初始化项目，由ModelHub XC社区提供模型

Model: alvanlii/wav2vec2-BERT-cantonese
Source: Original Platform

.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz 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

README.md

@@ -0,0 +1,74 @@
+---
+language:
+- zh
+license: apache-2.0
+datasets:
+- mozilla-foundation/common_voice_16_0
+model-index:
+- name: Wav2Vec2-BERT - Alvin
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: mozilla-foundation/common_voice_16_0 yue
+      type: mozilla-foundation/common_voice_16_0
+      config: yue
+      split: test
+      args: yue
+    metrics:
+    - name: CER
+      type: cer
+      value: 10.27
+---
+
+
+# Wav2Vec2-BERT - Alvin
+
+This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0). This has a CER of 10.27 on Common Voice 16 (yue) test set (without punctuations). 
+
+## Training and evaluation data
+For training, three datasets were used:
+- Common Voice 16 `zh-HK` and `yue` Train Set
+- CantoMap: Winterstein, Grégoire, Tang, Carmen and Lai, Regine (2020) "CantoMap: a Hong Kong Cantonese MapTask Corpus", in Proceedings of The 12th Language Resources and Evaluation Conference, Marseille: European Language Resources Association, p. 2899-2906.
+- Cantonse-ASR: Yu, Tiezheng, Frieske, Rita, Xu, Peng, Cahyawijaya, Samuel, Yiu, Cheuk Tung, Lovenia, Holy, Dai, Wenliang, Barezi, Elham, Chen, Qifeng, Ma, Xiaojuan, Shi, Bertram, Fung, Pascale (2022) "Automatic Speech Recognition Datasets in Cantonese: A Survey and New Dataset", 2022. Link: https://arxiv.org/pdf/2201.02419.pdf
+
+## Code Example
+```
+from transformers import pipeline
+bert_asr = pipeline(
+    "automatic-speech-recognition", model="alvanlii/wav2vec2-BERT-cantonese", device="cuda"
+)
+text = pipe(file)["text"]
+```
+or 
+```
+import torch
+import soundfile as sf
+from transformers import AutoModelForCTC, Wav2Vec2BertProcessor
+
+model_name = "alvanlii/wav2vec2-BERT-cantonese"
+
+asr_model = AutoModelForCTC.from_pretrained(model_name).to(device)
+processor = Wav2Vec2BertProcessor.from_pretrained(model_name)
+
+audio_input, _ = sf.read(file)
+
+inputs = processor([audio_input], sampling_rate=16_000).input_features
+features = torch.tensor(inputs)
+
+with torch.no_grad():
+  logits = asr_model(features).logits
+
+predicted_ids = torch.argmax(logits, dim=-1)
+predictions = processor.batch_decode(predicted_ids, skip_special_tokens=True)
+```
+
+## Training Hyperparameters
+- learning_rate: 5e-5
+- train_batch_size: 4 (on 1 3090)
+- eval_batch_size: 1
+- gradient_accumulation_steps: 32
+- total_train_batch_size: 32x4=128
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_warmup_steps: 1500

config.json

@@ -0,0 +1,82 @@
+{
+  "_name_or_path": "alvanlii/wav2vec2-BERT-cantonese",
+  "activation_dropout": 0.0,
+  "adapter_act": "relu",
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": true,
+  "apply_spec_augment": false,
+  "architectures": [
+    "Wav2Vec2BertForCTC"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 1,
+  "classifier_proj_size": 768,
+  "codevector_dim": 768,
+  "conformer_conv_dropout": 0.1,
+  "contrastive_logits_temperature": 0.1,
+  "conv_depthwise_kernel_size": 31,
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": true,
+  "diversity_loss_weight": 0.1,
+  "eos_token_id": 2,
+  "feat_proj_dropout": 0.1,
+  "feat_quantizer_dropout": 0.0,
+  "feature_projection_input_dim": 160,
+  "final_dropout": 0.1,
+  "hidden_act": "swish",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.1,
+  "left_max_position_embeddings": 64,
+  "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": 0.0,
+  "max_source_positions": 5000,
+  "model_type": "wav2vec2-bert",
+  "num_adapter_layers": 1,
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "output_hidden_size": 1024,
+  "pad_token_id": 2696,
+  "position_embeddings_type": "relative_key",
+  "proj_codevector_dim": 768,
+  "right_max_position_embeddings": 8,
+  "rotary_embedding_base": 10000,
+  "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": "4.38.2",
+  "use_intermediate_ffn_before_adapter": false,
+  "use_weighted_layer_sum": false,
+  "vocab_size": 2699,
+  "xvector_output_dim": 512
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:252cc61123d2c7b44d04cd7ae87d789e61e86f292d9339f05fabbffe0eec2813
+size 2433880476

optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a420558ee6dd550d0bf57ff4b79087225ff1a0db8441910732ed915ba6187fa
+size 2103705600

