search

learning-at-home / lean_transformer

Memory-efficient transformer. Work in progress.
MIT License
19 stars 3 forks source link

add attn cache #12

Closed xtinkt closed 2 years ago

codecov-commenter commented 2 years ago

Codecov Report

Merging #12 (90b1807) into main (e356040) will not change coverage. The diff coverage is 80.30%.

@@           Coverage Diff           @@
##             main      #12   +/-   ##
=======================================
  Coverage   79.61%   79.61%           
=======================================
  Files          16       16           
  Lines        1339     1339           
=======================================
  Hits         1066     1066           
  Misses        273      273
Impacted Files Coverage Δ
lean_transformer/models/gpt.py 76.79% <74.41%> (ø)
lean_transformer/transformer.py 89.02% <80.00%> (ø)
lean_transformer/attn.py 92.30% <94.11%> (ø)
lean_transformer/__init__.py 100.00% <100.00%> (ø)

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 666365e...90b1807. Read the comment docs.