server: fix incorrectly reported token probabilities #7125
JohannesGaessler
merged
2 commits into
ggml-org:master
from
JohannesGaessler:sampling-fix-p-sum
server: normalize token probabilities
e25ca431
ggerganov
approved these changes
on 2024-05-07
fix temperature == 0.0f
ae8235dc
Login to write a write a comment.
Login via GitHub