Skip to content
Merged
Changes from all commits
Commits
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
2 changes: 1 addition & 1 deletion train_thai_tiny_hf.py
Original file line number Diff line number Diff line change
Expand Up @@ -239,6 +239,7 @@ def collate(b):

# ---- greedy decode kaikki-test, corpus PER ----
model.eval()
inv = {v: k for k, v in vocab.items()}
preds = []
with torch.no_grad():
for i in range(0, len(test), 64):
Expand Down Expand Up @@ -266,7 +267,6 @@ def collate(b):
for row in ys[:, 1:].tolist():
toks = [c for c in row if c not in (pad, eos)]
preds.append("".join(inv[t] for t in toks))
inv = {v: k for k, v in vocab.items()}
per = corpus_per(list(zip(preds, [t for _, t in test])))
print(f"[gate] val PER (greedy, corpus) = {per:.4f}% gate<={GATE_PER}%", flush=True)

Expand Down
Loading