Add model files

README.md

@@ -0,0 +1,124 @@
+---
+language: eu
+datasets:
+- common_voice
+tags:
+- audio
+- automatic-speech-recognition
+- speech
+- xlsr-fine-tuning-week
+license: apache-2.0
+model-index:
+- name: XLSR Wav2Vec2 Basque Ivan G Torre
+  results:
+  - task: 
+      name: Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: Common Voice eu
+      type: common_voice
+      args: eu
+    metrics:
+       - name: Test WER
+         type: wer
+         value: 
+---
+
+# Wav2Vec2-Large-XLSR-53-euskera
+Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) in Euskera using the [Common Voice](https://huggingface.co/datasets/common_voice).
+When using this model, make sure that your speech input is sampled at 16kHz.
+
+## Usage
+
+The model can be used directly (without a language model) as follows:
+
+```python
+import torch
+import torchaudio
+from datasets import load_dataset
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+
+test_dataset = load_dataset("common_voice", "eu", split="test[:2%]").
+
+processor = Wav2Vec2Processor.from_pretrained("ivangtorre/wav2vec2-large-xlsr-53-basque")
+model = Wav2Vec2ForCTC.from_pretrained("ivangtorre/wav2vec2-large-xlsr-53-basque")
+
+resampler = torchaudio.transforms.Resample(48_000, 16_000)
+
+# Preprocessing the datasets.
+# We need to read the aduio 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
+
+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
+
+predicted_ids = torch.argmax(logits, dim=-1)
+
+print("Prediction:", processor.batch_decode(predicted_ids))
+print("Reference:", test_dataset["sentence"][:2])
+```
+
+
+## Evaluation
+
+The model can be evaluated as follows on the Basque test data of Common Voice.
+
+
+```python
+import torch
+import torchaudio
+from datasets import load_dataset, load_metric
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+import re
+
+test_dataset = load_dataset("common_voice", "eu", split="test")
+wer = load_metric("wer")
+
+processor = Wav2Vec2Processor.from_pretrained("ivangtorre/wav2vec2-large-xlsr-53-basque")
+model = Wav2Vec2ForCTC.from_pretrained("ivangtorre/wav2vec2-large-xlsr-53-basque")
+model.to("cuda")
+
+chars_to_ignore_regex = 'default=[",", "?", ".", "!", "-", ";", ":", '""', "%", "'", '"', "�"]'
+resampler = torchaudio.transforms.Resample(48_000, 16_000)
+
+# Preprocessing the datasets.
+# We need to read the aduio 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
+
+test_dataset = test_dataset.map(speech_file_to_array_fn)
+
+# Preprocessing the datasets.
+# We need to read the aduio files as arrays
+def evaluate(batch):
+  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
+
+    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**:
+
+
+## Training
+
+The Common Voice `train`, `validation` datasets were used for training.
+
+The script used for training can be found ???

config.json

@@ -0,0 +1,76 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "activation_dropout": 0.07,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.2,
+  "bos_token_id": 1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.2,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": true,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.04,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.05,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "mask_feature_length": 10,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.1,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_attention_heads": 16,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "pad_token_id": 30,
+  "transformers_version": "4.4.2",
+  "vocab_size": 31
+}

preprocessor_config.json

@@ -0,0 +1,8 @@
+{
+  "do_normalize": true,
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ab46384537a8ea6bbfd53bb2a4715eb721a0de52c3746a25aa3d5931eae1a37
+size 1262055847

