[sampler] fix mixed penalties in batch with async scheduling (#1594)

Signed-off-by: AlpinDale <alpindale@gmail.com>

Author: AlpinDale

aphrodite/v1/sample/ops/penalties.py

@@ -18,6 +18,14 @@ def apply_all_penalties(
     """
     _, vocab_size = logits.shape
     output_tokens_t = _convert_to_tensors(output_token_ids, vocab_size, logits.device)
+
+    # In the async scheduling case, rows that won't have penalties applied may contain
+    # -1 placeholder token ids. We must replace these with valid token ids so that the
+    # scatter done in apply_penalties is valid.
+    # NOTE: The penalties implementation is currently quite inefficient and
+    # will be reworked anyhow.
+    output_tokens_t.masked_fill_(output_tokens_t == -1, vocab_size)
+
     return apply_penalties(
         logits,
         prompt_token_ids,