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "feature_extractor_type": "SeamlessM4TFeatureExtractor",
+  "feature_size": 80,
+  "num_mel_bins": 80,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000,
+  "stride": 2
+}

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d41fc46d0cba18a124097418b7f25a99868be3af2511a03c16b6472d5ff5a56
+size 14244

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5951bb0dce0845c2de727d3fe5cb2c3377d931684f23d2aea053d66c153fdad
+size 1064

special_tokens_map.json

@@ -0,0 +1,16 @@
+{
+  "pad_token": {
+    "content": "[PAD]",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "[UNK]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

trainer_state.json

@@ -0,0 +1,341 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8165192041489382,
+  "eval_steps": 300,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.04,
+      "grad_norm": 21.039730072021484,
+      "learning_rate": 7.4e-06,
+      "loss": 2.2676,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "eval_cer": 0.2868164937892176,
+      "eval_loss": 0.6633031964302063,
+      "eval_runtime": 320.6214,
+      "eval_samples_per_second": 8.05,
+      "eval_steps_per_second": 8.05,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.525612831115723,
+      "learning_rate": 1.49e-05,
+      "loss": 1.8817,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "eval_cer": 0.2927711614803432,
+      "eval_loss": 0.6335302591323853,
+      "eval_runtime": 119.6817,
+      "eval_samples_per_second": 21.566,
+      "eval_steps_per_second": 21.566,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.7516608238220215,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.7845,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "eval_cer": 0.2928031758227686,
+      "eval_loss": 0.5966914296150208,
+      "eval_runtime": 123.0968,
+      "eval_samples_per_second": 20.967,
+      "eval_steps_per_second": 20.967,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.020584583282471,
+      "learning_rate": 2.9875000000000004e-05,
+      "loss": 1.7503,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "eval_cer": 0.29213087463183507,
+      "eval_loss": 0.5989494919776917,
+      "eval_runtime": 125.3567,
+      "eval_samples_per_second": 20.589,
+      "eval_steps_per_second": 20.589,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7881321907043457,
+      "learning_rate": 3.737500000000001e-05,
+      "loss": 1.7549,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "eval_cer": 0.29901395825329746,
+      "eval_loss": 0.6132485866546631,
+      "eval_runtime": 132.3811,
+      "eval_samples_per_second": 19.497,
+      "eval_steps_per_second": 19.497,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.710826873779297,
+      "learning_rate": 4.4875e-05,
+      "loss": 1.7214,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24,
+      "eval_cer": 0.29782942758355746,
+      "eval_loss": 0.6118220686912537,
+      "eval_runtime": 130.5327,
+      "eval_samples_per_second": 19.773,
+      "eval_steps_per_second": 19.773,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.127965927124023,
+      "learning_rate": 4.911181750186986e-05,
+      "loss": 1.6742,
+      "step": 2100
+    },
+    {
+      "epoch": 0.29,
+      "eval_cer": 0.2963567678319887,
+      "eval_loss": 0.6121346354484558,
+      "eval_runtime": 131.7081,
+      "eval_samples_per_second": 19.596,
+      "eval_steps_per_second": 19.596,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.801788091659546,
+      "learning_rate": 4.630703066566941e-05,
+      "loss": 1.6562,
+      "step": 2400
+    },
+    {
+      "epoch": 0.33,
+      "eval_cer": 0.3039761813292355,
+      "eval_loss": 0.6422853469848633,
+      "eval_runtime": 132.5985,
+      "eval_samples_per_second": 19.465,
+      "eval_steps_per_second": 19.465,
+      "step": 2400
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.572460174560547,
+      "learning_rate": 4.350224382946896e-05,
+      "loss": 1.6373,
+      "step": 2700
+    },
+    {
+      "epoch": 0.37,
+      "eval_cer": 0.2989499295684467,
+      "eval_loss": 0.639430046081543,
+      "eval_runtime": 133.5523,
+      "eval_samples_per_second": 19.326,
+      "eval_steps_per_second": 19.326,
+      "step": 2700
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.840504169464111,
+      "learning_rate": 4.069745699326851e-05,
+      "loss": 1.5944,
+      "step": 3000
+    },
+    {
+      "epoch": 0.41,
+      "eval_cer": 0.2967089255986682,
+      "eval_loss": 0.6272587776184082,
