mirror of
https://github.com/Vision-CAIR/MiniGPT-4.git
synced 2025-04-05 18:40:46 +00:00
LR resets every epoch during warmup to `warmup_start_lr` because it uses inner epoch step counter. We need to use `total_cur_step` to increment LR continuously from epoch to epoch. |
||
---|---|---|
.. | ||
common | ||
configs | ||
conversation | ||
datasets | ||
models | ||
processors | ||
runners | ||
tasks | ||
__init__.py |