Upload wav2vec2-base-vi-vlsp2020

.gitattributes

@@ -32,3 +32,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+quangnam.wav filter=lfs diff=lfs merge=lfs -text
+t2_0000006682.wav filter=lfs diff=lfs merge=lfs -text

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "</s>": 97,
+  "<s>": 96
+}

alphabet.json

@@ -0,0 +1 @@
+{"labels": [" ", "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z", "\u00e0", "\u00e1", "\u00e2", "\u00e3", "\u00e8", "\u00e9", "\u00ea", "\u00ec", "\u00ed", "\u00f2", "\u00f3", "\u00f4", "\u00f5", "\u00f9", "\u00fa", "\u00fd", "\u0103", "\u0111", "\u0129", "\u0169", "\u01a1", "\u01b0", "\u1ea1", "\u1ea3", "\u1ea5", "\u1ea7", "\u1ea9", "\u1eab", "\u1ead", "\u1eaf", "\u1eb1", "\u1eb3", "\u1eb5", "\u1eb7", "\u1eb9", "\u1ebb", "\u1ebd", "\u1ebf", "\u1ec1", "\u1ec3", "\u1ec5", "\u1ec7", "\u1ec9", "\u1ecb", "\u1ecd", "\u1ecf", "\u1ed1", "\u1ed3", "\u1ed5", "\u1ed7", "\u1ed9", "\u1edb", "\u1edd", "\u1edf", "\u1ee1", "\u1ee3", "\u1ee5", "\u1ee7", "\u1ee9", "\u1eeb", "\u1eed", "\u1eef", "\u1ef1", "\u1ef3", "\u1ef5", "\u1ef7", "\u1ef9", "\u2047", "", "<s>", "</s>"], "is_bpe": false}

config.json

@@ -0,0 +1,108 @@
+{
+  "_name_or_path": "./wav2vec2-base-vlsp2020/checkpoint-90000",
+  "activation_dropout": 0.0,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 256,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": false,
+  "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,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": false,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "group",
+  "feat_proj_dropout": 0.0,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.0,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.3,
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.1,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "model_type": "wav2vec2",
+  "num_adapter_layers": 3,
+  "num_attention_heads": 12,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 12,
+  "num_negatives": 100,
+  "output_hidden_size": 768,
+  "pad_token_id": 95,
+  "proj_codevector_dim": 256,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.23.1",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 98,
+  "xvector_output_dim": 512
+}

language_model/attrs.json

@@ -0,0 +1 @@
+{"alpha": 0.5, "beta": 1.5, "unk_score_offset": -10.0, "score_boundary": true}

language_model/vi_lm_5grams.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2e3e3264950427d249876ac3dd24c0f506b06f1cca3e2c4095e0f95e50b3133
+size 2471072676

