Merge branch 'main' of https://huggingface.co/othrif/wav2vec2-large-xlsr-arabic into main

README.md

@@ -23,7 +23,7 @@ model-index:
     metrics:
        - name: Test WER
          type: wer
-         value: ?? # TODO
+         value: 46.77
 ---
 
 # Wav2Vec2-Large-XLSR-53-Arabic
@@ -51,15 +51,15 @@ 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):
-	speech_array, sampling_rate = torchaudio.load(batch["path"])
-	batch["speech"] = resampler(speech_array).squeeze().numpy()
-	return batch
+    speech_array, sampling_rate = torchaudio.load(batch["path"])
+    batch["speech"] = resampler(speech_array).squeeze().numpy()
+    return batch
 
 test_dataset = test_dataset.map(speech_file_to_array_fn)
 inputs = processor(test_dataset["speech"][:2], sampling_rate=16_000, return_tensors="pt", padding=True)
 
 with torch.no_grad():
-	logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
+    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
 
 predicted_ids = torch.argmax(logits, dim=-1)
 
@@ -87,41 +87,41 @@ processor = Wav2Vec2Processor.from_pretrained("othrif/wav2vec2-large-xlsr-arabic
 model = Wav2Vec2ForCTC.from_pretrained("othrif/wav2vec2-large-xlsr-arabic")
 model.to("cuda")
 
-chars_to_ignore_regex = '[\؛\—\_get\«\»\ـ\ـ\,\?\.\!\-\;\:\"\“\%\‘\”\�\#\،\☭,\؟]'
+chars_to_ignore_regex = '[\\\\؛\\\\—\\\\_get\\\\«\\\\»\\\\ـ\\\\ـ\\\\,\\\\?\\\\.\\\\!\\\\-\\\\;\\\\:\\\\"\\\\“\\\\%\\\\‘\\\\”\\\\�\\\\#\\\\،\\\\☭,\\\\؟]'
 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()
-	speech_array, sampling_rate = torchaudio.load(batch["path"])
-	batch["speech"] = resampler(speech_array).squeeze().numpy()
-	return batch
+    batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
+    speech_array, sampling_rate = torchaudio.load(batch["path"])
+    batch["speech"] = resampler(speech_array).squeeze().numpy()
+    return batch
 
 test_dataset = test_dataset.map(speech_file_to_array_fn)
 
 # Preprocessing the datasets.
 # We need to read the audio files as arrays
 def evaluate(batch):
-	inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
+    inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
 
-	with torch.no_grad():
-		logits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits
+    with torch.no_grad():
+    logits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits
 
-	pred_ids = torch.argmax(logits, dim=-1)
-	batch["pred_strings"] = processor.batch_decode(pred_ids)
-	return batch
+    pred_ids = torch.argmax(logits, dim=-1)
+    batch["pred_strings"] = processor.batch_decode(pred_ids)
+    return batch
 
 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**: XX.XX %  # TODO
+**Test Result**: 46.77
 
 
 ## Training
 
 The Common Voice `train`, `validation` datasets were used for training.
 
-The script used for training can be found [here](https://colab.research.google.com/drive/1ZaQQU4rHB3hIosPdWLNknhVZd4HAcJwV?usp=sharing) 
+The script used for training can be found [here](https://huggingface.co/othrif/wav2vec2-large-xlsr-arabic/tree/main)