Comment Comment by alekandreev This is mostly about inference speed, while maintaining long context performance. parent alekandreev 2 years ago
