MMS-1B Wolof fine-tuned - WER 29.56% (full fine-tune, 20k samples)

README.md

@@ -0,0 +1,161 @@
+---
+language: wo
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- audio
+- wolof
+- mms
+- senegal
+- africa
+- low-resource
+datasets:
+- vonewman/wolof-audio-data
+metrics:
+- wer
+base_model: facebook/mms-1b-all
+model-index:
+- name: mms-1b-wolof-finetuned
+  results:
+  - task:
+      type: automatic-speech-recognition
+      name: Automatic Speech Recognition
+    dataset:
+      name: vonewman/wolof-audio-data
+      type: vonewman/wolof-audio-data
+    metrics:
+    - type: wer
+      value: 29.56
+      name: Word Error Rate
+---
+
+# MMS-1B Wolof Fine-Tuned 🇸🇳
+
+Fine-tuned version of [facebook/mms-1b-all](https://huggingface.co/facebook/mms-1b-all) for **Wolof Automatic Speech Recognition (ASR)**.
+
+This model transcribes Wolof speech audio into Wolof text. It's the result of full fine-tuning (all 964M parameters) on 20,000 quality-filtered audio samples.
+
+## 📊 Performance
+
+| Metric | Value |
+|--------|-------|
+| **WER (Word Error Rate)** | **29.56%** |
+| Test set | 500 examples |
+| Training samples | 20,000 |
+| Total audio | 25.6 hours |
+| Mode | Full fine-tuning (964M params) |
+
+## 🎯 Use Cases
+
+- 🎙️ Transcription of Wolof audio (radio, podcasts, conversations)
+- 🌍 First step in a Wolof → French translation pipeline
+- ♿ Accessibility tools for Wolof-speaking communities
+- 🔬 Research on low-resource African languages
+
+## 🚀 Quick Start
+
+```python
+from transformers import Wav2Vec2ForCTC, AutoProcessor
+import torch
+import torchaudio
+
+# Load model and processor
+model_id = "Sadou/mms-1b-wolof-finetuned"
+processor = AutoProcessor.from_pretrained(model_id)
+model = Wav2Vec2ForCTC.from_pretrained(model_id)
+
+# Load audio (must be 16kHz, mono)
+waveform, sr = torchaudio.load("your_audio.wav")
+if sr != 16000:
+    waveform = torchaudio.transforms.Resample(sr, 16000)(waveform)
+if waveform.shape[0] > 1:
+    waveform = waveform.mean(dim=0, keepdim=True)
+
+# Transcribe
+inputs = processor(waveform[0].numpy(), sampling_rate=16000, return_tensors="pt")
+
+with torch.no_grad():
+    logits = model(**inputs).logits
+
+predicted_ids = torch.argmax(logits, dim=-1)
+transcription = processor.batch_decode(predicted_ids)[0]
+print(f"Transcription: {transcription}")
+```
+
+## 📈 Training Details
+
+### Dataset
+- **Source**: [vonewman/wolof-audio-data](https://huggingface.co/datasets/vonewman/wolof-audio-data)
+- **Combines**: ALFFA + FLEURS + Urban Bus + Kallama
+- **Quality filters applied**:
+  - Audio duration: 2-15s
+  - Text: 3-30 words
+  - Speech rate: 1.0-4.5 words/sec
+  - Retention rate: 75.5%
+
+### Training Configuration
+- **Base model**: `facebook/mms-1b-all`
+- **Mode**: Full fine-tuning (all 964M parameters)
+- **Precision**: bf16 (Brain Float 16)
+- **Effective batch size**: 32 (16 × 2 gradient accumulation)
+- **Gradient checkpointing**: Enabled
+- **Learning rate**: 3e-5 (initial), then 5e-6 (final epochs)
+- **Epochs**: 6 (4 + 2 with LR decay)
+- **Hardware**: NVIDIA RTX PRO 6000 Blackwell
+
+### Training Progression
+
+| Step | WER | Phase |
+|------|-----|-------|
+| 300 | 47.67% | Initial training |
+| 900 | 38.86% | |
+| 1500 | 33.35% | |
+| 2100 | 30.77% | |
+| 2400 | 30.11% | Plateau detected |
+| 2700 | 30.56% | LR decay started |
+| 3300 | 29.56% | |
+| 3600 | **29.56%** | Final ✅ |
+
+## ⚠️ Limitations
+
+- **Code-switching**: Difficulty with French words mixed in Wolof speech
+- **Word segmentation**: Sometimes merges or splits words incorrectly
+- **Background noise**: Performance degrades on noisy audio
+- **Long audios**: For audios > 30s, use chunking with stride
+
+## 🔄 Long Audio Inference
+
+For audios longer than 30 seconds, use HuggingFace pipeline with chunking:
+
+```python
+from transformers import pipeline
+
+pipe = pipeline(
+    'automatic-speech-recognition',
+    model="Sadou/mms-1b-wolof-finetuned",
+    chunk_length_s=30,
+    stride_length_s=(4, 2),
+    device=0,
+)
+
+result = pipe("long_audio.wav")
+print(result['text'])
+```
+
+## 🙏 Acknowledgments
+
+- Meta AI for [MMS](https://huggingface.co/facebook/mms-1b-all)
+- [vonewman](https://huggingface.co/vonewman) for the Wolof audio dataset
+- [GalsenAI](https://huggingface.co/galsenai) and the Senegalese AI community
+
+## 📚 Citation
+
+```bibtex
+@misc{wolof-mms-2026,
+  author = {Sadou Barry},
+  title = {MMS-1B Wolof Fine-Tuned},
+  year = {2026},
+  publisher = {HuggingFace},
+  url = {https://huggingface.co/Sadou/mms-1b-wolof-finetuned}
+}
+```

