Hugging Face's logo

Kobi-01
/
tamil-qa-distilled

PEFT
Safetensors
Generated from Trainer
Model card Files Community

tamil-qa-distilled

This model is a fine-tuned version of xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 2.3458
  • Exact: 23.9056
  • F1: 40.0870
  • Total: 5848
  • Hasans Exact: 23.9056
  • Hasans F1: 40.0870
  • Hasans Total: 5848
  • Best Exact: 23.9056
  • Best Exact Thresh: 0.0
  • Best F1: 40.0870
  • Best F1 Thresh: 0.0

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: 0.0002
  • train_batch_size: 4
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 8
  • optimizer: Use paged_adamw_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss Exact F1 Total Hasans Exact Hasans F1 Hasans Total Best Exact Best Exact Thresh Best F1 Best F1 Thresh
5.9372 0.0302 250 5.5830 0.0342 5.2752 5848 0.0342 5.2752 5848 0.0342 0.0 5.2752 0.0
4.4435 0.0604 500 4.1190 5.5404 14.4289 5848 5.5404 14.4289 5848 5.5404 0.0 14.4289 0.0
4.1362 0.0905 750 3.8871 11.4227 22.5581 5848 11.4227 22.5581 5848 11.4227 0.0 22.5581 0.0
4.071 0.1207 1000 3.6548 12.6197 23.8673 5848 12.6197 23.8673 5848 12.6197 0.0 23.8673 0.0
3.4979 0.1509 1250 3.3795 12.7052 23.1488 5848 12.7052 23.1488 5848 12.7052 0.0 23.1488 0.0
3.5602 0.1811 1500 3.1890 16.9631 28.6981 5848 16.9631 28.6981 5848 16.9631 0.0 28.6981 0.0
3.1678 0.2112 1750 3.0806 17.1512 31.1741 5848 17.1512 31.1741 5848 17.1512 0.0 31.1741 0.0
3.3402 0.2414 2000 3.0524 20.2975 34.5701 5848 20.2975 34.5701 5848 20.2975 0.0 34.5701 0.0
3.2099 0.2716 2250 2.8973 17.4761 30.0263 5848 17.4761 30.0263 5848 17.4761 0.0 30.0263 0.0
2.9404 0.3018 2500 2.9263 20.3830 35.2615 5848 20.3830 35.2615 5848 20.3830 0.0 35.2615 0.0
3.5376 0.3319 2750 2.8066 19.1518 33.0098 5848 19.1518 33.0098 5848 19.1518 0.0 33.0098 0.0
2.8949 0.3621 3000 2.7580 18.1601 30.1449 5848 18.1601 30.1449 5848 18.1601 0.0 30.1449 0.0
3.1584 0.3923 3250 2.8535 21.9904 37.2070 5848 21.9904 37.2070 5848 21.9904 0.0 37.2070 0.0
2.7833 0.4225 3500 2.8549 21.7339 36.0716 5848 21.7339 36.0716 5848 21.7339 0.0 36.0716 0.0
2.9223 0.4526 3750 2.7775 19.8529 33.9325 5848 19.8529 33.9325 5848 19.8529 0.0 33.9325 0.0
3.0183 0.4828 4000 2.6326 17.8865 29.0245 5848 17.8865 29.0245 5848 17.8865 0.0 29.0245 0.0
2.8363 0.5130 4250 2.7080 22.2298 37.7239 5848 22.2298 37.7239 5848 22.2298 0.0 37.7239 0.0
2.7222 0.5432 4500 2.8187 22.1785 38.5146 5848 22.1785 38.5146 5848 22.1785 0.0 38.5146 0.0
2.6763 0.5733 4750 2.6953 22.5547 37.7873 5848 22.5547 37.7873 5848 22.5547 0.0 37.7873 0.0
2.5925 0.6035 5000 2.8178 22.6573 38.2837 5848 22.6573 38.2837 5848 22.6573 0.0 38.2837 0.0
2.6485 0.6337 5250 2.6554 22.1101 37.5204 5848 22.1101 37.5204 5848 22.1101 0.0 37.5204 0.0
3.0327 0.6639 5500 2.6043 21.9733 37.8703 5848 21.9733 37.8703 5848 21.9733 0.0 37.8703 0.0
2.7016 0.6940 5750 2.5752 22.1272 38.2156 5848 22.1272 38.2156 5848 22.1272 0.0 38.2156 0.0
