WeianMao / triattention
TriAttention — Efficient long reasoning with trigonometric KV cache compression. Enables OpenClaw local deployment on memory-constrained GPUs.
- Repository stars
- 868All-time #44,053
- Fork
- 79
- Last push
- updated 3 months ago
- Repository created
- since 2026
About
TriAttention — Efficient long reasoning with trigonometric KV cache compression. Enables OpenClaw local deployment on memory-constrained GPUs.Summarised by AI from the README; the original prevails
Author description: TriAttention — Efficient long reasoning with trigonometric KV cache compression. Enables OpenClaw local deployment on memory-constrained GPUs.
Basic listing based on official public information; functionality has not been independently evaluated.
- Pricing
- Pricing not verifiedCheck current official software, hosting, model and infrastructure pricing.
