Back to rankings

feifeibear / long-context-attention

USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inference

Repository stars
696All-time #50,459
Fork
87
Last push
updated 1 months ago
Repository created
since 2024
Apps & systemsGeneralPythonApache-2.0 · Permissive
View source

About

Original · GitHub
USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inference

Summarised by AI from the README; the original prevails

Author description: USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inference

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.

Star trend

69669769711-06 · 69610-0810-1310-1810-2310-2811-0211-06

First collected 2026-11-06Measured at 2026-11-06 09:46 UTCSourceLicense source

Compiled from public project information; not independently tested. · Repository Stars, not unique users · License tiers are derived from metadata. Check obligations for your use; unknown does not mean prohibited.