| | --- |
| | library_name: transformers |
| | language: |
| | - kh |
| | license: mit |
| | base_model: |
| | - facebook/nllb-200-distilled-600M |
| | tags: |
| | - generated_from_trainer |
| | datasets: |
| | - S-Sethisak |
| | metrics: |
| | - wer |
| | - bleu |
| | - cer |
| | model-index: |
| | - name: Khmer Orthographic Correction System using NLLB |
| | results: |
| | - task: |
| | name: Sequence-to-sequence Language Modeling |
| | type: text2text-generation |
| | dataset: |
| | name: khmer-orthography-correction-dataset |
| | type: S-Sethisak |
| | metrics: |
| | - name: Wer |
| | type: wer |
| | value: 0.3923043541154347 |
| | - name: Bleu |
| | type: bleu |
| | value: |
| | score: 60.71026653645501 |
| | counts: |
| | - 4985 |
| | - 558 |
| | - 166 |
| | - 33 |
| | totals: |
| | - 7537 |
| | - 805 |
| | - 267 |
| | - 63 |
| | precisions: |
| | - 66.14037415417275 |
| | - 69.3167701863354 |
| | - 62.172284644194754 |
| | - 52.38095238095238 |
| | bp: 0.9766598710135985 |
| | sys_len: 7537 |
| | ref_len: 7715 |
| | --- |
| | |
| | <!-- 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. --> |
| |
|
| | # Khmer Orthographic Correction System using NLLB |
| |
|
| | This model is a fine-tuned version of [facebook/mbart-large-50](https://huggingface.co/facebook/mbart-large-50) on the khmer-orthography-correction-dataset dataset. |
| | It achieves the following results on the evaluation set: |
| | - Loss: 0.1823 |
| | - Cer: 0.0473 |
| | - Wer: 0.3923 |
| | - Bleu: {'score': 60.71026653645501, 'counts': [4985, 558, 166, 33], 'totals': [7537, 805, 267, 63], 'precisions': [66.14037415417275, 69.3167701863354, 62.172284644194754, 52.38095238095238], 'bp': 0.9766598710135985, 'sys_len': 7537, 'ref_len': 7715} |
| |
|
| | ## 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: 2e-05 |
| | - train_batch_size: 16 |
| | - eval_batch_size: 16 |
| | - seed: 42 |
| | - gradient_accumulation_steps: 4 |
| | - total_train_batch_size: 64 |
| | - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
| | - lr_scheduler_type: linear |
| | - num_epochs: 10 |
| | - mixed_precision_training: Native AMP |
| |
|
| | ### Training results |
| |
|
| | | Training Loss | Epoch | Step | Validation Loss | Cer | Wer | Bleu | |
| | |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------:| |
| | | 0.9311 | 1.0 | 842 | 0.4968 | 0.1203 | 0.7082 | {'score': 34.801638642481215, 'counts': [2845, 373, 91, 19], 'totals': [7609, 877, 277, 64], 'precisions': [37.389932974109605, 42.53135689851767, 32.851985559566785, 29.6875], 'bp': 0.9861657142243254, 'sys_len': 7609, 'ref_len': 7715} | |
| | | 0.4966 | 2.0 | 1684 | 0.3594 | 0.0898 | 0.6000 | {'score': 44.88058034913873, 'counts': [3561, 447, 122, 24], 'totals': [7557, 825, 269, 63], 'precisions': [47.12187375942835, 54.18181818181818, 45.353159851301115, 38.095238095238095], 'bp': 0.9793092844178501, 'sys_len': 7557, 'ref_len': 7715} | |
| | | 0.3655 | 3.0 | 2526 | 0.2919 | 0.0767 | 0.5307 | {'score': 50.98256581813308, 'counts': [4034, 487, 138, 29], 'totals': [7549, 817, 270, 64], 'precisions': [53.437541396211415, 59.608323133414935, 51.111111111111114, 45.3125], 'bp': 0.9782503427651501, 'sys_len': 7549, 'ref_len': 7715} | |
| | | 0.327 | 4.0 | 3368 | 0.2529 | 0.0657 | 0.4862 | {'score': 53.914915512375295, 'counts': [4343, 511, 146, 30], 'totals': [7544, 812, 268, 64], 'precisions': [57.56892895015907, 62.93103448275862, 54.47761194029851, 46.875], 'bp': 0.977587946673324, 'sys_len': 7544, 'ref_len': 7715} | |
| | | 0.2712 | 5.0 | 4210 | 0.2253 | 0.0591 | 0.4484 | {'score': 55.04966846293549, 'counts': [4604, 528, 152, 30], 'totals': [7547, 815, 272, 66], 'precisions': [61.00437259838346, 64.78527607361963, 55.88235294117647, 45.45454545454545], 'bp': 0.9779854358166019, 'sys_len': 7547, 'ref_len': 7715} | |
| | | 0.2406 | 6.0 | 5052 | 0.2079 | 0.0530 | 0.4289 | {'score': 58.09076687466732, 'counts': [4731, 536, 158, 32], 'totals': [7537, 805, 268, 63], 'precisions': [62.7703330237495, 66.58385093167702, 58.95522388059702, 50.79365079365079], 'bp': 0.9766598710135985, 'sys_len': 7537, 'ref_len': 7715} | |
| | | 0.2243 | 7.0 | 5894 | 0.1962 | 0.0508 | 0.4118 | {'score': 59.13761418304522, 'counts': [4851, 546, 162, 32], 'totals': [7538, 806, 267, 63], 'precisions': [64.35394003714514, 67.74193548387096, 60.674157303370784, 50.79365079365079], 'bp': 0.976792504783367, 'sys_len': 7538, 'ref_len': 7715} | |
| | | 0.2067 | 8.0 | 6736 | 0.1879 | 0.0482 | 0.3997 | {'score': 60.725787252253205, 'counts': [4939, 556, 167, 34], 'totals': [7540, 808, 268, 64], 'precisions': [65.50397877984085, 68.81188118811882, 62.3134328358209, 53.125], 'bp': 0.977057720781772, 'sys_len': 7540, 'ref_len': 7715} | |
| | | 0.1962 | 9.0 | 7578 | 0.1841 | 0.0480 | 0.3939 | {'score': 60.53730944860592, 'counts': [4972, 556, 165, 33], 'totals': [7536, 804, 267, 63], 'precisions': [65.97664543524417, 69.1542288557214, 61.79775280898876, 52.38095238095238], 'bp': 0.9765272200606311, 'sys_len': 7536, 'ref_len': 7715} | |
| | | 0.194 | 10.0 | 8420 | 0.1823 | 0.0473 | 0.3923 | {'score': 60.71026653645501, 'counts': [4985, 558, 166, 33], 'totals': [7537, 805, 267, 63], 'precisions': [66.14037415417275, 69.3167701863354, 62.172284644194754, 52.38095238095238], 'bp': 0.9766598710135985, 'sys_len': 7537, 'ref_len': 7715} | |
| |
|
| |
|
| | ### Framework versions |
| |
|
| | - Transformers 4.57.2 |
| | - Pytorch 2.9.0+cu126 |
| | - Datasets 4.0.0 |
| | - Tokenizers 0.22.1 |
