness, security, data loss, concurrency, broken callers, missing tests), verifies every finding against the code before reporting it, and returns a short severity-ranked list with a concrete failure scenario for each. Use it for any language; not for writing new code or for style-only cleanups.
- autonomous-agentsAutonomous agents are AI systems that can independently decompose goals, plan actions, execute tools, and self-correct without constant human guidance. The challenge isn't making them capable - it's making them reliable. Every extra decision multiplies failure probability. This skill covers agent loops (ReAct, Plan-Execute), goal decomposition, reflection patterns, and production reliability. Key insight: compounding error rates kill autonomous agents. A 95% success rate per step drops to 60% b
- create-planCreate a concise plan. Use when a user explicitly asks for a plan related to a coding task.