Add more info

README.md

@@ -9,10 +9,10 @@ tags:
 - xlsr-fine-tuning-week
 license: apache-2.0
 widget:
-- label: Malromur sample 11
-  src: https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/resolve/main/sample11.flac
-- label: Malromur sample 74
-  src: https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/resolve/main/sample74.flac
+- label: Malromur sample 1608
+  src: https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/resolve/main/sample1608.flac
+- label: Malromur sample 3860
+  src: https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/resolve/main/sample3860.flac
 model-index:
 - name: XLSR Wav2Vec2 Icelandic by Mehrdad Farahani
   results:
@@ -26,7 +26,7 @@ model-index:
     metrics:
        - name: Test WER
          type: wer
-         value: 12.00
+         value: 10.74
         
 ---
 
@@ -108,7 +108,7 @@ def predict(batch):
         
     pred_ids = torch.argmax(logits, dim=-1)
 
-    batch["predicted"] = processor.batch_decode(pred_ids)[0]
+    batch["predicted"] = processor.batch_decode(pred_ids)
     return batch
 
 
@@ -119,16 +119,16 @@ model = Wav2Vec2ForCTC.from_pretrained("m3hrdadfi/wav2vec2-large-xlsr-icelandic"
 dataset = load_dataset("csv", data_files={"test": "./malromur_test.csv"})["test"]
 dataset = dataset.map(
     normalizer, 
-    fn_kwargs={"remove_extra_space": True},
-    remove_columns=list(set(dataset.column_names) - set(['sentence', 'path']))
+    fn_kwargs={"do_lastspace_removing": True, "text_key_name": "cleaned_sentence"},
+    remove_columns=list(set(dataset.column_names) - set(['cleaned_sentence', 'path']))
 )
 
 dataset = dataset.map(speech_file_to_array_fn)
-result = dataset.map(predict)
+result = dataset.map(predict, batched=True, batch_size=8)
 
 max_items = np.random.randint(0, len(result), 20).tolist()
 for i in max_items:
-    reference, predicted =  result["sentence"][i], result["predicted"][i]
+    reference, predicted =  result["cleaned_sentence"][i], result["predicted"][i]
     print("reference:", reference)
     print("predicted:", predicted)
     print('---')
@@ -136,13 +136,72 @@ for i in max_items:
 
 **Output:**
 ```text
-SOON
+reference: lögregla rakti sporin í snjónum
+predicted: lögregla rakti sporinn í snjónum
+---
+reference: vaðlatúni
+predicted: vaðlatúni
+---
+reference: mykjunesi
+predicted: mikjunesi
+---
+reference: miðey
+predicted: miðey
+---
+reference: tveir mótmæla við stjórnarráðsbygginguna
+predicted: tveir mótmæla við stjórnarráðsbegginguna
+---
+reference: furðustrandir mest selda bók ársins
+predicted: furðustrandir mest seldabók ársins
+---
+reference: flekar brenndir í kvöld
+predicted: flekar brenndir í kvöld
+---
+reference: ástæðan er sögð eldgosið í grímsvötnum
+predicted: ástæðan er sögð eldgosið í grímsvötnum
+---
+reference: birtingur
+predicted: birtingur
+---
+reference: tvöþúsund og átján
+predicted: tvöþúsund og átján
+---
+reference: einfríður
+predicted: einfríður
+---
+reference: dalhúsum
+predicted: dalhúsum
+---
+reference: sex stútar á ferð
+predicted: sex stútar á ferð
+---
+reference: eyjamenn áfram í toppbaráttu
+predicted: eyjamenn áfram í toppbaráttu
+---
+reference: þetta októberkvöld sýndi sitt rétta andlit með hráslagakulda frá vatninu
+predicted: þetta októberkvöld sýnsint réttla andlit með hráslagakulda frá vatninu
+---
+reference: jes
+predicted: js
+---
+reference: hersveitirnar benda hvor á aðra
+predicted: hersveitirnar benda hvor á aðra
+---
+reference: þetta er hráskinnsleikur stórvelda eins og hver maður vissi
+predicted: þetta er hráskinnsleikur stórvelda eins og hver maður vissi
+---
+reference: umferð efstu deildar hófst
+predicted: umferð efstu deildar hófst
+---
+reference: freisting is
+predicted: freisting is
+---
 ```
 
 
 ## Evaluation
 
-The model can be evaluated as follows on the test data of Common Voice.
+The model can be evaluated as follows on the test data of Malromur.
 
 ```python
 import librosa
@@ -180,7 +239,7 @@ def predict(batch):
         
     pred_ids = torch.argmax(logits, dim=-1)
 
-    batch["predicted"] = processor.batch_decode(pred_ids)[0]
+    batch["predicted"] = processor.batch_decode(pred_ids)
     return batch
 
 
@@ -191,21 +250,21 @@ model = Wav2Vec2ForCTC.from_pretrained("m3hrdadfi/wav2vec2-large-xlsr-icelandic"
 dataset = load_dataset("csv", data_files={"test": "./malromur_test.csv"})["test"]
 dataset = dataset.map(
     normalizer, 
-    fn_kwargs={"remove_extra_space": True},
-    remove_columns=list(set(dataset.column_names) - set(['sentence', 'path']))
+    fn_kwargs={"do_lastspace_removing": True, "text_key_name": "cleaned_sentence"},
+    remove_columns=list(set(dataset.column_names) - set(['cleaned_sentence', 'path']))
 )
 
 dataset = dataset.map(speech_file_to_array_fn)
-result = dataset.map(predict)
+result = dataset.map(predict, batched=True, batch_size=8)
 
 wer = load_metric("wer")
 
-print("WER: {:.2f}".format(100 * wer.compute(predictions=result["predicted"], references=result["sentence"])))
+print("WER: {:.2f}".format(100 * wer.compute(predictions=result["predicted"], references=result["cleaned_sentence"])))
 ```
-]
+
 
 **Test Result**: 
-- WER: 12.00%
+- WER: 10.74%
 
 
 ## Training & Report