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|"}

trainer_state.json

@@ -0,0 +1,240 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 18.95583596214511,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00011848341232227489,
+      "loss": 5.8456,
+      "step": 500
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 2.693875551223755,
+      "eval_runtime": 1054.431,
+      "eval_samples_per_second": 4.905,
+      "eval_wer": 1.0,
+      "step": 500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023696682464454977,
+      "loss": 0.5706,
+      "step": 1000
+    },
+    {
+      "epoch": 2.37,
+      "eval_loss": 0.19720512628555298,
+      "eval_runtime": 965.5246,
+      "eval_samples_per_second": 5.357,
+      "eval_wer": 0.4136704756426013,
+      "step": 1000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0003554502369668247,
+      "loss": 0.2135,
+      "step": 1500
+    },
+    {
+      "epoch": 3.55,
+      "eval_loss": 0.17894236743450165,
+      "eval_runtime": 829.9805,
+      "eval_samples_per_second": 6.231,
+      "eval_wer": 0.3831812357817117,
+      "step": 1500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00039178515007898897,
+      "loss": 0.1761,
+      "step": 2000
+    },
+    {
+      "epoch": 4.74,
+      "eval_loss": 0.1792241483926773,
+      "eval_runtime": 839.1507,
+      "eval_samples_per_second": 6.163,
+      "eval_wer": 0.3637789922338729,
+      "step": 2000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0003786203264876251,
+      "loss": 0.1403,
+      "step": 2500
+    },
+    {
+      "epoch": 5.92,
+      "eval_loss": 0.16667363047599792,
+      "eval_runtime": 773.0615,
+      "eval_samples_per_second": 6.69,
+      "eval_wer": 0.3549407734748843,
+      "step": 2500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00036545550289626125,
+      "loss": 0.1206,
+      "step": 3000
+    },
+    {
+      "epoch": 7.11,
+      "eval_loss": 0.16922971606254578,
+      "eval_runtime": 754.7522,
+      "eval_samples_per_second": 6.853,
+      "eval_wer": 0.35033862406191985,
+      "step": 3000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.00035229067930489733,
+      "loss": 0.1065,
+      "step": 3500
+    },
+    {
+      "epoch": 8.29,
+      "eval_loss": 0.17697516083717346,
+      "eval_runtime": 830.9983,
+      "eval_samples_per_second": 6.224,
+      "eval_wer": 0.34748842925502704,
+      "step": 3500
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.00033912585571353347,
+      "loss": 0.0949,
+      "step": 4000
+    },
+    {
+      "epoch": 9.48,
+      "eval_loss": 0.17623142898082733,
+      "eval_runtime": 808.622,
+      "eval_samples_per_second": 6.396,
+      "eval_wer": 0.34633789190178593,
+      "step": 4000
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.0003259610321221696,
+      "loss": 0.0834,
+      "step": 4500
+    },
+    {
+      "epoch": 10.66,
+      "eval_loss": 0.17101719975471497,
+      "eval_runtime": 805.7401,
+      "eval_samples_per_second": 6.419,
+      "eval_wer": 0.33004732892294014,
+      "step": 4500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.0003127962085308057,
+      "loss": 0.0781,
+      "step": 5000
+    },
+    {
+      "epoch": 11.85,
+      "eval_loss": 0.1730242371559143,
+      "eval_runtime": 801.4866,
+      "eval_samples_per_second": 6.453,
+      "eval_wer": 0.3337342781685537,
+      "step": 5000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.00029963138493944183,
+      "loss": 0.0702,
+      "step": 5500
+    },
+    {
+      "epoch": 13.03,
+      "eval_loss": 0.1908612698316574,
+      "eval_runtime": 817.2678,
+      "eval_samples_per_second": 6.328,
+      "eval_wer": 0.3248437622571451,
+      "step": 5500
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.00028646656134807797,
+      "loss": 0.065,
+      "step": 6000
+    },
+    {
+      "epoch": 14.22,
+      "eval_loss": 0.19133996963500977,
+      "eval_runtime": 815.1424,
+      "eval_samples_per_second": 6.345,
+      "eval_wer": 0.3376304160238475,
+      "step": 6000
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 0.00027330173775671406,
+      "loss": 0.0634,
+      "step": 6500
+    },
+    {
+      "epoch": 15.4,
+      "eval_loss": 0.17903825640678406,
+      "eval_runtime": 788.5559,
+      "eval_samples_per_second": 6.559,
+      "eval_wer": 0.32520984232408545,
+      "step": 6500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.0002601369141653502,
+      "loss": 0.0574,
+      "step": 7000
+    },
+    {
+      "epoch": 16.59,
+      "eval_loss": 0.2020258903503418,
+      "eval_runtime": 1572.1771,
+      "eval_samples_per_second": 3.29,
+      "eval_wer": 0.33182543210522186,
+      "step": 7000
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.00024697209057398633,
+      "loss": 0.0539,
+      "step": 7500
+    },
+    {
+      "epoch": 17.77,
+      "eval_loss": 0.18831460177898407,
+      "eval_runtime": 887.0107,
+      "eval_samples_per_second": 5.831,
+      "eval_wer": 0.3172083780038177,
+      "step": 7500
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.00023380726698262242,
+      "loss": 0.0518,
+      "step": 8000
+    },
+    {
+      "epoch": 18.96,
+      "eval_loss": 0.19699327647686005,
+      "eval_runtime": 927.4568,
+      "eval_samples_per_second": 5.577,
+      "eval_wer": 0.31524723478806577,
+      "step": 8000
+    }
+  ],
+  "max_steps": 16880,
+  "num_train_epochs": 40,
+  "total_flos": 3.909573475423295e+19,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3ff2eca68c59215249e3d45e21b893cb00ca2fcf912465cd2d8b55458c36c22
+size 2287

vocab.json

@@ -0,0 +1 @@
+{"q": 0, "b": 1, "r": 2, "h": 4, "j": 5, "f": 6, "a": 7, "u": 8, "x": 9, "p": 10, "w": 11, "z": 12, "k": 13, "e": 14, "l": 15, "y": 16, "t": 17, "g": 18, "c": 19, "d": 20, "o": 21, "n": 22, "ñ": 23, "v": 24, "m": 25, "s": 26, "i": 27, "|": 3, "[UNK]": 28, "[PAD]": 29}