Changed `num_beams` to `num_beams // num_beam_groups` when initialising `PrefixConstrainedLogitsProcessor` in `_get_logits_processor` to fix compatibility issue when constrained decoding is used together with grouped beam search (#10475)

src/transformers/generation_utils.py

@@ -605,7 +605,7 @@ class GenerationMixin:
         if min_length is not None and eos_token_id is not None and min_length > -1:
             processors.append(MinLengthLogitsProcessor(min_length, eos_token_id))
         if prefix_allowed_tokens_fn is not None:
-            processors.append(PrefixConstrainedLogitsProcessor(prefix_allowed_tokens_fn, num_beams))
+            processors.append(PrefixConstrainedLogitsProcessor(prefix_allowed_tokens_fn, num_beams // num_beam_groups))
         if forced_bos_token_id is not None:
             processors.append(ForcedBOSTokenLogitsProcessor(forced_bos_token_id))
         if forced_eos_token_id is not None: