Update README.md

README.md

@@ -1,105 +1,47 @@
 ---
+language:
+- da
 license: apache-2.0
-tags:
-- generated_from_trainer
+tasks:
+- automatic-speech-recognition
 datasets:
-- common_voice
+- common_voice_8_0
+metrics:
+- wer
 model-index:
 - name: wav2vec2-xls-r-300m-cv8-da
-  results: []
+  results:
+  - task: 
+      type: automatic-speech-recognition
+    dataset:
+      type: Alvenir/alvenir_asr_da_eval
+      name: Alvenir ASR test dataset
+    metrics:
+      - type: wer
+        value: 25.80
+  - task: 
+      type: automatic-speech-recognition
+    dataset:
+      type: mozilla-foundation/common_voice_8_0
+      args: da
+      name: Danish Common Voice 8.0
+    metrics:
+      - type: wer
+        value: 26.45
 ---
 
-<!-- This model card has been generated automatically according to the information the Trainer had access to. You
-should probably proofread and complete it, then remove this comment. -->
-
-# wav2vec2-xls-r-300m-cv8-da
-
-This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice dataset.
-It achieves the following results on the evaluation set:
-- Loss: 287.3684
-- Wer: 0.2978
+# XLS-R-300m-CV8-da
 
 ## Model description
 
-More information needed
-
-## Intended uses & limitations
-
-More information needed
-
-## Training and evaluation data
-
-More information needed
-
-## Training procedure
-
-### Training hyperparameters
-
-The following hyperparameters were used during training:
-- learning_rate: 4e-05
-- train_batch_size: 4
-- eval_batch_size: 8
-- seed: 4242
-- gradient_accumulation_steps: 8
-- total_train_batch_size: 32
-- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
-- lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 500
-- num_epochs: 500
-- mixed_precision_training: Native AMP
-
-### Training results
+This model is a fine-tuned version of the multilingual acoustic model [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the Danish part of [Common Voice 8.0](https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0), containing ~6 crowdsourced hours of read-aloud Danish speech.
 
-| Training Loss | Epoch  | Step  | Validation Loss | Wer    |
-|:-------------:|:------:|:-----:|:---------------:|:------:|
-| 736.2597      | 5.55   | 300   | 1418.4758       | 1.0    |
-| 577.3605      | 11.11  | 600   | 1267.1588       | 1.0    |
-| 476.434       | 16.66  | 900   | 842.5760        | 1.0028 |
-| 215.5165      | 22.22  | 1200  | 356.9064        | 0.6405 |
-| 138.4598      | 27.77  | 1500  | 256.0349        | 0.4889 |
-| 104.3224      | 33.33  | 1800  | 228.9147        | 0.4435 |
-| 82.2922       | 38.88  | 2100  | 208.3896        | 0.4134 |
-| 68.0471       | 44.44  | 2400  | 205.3142        | 0.3976 |
-| 57.6613       | 49.99  | 2700  | 208.4677        | 0.3935 |
-| 52.6747       | 55.55  | 3000  | 213.9339        | 0.3928 |
-| 43.4302       | 61.11  | 3300  | 212.9035        | 0.3808 |
-| 40.6623       | 66.66  | 3600  | 221.2281        | 0.3850 |
-| 34.9752       | 72.22  | 3900  | 224.3349        | 0.3795 |
-| 33.6799       | 77.77  | 4200  | 218.9585        | 0.3696 |
-| 31.5797       | 83.33  | 4500  | 211.9424        | 0.3713 |
-| 28.2278       | 88.88  | 4800  | 226.6527        | 0.3705 |
-| 26.6871       | 94.44  | 5100  | 227.9236        | 0.3689 |
-| 25.8799       | 99.99  | 5400  | 227.0796        | 0.3653 |
-| 25.0987       | 105.55 | 5700  | 231.7458        | 0.3640 |
-| 22.328        | 111.11 | 6000  | 236.6981        | 0.3644 |
-| 21.8799       | 116.66 | 6300  | 219.9040        | 0.3573 |
-| 20.1069       | 122.22 | 6600  | 230.2603        | 0.3592 |
-| 20.8641       | 127.77 | 6900  | 248.2307        | 0.3638 |
-| 18.6306       | 133.33 | 7200  | 239.0962        | 0.3583 |
-| 18.2617       | 138.88 | 7500  | 245.4144        | 0.3549 |
-| 17.5175       | 144.44 | 7800  | 258.2551        | 0.3617 |
-| 56.1866       | 149.99 | 8100  | 278.0422        | 0.3384 |
-| 28.7438       | 155.55 | 8400  | 271.7619        | 0.3107 |
-| 23.462        | 161.11 | 8700  | 279.4823        | 0.3091 |
-| 20.5816       | 166.66 | 9000  | 274.7589        | 0.3070 |
-| 18.7921       | 172.22 | 9300  | 265.1793        | 0.3024 |
-| 16.5106       | 177.77 | 9600  | 269.0261        | 0.2977 |
-| 17.1239       | 183.33 | 9900  | 261.8192        | 0.2956 |
-| 15.4832       | 188.88 | 10200 | 271.2451        | 0.2982 |
-| 16.2146       | 194.44 | 10500 | 269.5397        | 0.2957 |
-| 13.7332       | 199.99 | 10800 | 276.0069        | 0.3017 |
-| 13.914        | 205.55 | 11100 | 278.2491        | 0.3023 |
-| 13.7551       | 211.11 | 11400 | 260.8341        | 0.2900 |
-| 11.6575       | 216.66 | 11700 | 276.8361        | 0.2906 |
-| 12.4438       | 222.22 | 12000 | 280.0102        | 0.2954 |
-| 13.1577       | 227.77 | 12300 | 268.7986        | 0.2949 |
-| 12.2083       | 233.33 | 12600 | 272.5989        | 0.2949 |
-| 11.1147       | 238.88 | 12900 | 287.3684        | 0.2978 |
 
+## Performance
 
-### Framework versions
+The model achieves the following WER scores (lower is better):
 
-- Transformers 4.16.2
-- Pytorch 1.8.1+cu101
-- Datasets 1.18.3
-- Tokenizers 0.11.0
+| **Dataset** | **WER without LM** | **WER with 5-gram LM** |
+| :---:   | ---: | ---: |
+| [Danish part of Common Voice 8.0](https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0/viewer/da/train) | 31.33 | 26.45 |
+| [Alvenir test set](https://huggingface.co/datasets/Alvenir/alvenir_asr_da_eval) | 30.54 | 25.80 |