model_handling.py

@@ -0,0 +1,163 @@
+from transformers import Wav2Vec2PreTrainedModel, Wav2Vec2Model
+from torch import nn
+import warnings
+import torch
+from transformers.modeling_outputs import CausalLMOutput
+from collections import OrderedDict
+
+_HIDDEN_STATES_START_POSITION = 2
+
+
+class Wav2Vec2ForCTC(Wav2Vec2PreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+
+        self.wav2vec2 = Wav2Vec2Model(config)
+        self.dropout = nn.Dropout(config.final_dropout)
+
+        self.feature_transform = nn.Sequential(OrderedDict([
+            ('linear1', nn.Linear(config.hidden_size, config.hidden_size)),
+            ('bn1', nn.BatchNorm1d(config.hidden_size)),
+            ('activation1', nn.LeakyReLU()),
+            ('drop1', nn.Dropout(config.final_dropout)),
+            ('linear2', nn.Linear(config.hidden_size, config.hidden_size)),
+            ('bn2', nn.BatchNorm1d(config.hidden_size)),
+            ('activation2', nn.LeakyReLU()),
+            ('drop2', nn.Dropout(config.final_dropout)),
+            ('linear3', nn.Linear(config.hidden_size, config.hidden_size)),
+            ('bn3', nn.BatchNorm1d(config.hidden_size)),
+            ('activation3', nn.LeakyReLU()),
+            ('drop3', nn.Dropout(config.final_dropout))
+        ]))
+
+        if config.vocab_size is None:
+            raise ValueError(
+                f"You are trying to instantiate {self.__class__} with a configuration that "
+                "does not define the vocabulary size of the language model head. Please "
+                "instantiate the model as follows: `Wav2Vec2ForCTC.from_pretrained(..., vocab_size=vocab_size)`. "
+                "or define `vocab_size` of your model's configuration."
+            )
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size)
+
+        self.is_wav2vec_freeze = False
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def freeze_feature_extractor(self):
+        """
+        Calling this function will disable the gradient computation for the feature encoder so that its parameter will
+        not be updated during training.
+        """
+        warnings.warn(
+            "The method `freeze_feature_extractor` is deprecated and will be removed in Transformers v5."
+            "Please use the equivalent `freeze_feature_encoder` method instead.",
+            FutureWarning,
+        )
+        self.freeze_feature_encoder()
+
+    def freeze_feature_encoder(self):
+        """
+        Calling this function will disable the gradient computation for the feature encoder so that its parameter will
+        not be updated during training.
+        """
+        self.wav2vec2.feature_extractor._freeze_parameters()
+
+    def freeze_wav2vec(self, is_freeze=True):
+        """
+        Calling this function will disable the gradient computation for the feature extractor so that its parameter
+        will not be updated during training.
+        """
+        if is_freeze:
+            self.is_wav2vec_freeze = True
+            for param in self.wav2vec2.parameters():
+                param.requires_grad = False
+        else:
+            self.is_wav2vec_freeze = False
+            for param in self.wav2vec2.parameters():
+                param.requires_grad = True
+        self.freeze_feature_encoder()
+
+        model_total_params = sum(p.numel() for p in self.parameters())
+        model_total_params_trainable = sum(p.numel() for p in self.parameters() if p.requires_grad)
+        print("model_total_params: {}\nmodel_total_params_trainable: {}".format(model_total_params,
+                                                                                model_total_params_trainable))
+
+    def forward(
+            self,
+            input_values,
+            attention_mask=None,
+            output_attentions=None,
+            output_hidden_states=None,
+            return_dict=None,
+            labels=None,
+    ):
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size, target_length)`, *optional*):
+            Labels for connectionist temporal classification. Note that `target_length` has to be smaller or equal to
+            the sequence length of the output logits. Indices are selected in `[-100, 0, ..., config.vocab_size - 1]`.
+            All labels set to `-100` are ignored (masked), the loss is only computed for labels in `[0, ...,
+            config.vocab_size - 1]`.
+        """
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        outputs = self.wav2vec2(
+            input_values,
+            attention_mask=attention_mask,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = outputs[0]
+        hidden_states = self.dropout(hidden_states)
+
+        B, T, F = hidden_states.size()
+        hidden_states = hidden_states.view(B * T, F)
+
+        hidden_states = self.feature_transform(hidden_states)
+
+        hidden_states = hidden_states.view(B, T, F)
+
+        logits = self.lm_head(hidden_states)
+
+        loss = None
+        if labels is not None:
+
+            if labels.max() >= self.config.vocab_size:
+                raise ValueError(f"Label values must be <= vocab_size: {self.config.vocab_size}")
+
+            # retrieve loss input_lengths from attention_mask
+            attention_mask = (
+                attention_mask if attention_mask is not None else torch.ones_like(input_values, dtype=torch.long)
+            )
+            input_lengths = self._get_feat_extract_output_lengths(attention_mask.sum(-1)).to(torch.long)
+
+            # assuming that padded tokens are filled with -100
+            # when not being attended to
+            labels_mask = labels >= 0
+            target_lengths = labels_mask.sum(-1)
+            flattened_targets = labels.masked_select(labels_mask)
+
+            # ctc_loss doesn't support fp16
+            log_probs = nn.functional.log_softmax(logits, dim=-1, dtype=torch.float32).transpose(0, 1)
+
+            with torch.backends.cudnn.flags(enabled=False):
+                loss = nn.functional.ctc_loss(
+                    log_probs,
+                    flattened_targets,
+                    input_lengths,
+                    target_lengths,
+                    blank=self.config.pad_token_id,
+                    reduction=self.config.ctc_loss_reduction,
+                    zero_infinity=self.config.ctc_zero_infinity,
+                )
+
+        if not return_dict:
+            output = (logits,) + outputs[_HIDDEN_STATES_START_POSITION:]
+            return ((loss,) + output) if loss is not None else output
+
+        return CausalLMOutput(
+            loss=loss, logits=logits, hidden_states=outputs.hidden_states, attentions=outputs.attentions
+        )

