Add LoRA for Zipformer

egs/librispeech/ASR/README.md

@@ -36,6 +36,7 @@ The following table lists the differences among them.
 | `lstm_transducer_stateless3`          | LSTM | Embedding + Conv1d | Using LSTM with mechanisms in reworked model + gradient filter + delay penalty |
 | `zipformer`                           | Upgraded Zipformer | Embedding + Conv1d | The latest recipe |
 | `zipformer_adapter`                           | Upgraded Zipformer | Embedding + Conv1d | It supports domain adaptation of Zipformer using parameter efficient adapters |
+| `zipformer_adapter`                           | Upgraded Zipformer | Embedding + Conv1d | Finetune Zipformer with LoRA  |
 
 The decoder in `transducer_stateless` is modified from the paper
 [Rnn-Transducer with Stateless Prediction Network](https://ieeexplore.ieee.org/document/9054419/).

egs/librispeech/ASR/tdnn_lstm_ctc/asr_datamodule.py

@@ -479,18 +479,14 @@ def test_other_cuts(self) -> CutSet:
     @lru_cache()
     def gigaspeech_subset_small_cuts(self) -> CutSet:
         logging.info("About to get Gigaspeech subset-S cuts")
-        return load_manifest_lazy(self.args.manifest_dir / "gigaspeech_cuts_S.jsonl.gz")
+        return load_manifest_lazy(self.args.manifest_dir / "cuts_S.jsonl.gz")
 
     @lru_cache()
     def gigaspeech_dev_cuts(self) -> CutSet:
         logging.info("About to get Gigaspeech dev cuts")
-        return load_manifest_lazy(
-            self.args.manifest_dir / "gigaspeech_cuts_DEV.jsonl.gz"
-        )
+        return load_manifest_lazy(self.args.manifest_dir / "cuts_DEV.jsonl.gz")
 
     @lru_cache()
     def gigaspeech_test_cuts(self) -> CutSet:
         logging.info("About to get Gigaspeech test cuts")
-        return load_manifest_lazy(
-            self.args.manifest_dir / "gigaspeech_cuts_TEST.jsonl.gz"
-        )
+        return load_manifest_lazy(self.args.manifest_dir / "cuts_TEST.jsonl.gz")

egs/librispeech/ASR/zipformer_lora/asr_datamodule.py

@@ -0,0 +1 @@
+../tdnn_lstm_ctc/asr_datamodule.py

egs/librispeech/ASR/zipformer_lora/beam_search.py

@@ -0,0 +1 @@
+../pruned_transducer_stateless2/beam_search.py