+      "eval_runtime": 134.0186,
+      "eval_samples_per_second": 19.259,
+      "eval_steps_per_second": 19.259,
+      "step": 3000
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.972968816757202,
+      "learning_rate": 3.7892670157068066e-05,
+      "loss": 1.5654,
+      "step": 3300
+    },
+    {
+      "epoch": 0.45,
+      "eval_cer": 0.2903700857984377,
+      "eval_loss": 0.5716381669044495,
+      "eval_runtime": 134.8224,
+      "eval_samples_per_second": 19.144,
+      "eval_steps_per_second": 19.144,
+      "step": 3300
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.683291435241699,
+      "learning_rate": 3.5087883320867614e-05,
+      "loss": 1.5333,
+      "step": 3600
+    },
+    {
+      "epoch": 0.49,
+      "eval_cer": 0.29011397105903447,
+      "eval_loss": 0.573725163936615,
+      "eval_runtime": 134.5796,
+      "eval_samples_per_second": 19.178,
+      "eval_steps_per_second": 19.178,
+      "step": 3600
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 8.39251708984375,
+      "learning_rate": 3.228309648466717e-05,
+      "loss": 1.5252,
+      "step": 3900
+    },
+    {
+      "epoch": 0.53,
+      "eval_cer": 0.29501216545012166,
+      "eval_loss": 0.5687663555145264,
+      "eval_runtime": 135.3645,
+      "eval_samples_per_second": 19.067,
+      "eval_steps_per_second": 19.067,
+      "step": 3900
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.761993169784546,
+      "learning_rate": 2.9478309648466717e-05,
+      "loss": 1.5017,
+      "step": 4200
+    },
+    {
+      "epoch": 0.57,
+      "eval_cer": 0.2957164809834806,
+      "eval_loss": 0.5565311312675476,
+      "eval_runtime": 135.2285,
+      "eval_samples_per_second": 19.086,
+      "eval_steps_per_second": 19.086,
+      "step": 4200
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 23.93025016784668,
+      "learning_rate": 2.667352281226627e-05,
+      "loss": 1.4707,
+      "step": 4500
+    },
+    {
+      "epoch": 0.61,
+      "eval_cer": 0.28828915354078627,
+      "eval_loss": 0.5579658150672913,
+      "eval_runtime": 135.1123,
+      "eval_samples_per_second": 19.103,
+      "eval_steps_per_second": 19.103,
+      "step": 4500
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7994115352630615,
+      "learning_rate": 2.386873597606582e-05,
+      "loss": 1.4578,
+      "step": 4800
+    },
+    {
+      "epoch": 0.65,
+      "eval_cer": 0.2855359200922013,
+      "eval_loss": 0.5352594256401062,
+      "eval_runtime": 135.7058,
+      "eval_samples_per_second": 19.019,
+      "eval_steps_per_second": 19.019,
+      "step": 4800
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9403347969055176,
+      "learning_rate": 2.1073298429319373e-05,
+      "loss": 1.4236,
+      "step": 5100
+    },
+    {
+      "epoch": 0.69,
+      "eval_cer": 0.28758483800742735,
+      "eval_loss": 0.5495265126228333,
+      "eval_runtime": 136.3492,
+      "eval_samples_per_second": 18.929,
+      "eval_steps_per_second": 18.929,
+      "step": 5100
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.021031379699707,
+      "learning_rate": 1.8268511593118924e-05,
+      "loss": 1.3954,
+      "step": 5400
+    },
+    {
+      "epoch": 0.73,
+      "eval_cer": 0.2876808810347036,
+      "eval_loss": 0.5390140414237976,
+      "eval_runtime": 138.4653,
+      "eval_samples_per_second": 18.64,
+      "eval_steps_per_second": 18.64,
+      "step": 5400
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 9.61117172241211,
+      "learning_rate": 1.5463724756918475e-05,
+      "loss": 1.3947,
+      "step": 5700
+    },
+    {
+      "epoch": 0.78,
+      "eval_cer": 0.28633627865283645,
+      "eval_loss": 0.5187196731567383,
+      "eval_runtime": 136.0715,
+      "eval_samples_per_second": 18.968,
+      "eval_steps_per_second": 18.968,
+      "step": 5700
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.387241840362549,
+      "learning_rate": 1.2658937920718025e-05,
+      "loss": 1.3678,
+      "step": 6000
+    },
+    {
+      "epoch": 0.82,
+      "eval_cer": 0.27993341016775514,
+      "eval_loss": 0.5071456432342529,
+      "eval_runtime": 136.2594,
+      "eval_samples_per_second": 18.942,
+      "eval_steps_per_second": 18.942,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 300,
+  "max_steps": 7348,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 600,
+  "total_flos": 1.8710735171848015e+20,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55d26536955cba130ab0ce5e51b836904166383e2bf233c2c401fde5e5113deb
+size 4856