config.json

@@ -0,0 +1,108 @@
+{
+  "activation_dropout": 0.05,
+  "adapter_attn_dim": 16,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.05,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 1024,
+  "contrastive_logits_temperature": 0.1,
+  "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,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "dtype": "float32",
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.05,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.05,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.05,
+  "hidden_size": 1280,
+  "initializer_range": 0.02,
+  "intermediate_size": 5120,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.05,
+  "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": 16,
+  "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": 48,
+  "num_negatives": 100,
+  "output_hidden_size": 1280,
+  "pad_token_id": 0,
+  "proj_codevector_dim": 1024,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "transformers_version": "5.0.0",
+  "use_cache": false,
+  "use_weighted_layer_sum": false,
+  "vocab_size": 87,
+  "xvector_output_dim": 512
+}

metrics.json

@@ -0,0 +1,12 @@
+{
+  "final_wer": 0.2956,
+  "train_size": 20000,
+  "test_size": 500,
+  "epochs": 6,
+  "mode": "full_fine_tune",
+  "lr_initial": 3e-05,
+  "lr_final": 5e-06,
+  "base_model": "facebook/mms-1b-all",
+  "language": "wol",
+  "dataset": "vonewman/wolof-audio-data"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11559aacad0861f7fd06e1ebe56bdb7d0ed5e522b0935716fd86aaddb9b452a7
+size 3859177812

processor_config.json

@@ -0,0 +1,12 @@
+{
+  "feature_extractor": {
+    "do_normalize": true,
+    "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+    "feature_size": 1,
+    "padding_side": "right",
+    "padding_value": 0,
+    "return_attention_mask": true,
+    "sampling_rate": 16000
+  },
+  "processor_class": "Wav2Vec2Processor"
+}

tokenizer_config.json

@@ -0,0 +1,54 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": null,
+  "backend": "custom",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": true,
+  "do_lower_case": false,
+  "eos_token": "</s>",
+  "is_local": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "word_delimiter_token": "|"
+  },
+  "pad_token": "<pad>",
+  "processor_class": "Wav2Vec2Processor",
+  "replace_word_delimiter_char": " ",
+  "target_lang": "wol",
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "unk_token": "<unk>",
+  "word_delimiter_token": "|"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ccb3f489033d1906cbdc076b633aa8f5ce73efa6e7f0d5411b18e507b91d409
+size 5137