Clean artifacts and update example pipeline

2026-01-22 16:32:51 +08:00
parent c0639386be
commit c3f750cd9d
20 changed files with 651 additions and 30826 deletions
--- a/example/data_utils.py
+++ b/example/data_utils.py
@@ -66,6 +66,8 @@ def build_vocab(
    vocab = {}
    for c in disc_cols:
        tokens = sorted(values[c])
+        if "<UNK>" not in tokens:
+            tokens.append("<UNK>")
        vocab[c] = {tok: idx for idx, tok in enumerate(tokens)}
    return vocab

@@ -105,7 +107,7 @@ def windowed_batches(
    batches_yielded = 0
    for row in iter_rows(path):
        cont_row = [float(row[c]) for c in cont_cols]
-        disc_row = [vocab[c][row[c]] for c in disc_cols]
+        disc_row = [vocab[c].get(row[c], vocab[c]["<UNK>"]) for c in disc_cols]
        seq_cont.append(cont_row)
        seq_disc.append(disc_row)
        if len(seq_cont) == seq_len: