TemporalMesh Transformer: 29.4 PPL at 48% compute — dynamic graph attention + adaptive exit gates (open-source)
#23 opened about 5 hours ago
by
vigneshwar234
Comparable between languages
1
#21 opened about 2 years ago
by
VityaVitalich
List Punctuation
#6 opened over 3 years ago
by
Tristan