update users/raissi

rwth-i6 · Dec 11, 2024 · aae42ae · aae42ae
1 parent 180646d
commit aae42ae
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 6 deletions.
diff --git a/users/raissi/experiments/domain_mismtach/medline/base_args.py b/users/raissi/experiments/domain_mismtach/medline/base_args.py
@@ -14,6 +14,7 @@
 import i6_core.corpus as corpus_recipes
 import i6_core.meta as meta
 from i6_core.audio import BlissChangeEncodingJob
+from i6_core.lexicon import MergeLexiconJob
 
 from i6_experiments.common.baselines.librispeech.data import CorpusData
 from i6_experiments.common.datasets.librispeech import (
@@ -143,7 +144,7 @@ class DATASET:
         ),
         corpus=wmt22_medline_noise07,
         lm=tk.Path(f"{PREPATH_ASR3}/lm/v2/only_medline/ufal_v1_lm_3more.gz", cached=True, hash_overwrite="v22_lm"),
-        description="lexicon uses both LBS and medline data with words repeating 3 or more",
+        description="lexicon uses only medline data with words repeating 3 or more",
     ),
     0.3: DATASET(
         lexicon_with_unk=tk.Path(
@@ -158,7 +159,7 @@ class DATASET:
         ),
         corpus=wmt22_medline_noise03,
         lm=tk.Path(f"{PREPATH_ASR3}/lm/v2/only_medline/ufal_v1_lm_3more.gz", cached=True, hash_overwrite="v22_lm"),
-        description="lexicon uses both LBS and medline data with words repeating 3 or more",
+        description="lexicon uses only medline data with words repeating 3 or more",
     ),
 }
 
@@ -295,6 +296,11 @@ def get_corpus_data_inputs(
         if add_unknown_for_medline_lex
         else MEDLINE_DATA["dev"][version][noise].lexicon_no_unk
     )
+    if version > 1:
+        seed_lexicon = "seed_withunk.xml.gz" if add_unknown_for_medline_lex else "seed_nounk.xml.gz"
+        seed_lexicon_path = tk.Path(("/").join([f"{PREPATH_ASR3}", f"lexicon/seed_lbs_lexicon_nolemmata/{seed_lexicon}"]), hash_overwrite=f"seed_{seed_lexicon}")
+        med_lex = MergeLexiconJob([seed_lexicon_path, med_lex]).out_bliss_lexicon
+
     oov_lexicon_medline = {
         "filename": med_lex,
         "normalize_pronunciation": False,

diff --git a/users/raissi/setups/librispeech/decoder/LBS_factored_hybrid_search.py b/users/raissi/setups/librispeech/decoder/LBS_factored_hybrid_search.py
@@ -76,7 +76,7 @@ def __init__(
             gpu=gpu,
         )
         self.trafo_lm_config = self.get_eugen_trafo_with_quant_and_compress_config()
-        self.lstm_lm_config = self.get_kazuki_lstm_config()
+        #self.lstm_lm_config = self.get_kazuki_lstm_config()
 
 
     def get_kazuki_lstm_config(
@@ -88,14 +88,13 @@ def get_kazuki_lstm_config(
     ) -> rasr.RasrConfig:
 
         assert self.library_path is not None
-        dependency_path = Path("/work/asr4/raissi/setups/librispeech/960-ls/dependencies/trafo-lm_kazuki/IS2019", hash_overwrite="LBS_LM_KAZUKI")
 
         trafo_config = rasr.RasrConfig()
 
         #model and graph info
         trafo_config.loader.type = "meta"
-        trafo_config.loader.meta_graph_file = tk.Path("/u/rossenbach/experiments/asru_ls100_full_context_transducer/work/crnn/compile/CompileTFGraphJob.0dxq1DSvOxuN/output/graph.meta", cached=True)
-        trafo_config.loader.saved_model_file = DelayedFormat("/u/zhou/asr-exps/librispeech/dependencies/kazuki_lstmlm_27062019/network.040")
+        trafo_config.loader.meta_graph_file = tk.Path("/u/raissi/Desktop/debug/lm/graph.meta", cached=True)
+        trafo_config.loader.saved_model_file = DelayedFormat("/work/asr4/rossenbach/custom_projects/kazuki_replicate_lm_training/net-model/network.029")
         trafo_config.loader.required_libraries = self.library_path
 
         trafo_config.type = "tfrnn"