Wheelodex » Projects » flash-attention-softmax-n

flash-attention-softmax-n

View on PyPI — Reverse Dependencies (0)

0.3.2 flash_attention_softmax_n-0.3.2-py3-none-any.whl

Wheel Details

Project: flash-attention-softmax-n
Version: 0.3.2
Filename: flash_attention_softmax_n-0.3.2-py3-none-any.whl
Download: [link]
Size: 34627
MD5: 8637e06aea92f8e2c51b60fe9fdeaffd
SHA256: f41d9dabe136d0c74a35ba247bb88b4f97cb281b4c5c5a1249af0cc790ae9596
Uploaded: 2023-11-21 14:15:27 +0000

dist-info

METADATA · WHEEL · RECORD · top_level.txt

METADATA

Metadata-Version: 2.1
Name: flash-attention-softmax-n
Version: 0.3.2
Summary: CUDA and Triton implementations of Flash Attention with SoftmaxN.
Author: Christopher W. Murphy
Author-Email: murphtron5000[at]gmail.com
Home-Page: https://github.com/softmax1/Flash-Attention-Softmax-N
License: GPLv3
Keywords: artificial intelligence,attention mechanism,transformers
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3)
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Requires-Python: >=3.9
Requires-Dist: torch (>=2.0.0)
Requires-Dist: einops (>=0.6.1)
Requires-Dist: mosaicml (>=0.16.0); extra == "surgery"
Requires-Dist: transformers (<4.33,>=4.11); extra == "surgery"
Requires-Dist: triton (>=2.0.0); extra == "triton"
Provides-Extra: surgery
Provides-Extra: triton
Description-Content-Type: text/markdown
License-File: LICENSE
[Description omitted; length: 9390 characters]

WHEEL

Wheel-Version: 1.0
Generator: bdist_wheel (0.41.3)
Root-Is-Purelib: true
Tag: py3-none-any

RECORD

Path Digest Size
flash_attention_softmax_n/__init__.py sha256=dBob_wgxykXAsu-yKq5vMcs2n7OzpLI0hY1IwbxCJsI 430
flash_attention_softmax_n/analysis/__init__.py sha256=piNsbV0o8wuo3sownDkJfeCHl1Nj3JAkF-zPQHZ1Ru8 169
flash_attention_softmax_n/analysis/hooks.py sha256=SDjdimi-w4OFr3eULBO01nHJ_lwhPwvH3uS8oS0R4IY 3249
flash_attention_softmax_n/analysis/io.py sha256=gmoexTkZca42nYvVEThBll-Jch5poG1yyTU0VngDtTk 355
flash_attention_softmax_n/analysis/statistics.py sha256=YzQDjql3r9QbtkjzgFSoVzTFS7GlabvX3PyBM2bXwWs 2048
flash_attention_softmax_n/core/__init__.py sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU 0
flash_attention_softmax_n/core/flash_attn.py sha256=63aJtvBjTpeMzomFnMWuKlUnd8EDman9Q46BQt-e5wU 4867
flash_attention_softmax_n/core/flash_attn_triton.py sha256=ITS9o_AdqhJfYP0Uy69BdcW3VhCmo790PIESafs125c 13613
flash_attention_softmax_n/core/functional.py sha256=Rmk9UxwTK9HRaTQ29f-qxL6RpDSGlOla3uxLctu5Fik 4197
flash_attention_softmax_n/surgery/__init__.py sha256=Seztl0B3-Sr-Te4UXs7Wx1mW61pvosm_3AfSpwl3rXA 207
flash_attention_softmax_n/surgery/attention_softmax_n.py sha256=LElDorVPiEVj3o9_XcE_ntZbQCnnifObtYBB_iPy4ZM 3454
flash_attention_softmax_n/surgery/surgery_functions/__init__.py sha256=326jLTQ_tP_gJ5JRUYmcCrFOHTDnoO4dU8YqSSWXALM 324
flash_attention_softmax_n/surgery/surgery_functions/_bert.py sha256=B4nX9NrmTJzZHIyYwD8CMwWQSeMSgKJ-xShWQqA1jh0 6163
flash_attention_softmax_n/surgery/surgery_functions/_xlnet.py sha256=JijO6PPUGznx41BiHIBEcLIzC_sVcyEPRMLcci0Dzn4 2440
flash_attention_softmax_n/surgery/surgery_functions/utils.py sha256=SyMYJyZByDitQGHNf9_yhyVEBDJKiopjU1IgO1l2RUg 4220
flash_attention_softmax_n-0.3.2.dist-info/LICENSE sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY 35149
flash_attention_softmax_n-0.3.2.dist-info/METADATA sha256=2ceK67jiq4O5UNdaYUHtIAP1wqRQLPEQ91Z9LTLB5y8 10506
flash_attention_softmax_n-0.3.2.dist-info/WHEEL sha256=Xo9-1PvkuimrydujYJAjF7pCkriuXBpUPEjma1nZyJ0 92
flash_attention_softmax_n-0.3.2.dist-info/top_level.txt sha256=BBka1pdHqccXleM5mQKSEFT4HECz1awVoRY_fOKBfTw 26
flash_attention_softmax_n-0.3.2.dist-info/RECORD — —

top_level.txt

flash_attention_softmax_n