diff --git a/whisper/decoding.py b/whisper/decoding.py
index 457ee7ccb..ecd98a455 100644
--- a/whisper/decoding.py
+++ b/whisper/decoding.py
@@ -146,6 +146,10 @@ def __init__(self, model: "Whisper", initial_token_length: int):
         self.kv_cache = {}
         self.hooks = []
 
+        key_modules = [block.attn.key for block in self.model.decoder.blocks]
+        value_modules = [block.attn.value for block in self.model.decoder.blocks]
+        self.kv_modules = key_modules + value_modules
+
     def logits(self, tokens: Tensor, audio_features: Tensor) -> Tensor:
         if not self.kv_cache:
             self.kv_cache, self.hooks = self.model.install_kv_cache_hooks()
@@ -164,9 +168,10 @@ def cleanup_caching(self):
         self.hooks = []
 
     def rearrange_kv_cache(self, source_indices):
-        for module, tensor in self.kv_cache.items():
-            # update the key/value cache to contain the selected sequences
-            self.kv_cache[module] = tensor[source_indices].detach()
+        if source_indices != list(range(len(source_indices))):
+            for module in self.kv_modules:
+                # update the key/value cache to contain the selected sequences
+                self.kv_cache[module] = self.kv_cache[module][source_indices].detach()
 
 
 class SequenceRanker:
@@ -668,7 +673,6 @@ def _detect_language(self, audio_features: Tensor, tokens: Tensor):
         return languages, lang_probs
 
     def _main_loop(self, audio_features: Tensor, tokens: Tensor):
-        assert audio_features.shape[0] == tokens.shape[0]
         n_batch = tokens.shape[0]
         sum_logprobs: Tensor = torch.zeros(n_batch, device=audio_features.device)
         no_speech_probs = [np.nan] * n_batch
@@ -721,8 +725,7 @@ def run(self, mel: Tensor) -> List[DecodingResult]:
                 )
             ]
 
-        # repeat the audio & text tensors by the group size, for beam search or best-of-n sampling
-        audio_features = audio_features.repeat_interleave(self.n_group, dim=0)
+        # repeat text tensors by the group size, for beam search or best-of-n sampling
         tokens = tokens.repeat_interleave(self.n_group, dim=0).to(audio_features.device)
 
         # call the main sampling loop