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "processor_class": "Wav2Vec2ProcessorWithLM",
+  "return_attention_mask": false,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e85ff35346e235b16fbe3821091323c377113828e8fdffd1c151c928035530b
+size 384987309

quangnam.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06a65d90eaef5d174751693b4ae0383c472ff95267713fb01c08802c56b143e2
+size 159404

special_tokens_map.json

@@ -0,0 +1,134 @@
+{
+  "additional_special_tokens": [
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    }
+  ],
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

t2_0000006682.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e6821f71b608ec9fa947571af7f631cad1f9d45b22287e22e70e7411c62b802
+size 49644

tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token": "<s>",
+  "do_lower_case": false,
+  "eos_token": "</s>",
+  "name_or_path": "../lyric_alignment/model-bin/",
+  "pad_token": "<pad>",
+  "processor_class": "Wav2Vec2ProcessorWithLM",
+  "replace_word_delimiter_char": " ",
+  "special_tokens_map_file": null,
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "tokenizer_file": null,
+  "unk_token": "<unk>",
+  "word_delimiter_token": "|"
+}

vocab.json

@@ -0,0 +1,98 @@
+{
+  "<pad>": 95,
+  "<unk>": 94,
+  "a": 1,
+  "b": 2,
+  "c": 3,
+  "d": 4,
+  "e": 5,
+  "f": 6,
+  "g": 7,
+  "h": 8,
+  "i": 9,
+  "j": 10,
+  "k": 11,
+  "l": 12,
+  "m": 13,
+  "n": 14,
+  "o": 15,
+  "p": 16,
+  "q": 17,
+  "r": 18,
+  "s": 19,
+  "t": 20,
+  "u": 21,
+  "v": 22,
+  "w": 23,
+  "x": 24,
+  "y": 25,
+  "z": 26,
+  "|": 0,
+  "à": 27,
+  "á": 28,
+  "â": 29,
+  "ã": 30,
+  "è": 31,
+  "é": 32,
+  "ê": 33,
+  "ì": 34,
+  "í": 35,
+  "ò": 36,
+  "ó": 37,
+  "ô": 38,
+  "õ": 39,
+  "ù": 40,
+  "ú": 41,
+  "ý": 42,
+  "ă": 43,
+  "đ": 44,
+  "ĩ": 45,
+  "ũ": 46,
+  "ơ": 47,
+  "ư": 48,
+  "ạ": 49,
+  "ả": 50,
+  "ấ": 51,
+  "ầ": 52,
+  "ẩ": 53,
+  "ẫ": 54,
+  "ậ": 55,
+  "ắ": 56,
+  "ằ": 57,
+  "ẳ": 58,
+  "ẵ": 59,
+  "ặ": 60,
+  "ẹ": 61,
+  "ẻ": 62,
+  "ẽ": 63,
+  "ế": 64,
+  "ề": 65,
+  "ể": 66,
+  "ễ": 67,
+  "ệ": 68,
+  "ỉ": 69,
+  "ị": 70,
+  "ọ": 71,
+  "ỏ": 72,
+  "ố": 73,
+  "ồ": 74,
+  "ổ": 75,
+  "ỗ": 76,
+  "ộ": 77,
+  "ớ": 78,
+  "ờ": 79,
+  "ở": 80,
+  "ỡ": 81,
+  "ợ": 82,
+  "ụ": 83,
+  "ủ": 84,
+  "ứ": 85,
+  "ừ": 86,
+  "ử": 87,
+  "ữ": 88,
+  "ự": 89,
+  "ỳ": 90,
+  "ỵ": 91,
+  "ỷ": 92,
+  "ỹ": 93
+}