Updated model

README.md

@@ -23,7 +23,7 @@ model-index:
     metrics:
        - name: Test WER
          type: wer
-         value: 22.05
+         value: 21.72
 ---
 
 # Wav2Vec2-Large-XLSR-53-Dutch
@@ -86,13 +86,13 @@ processor = Wav2Vec2Processor.from_pretrained("nithinholla/wav2vec2-large-xlsr-5
 model = Wav2Vec2ForCTC.from_pretrained("nithinholla/wav2vec2-large-xlsr-53-dutch")
 model.to("cuda")
 
-chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\'\�\(\)\&\–\—\—\…\´\’]'
+chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\'\�\(\)\&\–\—\=\…]'
 resampler = torchaudio.transforms.Resample(48_000, 16_000)
 
 # Preprocessing the datasets.
 # We need to read the audio files as arrays
 def speech_file_to_array_fn(batch):
-	batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
+	batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower().replace("´", "'").replace("’", "'")
 	speech_array, sampling_rate = torchaudio.load(batch["path"])
 	batch["speech"] = resampler(speech_array).squeeze().numpy()
 	return batch
@@ -116,7 +116,7 @@ result = test_dataset.map(evaluate, batched=True, batch_size=8)
 print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
 ```
 
-**Test Result**: 22.05 %
+**Test Result**: 21.72 %
 
 
 ## Training

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "_name_or_path": "/workspace/models/nl_third/checkpoint-13000",
   "activation_dropout": 0.1,
   "apply_spec_augment": true,
   "architectures": [
@@ -70,7 +70,7 @@
   "num_conv_pos_embeddings": 128,
   "num_feat_extract_layers": 7,
   "num_hidden_layers": 24,
-  "pad_token_id": 39,
+  "pad_token_id": 41,
   "transformers_version": "4.5.0.dev0",
-  "vocab_size": 40
+  "vocab_size": 42
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e7af24042f0f1585a49c5952219c54d10f5eedb1880d5b2a9a1a4d0492568c0e
-size 1262097815
+oid sha256:0d66e36e390922ffe5ab48bcd36d637e94d3058955e987c62e41a04a276d368e
+size 1262106007

vocab.json

@@ -1 +1 @@
-{"t": 0, "q": 1, "b": 2, "r": 3, "p": 4, "ë": 5, "h": 6, "d": 7, "a": 8, "é": 10, "g": 11, "f": 12, "ü": 13, "i": 14, "n": 15, "x": 16, "á": 17, "ê": 18, "l": 19, "à": 20, "s": 21, "ó": 22, "w": 23, "u": 24, "k": 25, "j": 26, "y": 27, "è": 28, "z": 29, "m": 30, "v": 31, "e": 32, "ú": 33, "o": 34, "ï": 35, "c": 36, "ö": 37, "|": 9, "[UNK]": 38, "[PAD]": 39}
+{"x": 0, "n": 1, "u": 2, "b": 3, "p": 5, "g": 6, "ï": 7, "o": 8, "i": 9, "c": 10, "m": 11, "a": 12, "l": 13, "t": 14, "ü": 15, "é": 16, "á": 17, "e": 18, "r": 19, "f": 20, "w": 21, "´": 22, "v": 23, "ö": 24, "z": 25, "y": 26, "d": 27, "ó": 28, "h": 29, "s": 30, "q": 31, "k": 32, "'": 33, "ê": 34, "à": 35, "ë": 36, "è": 37, "j": 38, "ú": 39, "|": 4, "[UNK]": 40, "[PAD]": 41}