Skip to content

Fix loading configured self-play checkpoints #673

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
bagel786 wants to merge 1 commit into
base: 5c
Choose a base branch
from
+10 −0
Open

Fix loading configured self-play checkpoints #673

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions src/pufferl.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1735,6 +1735,9 @@ void pufferl_load_policy(PuffeRL* pufferl, int i, const char* path) {
Policy* pol = &pufferl->policies[i];
puf_load_weights_into(pol->master_weights, pol->param,
pufferl->default_stream, path);
if (i == 0 && pufferl->hypers.async) {
puf_copy(&pufferl->actor_param, &pol->param, pufferl->default_stream);
}
cudaDeviceSynchronize();
}

Expand Down Expand Up @@ -2959,6 +2962,13 @@ TrainResult run_train(Ini* ini, TrainContext* ctx) {
}

PuffeRL* pufferl = create_pufferl(ini, ctx);
char load_path_buf[4096];
const char* load_path = puf_checkpoint_path_key(
ini, "load_model_path", load_path_buf, sizeof(load_path_buf));
if (load_path) {
pufferl_load_policy(pufferl, 0, load_path);
}

Selfplay selfplay = {0};
if (use_selfplay) {
char initial_checkpoint[4096];
Expand Down