Jeff-Code: a 0.8B model makes Qwen 3.8-27B coding 47% faster, same pass rate

2 pointsposted 7 hours ago
by firelex
(github.com)

1 Comments