PyPI page
Home page
Author:
Leonard Lin
Summary:
Production-speed compact Dynamic Memory Sparsification (DMS) for KV cache compression
Latest version:
0.2.0
Required dependencies:
flash-attn
|
huggingface-hub
|
numpy
|
safetensors
|
torch
|
tqdm
|
transformers
|
triton
|
xxhash
Optional dependencies:
datasets
|
fast-hadamard-transform
|
pytest
Downloads last day:
1
Downloads last week:
18
Downloads last month:
138