flash-attention Skill
Speed up long-sequence transformer training and inference. Published by NousResearch in hermes-agent.
Decision snapshot
Is this a fit?
Developers using hermes-agent, Includes SKILL.md, Reusable instructions
Compatibility not yet detected.
Permission behavior not yet detected.
Copy skill directory
21 days ago · MIT license
No specific cautions were detected. Review the source and requested permissions before installing.
What is flash-attention Skill?
Speed up long-sequence transformer training and inference. Published by NousResearch in hermes-agent. This profile combines repository metadata with install, compatibility, and usage signals so developers can quickly decide whether it fits their agent workflow before opening the source repository.
Automated repository signals based on public metadata such as recency, license, installation evidence, and adoption. These are not a security audit or endorsement. See how SkillIndex evaluates profiles.
Key capabilities
- Includes SKILL.md support
- Reusable instructions support
- Developers using hermes-agent
Declared skill metadata
- Declared author: Orchestra Research
- Declared license: MIT
- Source file: optional-skills/mlops/flash-attention/SKILL.md
These fields retain source and confidence evidence from the indexed SKILL.md.
Compatibility and setup
- Install or run with Copy skill directory
When to use flash-attention Skill
- Use it for developers using hermes-agent.
Built with
Editorial notes
Source
- Creator: NousResearch
- Repository: NousResearch/hermes-agent
- Skill file: optional-skills/mlops/flash-attention/SKILL.md
What it does
Speed up long-sequence transformer training and inference.
Skill instructions
Flash Attention - Fast Memory-Efficient Attention Quick start Flash Attention provides 2-4x speedup and 10-20x memory reduction for transformer attention through IO-aware tiling and recomputation. PyTorch native (easiest, PyTorch 2.2+): python import torch import torch.nn.functional as F q = torch.randn(2, 8, 512, 64, device='cuda', dtype=torch.float16) [batch, heads, seq, dim] k = torch.randn(2, 8, 512, 64, device='cuda', dtype=torch.float16) v = torch.randn(2, 8, 512, 64, device='cuda', dtype=torch.float16) Automatically uses Flash Attention if available out = F.scaleddotproductattention(q, k, v) flash-attn library (more features): bash pip install flash-attn --no-build-isolation python from flashattn import flashattnfunc q, k, v: [batch, seqlen, nheads, headdim] out = flashattnfunc(q, k, v, dropoutp=0.0, causal=True) Common workflows Workflow 1: Enable in existing PyTorch model Copy this checklist: Flash Attention Integration: - [ ] Step 1: Check PyTorch version (≥2.2) - [ ] Step 2:
Verified compatibility and discovery
Frequently asked questions
What is flash-attention?
flash-attention is a open-source AI agent skill with Copy skill directory. Speed up long-sequence transformer training and inference.
Who is flash-attention best for?
flash-attention is best for reusing agent instructions, scripts, and references.
How do I install flash-attention?
Install or run flash-attention using Copy skill directory. Check flash-attention for the latest setup command.
Is flash-attention actively maintained?
flash-attention may need a closer maintenance check before production use.
Project health auto-fetched from the source repository.
Maintain this resource?
Review this source-backed profile, send a correction with evidence, or link to it from your documentation. Claims verify your relationship to the project; profile facts still require source evidence and editorial review.
