zhihu/cuBERT

Fast implementation of BERT inference directly on NVIDIA (CUDA, CUBLAS) and Intel MKL

47
/ 100
Emerging

Eliminates TensorFlow framework overhead through direct kernel implementation, supporting mixed-precision inference (fp16 computation with fp32 accuracy) on Volta/Turing GPUs for 2x speedup. Provides thread-safe serving via `BertM` wrapper with configurable request and operation-level parallelism, plus language bindings for Python (Cython) and Java (JNA) alongside the core C++ API.

549 stars. No commits in the last 6 months.

Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 21 / 25

How are scores calculated?

Stars

549

Forks

84

Language

C++

License

MIT

Last pushed

Nov 18, 2020

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/zhihu/cuBERT"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.