jonatasgrosman commited on
Commit
08a76a0
1 Parent(s): e4ab2bd

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -7
README.md CHANGED
@@ -133,14 +133,14 @@ test_dataset = test_dataset.map(speech_file_to_array_fn)
133
  # Preprocessing the datasets.
134
  # We need to read the audio files as arrays
135
  def evaluate(batch):
136
- \tinputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
137
 
138
- \twith torch.no_grad():
139
- \t\tlogits = model(inputs.input_values.to(DEVICE), attention_mask=inputs.attention_mask.to(DEVICE)).logits
140
 
141
- \tpred_ids = torch.argmax(logits, dim=-1)
142
- \tbatch["pred_strings"] = processor.batch_decode(pred_ids)
143
- \treturn batch
144
 
145
  result = test_dataset.map(evaluate, batched=True, batch_size=8)
146
 
@@ -153,7 +153,7 @@ print(f"CER: {cer.compute(predictions=predictions, references=references, chunk_
153
 
154
  **Test Result**:
155
 
156
- My model may report better scores than others because of some specificity of my evaluation script, so I ran the same evaluation script on other models (on 2021-04-22) to make a fairer comparison.
157
 
158
  | Model | WER | CER |
159
  | ------------- | ------------- | ------------- |
133
  # Preprocessing the datasets.
134
  # We need to read the audio files as arrays
135
  def evaluate(batch):
136
+ \\tinputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
137
 
138
+ \\twith torch.no_grad():
139
+ \\t\\tlogits = model(inputs.input_values.to(DEVICE), attention_mask=inputs.attention_mask.to(DEVICE)).logits
140
 
141
+ \\tpred_ids = torch.argmax(logits, dim=-1)
142
+ \\tbatch["pred_strings"] = processor.batch_decode(pred_ids)
143
+ \\treturn batch
144
 
145
  result = test_dataset.map(evaluate, batched=True, batch_size=8)
146
 
153
 
154
  **Test Result**:
155
 
156
+ In the table below I report the Word Error Rate (WER) and the Character Error Rate (CER) of the model. I ran the evaluation script described above on other models as well (on 2021-04-22). Note that the table below may show different results from those already reported, this may have been caused due to some specificity of the other evaluation scripts used.
157
 
158
  | Model | WER | CER |
159
  | ------------- | ------------- | ------------- |