2.6785 0.7242 6000 2.6829 22.5889 38.4805 5848 22.5889 38.4805 5848 22.5889 0.0 38.4805 0.0
2.7414 0.7544 6250 2.5243 21.8023 36.6811 5848 21.8023 36.6811 5848 21.8023 0.0 36.6811 0.0
2.6025 0.7846 6500 2.4759 22.2811 36.9185 5848 22.2811 36.9185 5848 22.2811 0.0 36.9185 0.0
2.6305 0.8147 6750 2.6849 23.3926 39.6983 5848 23.3926 39.6983 5848 23.3926 0.0 39.6983 0.0
2.6485 0.8449 7000 2.5377 23.2900 38.9523 5848 23.2900 38.9523 5848 23.2900 0.0 38.9523 0.0
2.6644 0.8751 7250 2.5613 22.6915 37.9573 5848 22.6915 37.9573 5848 22.6915 0.0 37.9573 0.0
2.4504 0.9053 7500 2.5179 23.1019 39.2826 5848 23.1019 39.2826 5848 23.1019 0.0 39.2826 0.0
2.7274 0.9354 7750 2.4839 23.3242 39.6564 5848 23.3242 39.6564 5848 23.3242 0.0 39.6564 0.0
2.732 0.9656 8000 2.4473 22.6915 38.1896 5848 22.6915 38.1896 5848 22.6915 0.0 38.1896 0.0
2.5412 0.9958 8250 2.4536 22.8112 38.1155 5848 22.8112 38.1155 5848 22.8112 0.0 38.1155 0.0
2.4763 1.0260 8500 2.4868 23.6491 39.5944 5848 23.6491 39.5944 5848 23.6491 0.0 39.5944 0.0
2.6547 1.0561 8750 2.5451 23.8030 40.6998 5848 23.8030 40.6998 5848 23.8030 0.0 40.6998 0.0
2.5899 1.0863 9000 2.4332 22.6915 37.8246 5848 22.6915 37.8246 5848 22.6915 0.0 37.8246 0.0
2.5589 1.1165 9250 2.3779 22.0417 36.8825 5848 22.0417 36.8825 5848 22.0417 0.0 36.8825 0.0
2.3319 1.1467 9500 2.3715 22.7086 38.7405 5848 22.7086 38.7405 5848 22.7086 0.0 38.7405 0.0
2.2501 1.1768 9750 2.4244 23.4781 39.4466 5848 23.4781 39.4466 5848 23.4781 0.0 39.4466 0.0
2.337 1.2070 10000 2.3661 22.5718 36.8759 5848 22.5718 36.8759 5848 22.5718 0.0 36.8759 0.0
2.5074 1.2372 10250 2.6080 25.1368 42.1434 5848 25.1368 42.1434 5848 25.1368 0.0 42.1434 0.0
2.531 1.2674 10500 2.3560 21.6484 36.3479 5848 21.6484 36.3479 5848 21.6484 0.0 36.3479 0.0
2.513 1.2975 10750 2.4430 23.2387 39.3702 5848 23.2387 39.3702 5848 23.2387 0.0 39.3702 0.0
2.2461 1.3277 11000 2.5602 23.6149 40.3775 5848 23.6149 40.3775 5848 23.6149 0.0 40.3775 0.0
2.5934 1.3579 11250 2.4806 22.8625 39.5232 5848 22.8625 39.5232 5848 22.8625 0.0 39.5232 0.0
2.2671 1.3881 11500 2.4027 22.7599 37.6504 5848 22.7599 37.6504 5848 22.7599 0.0 37.6504 0.0
2.3451 1.4182 11750 2.4227 23.1703 39.2540 5848 23.1703 39.2540 5848 23.1703 0.0 39.2540 0.0
2.6997 1.4484 12000 2.3992 23.2216 39.6402 5848 23.2216 39.6402 5848 23.2216 0.0 39.6402 0.0
2.5253 1.4786 12250 2.4076 24.1621 40.3112 5848 24.1621 40.3112 5848 24.1621 0.0 40.3112 0.0
2.6091 1.5088 12500 2.3743 22.4179 38.6880 5848 22.4179 38.6880 5848 22.4179 0.0 38.6880 0.0
2.3445 1.5389 12750 2.3862 23.9056 39.8296 5848 23.9056 39.8296 5848 23.9056 0.0 39.8296 0.0
2.4083 1.5691 13000 2.5380 24.8461 41.8272 5848 24.8461 41.8272 5848 24.8461 0.0 41.8272 0.0
2.2722 1.5993 13250 2.3600 23.5123 39.5265 5848 23.5123 39.5265 5848 23.5123 0.0 39.5265 0.0
2.4945 1.6295 13500 2.3862 23.9056 39.8474 5848 23.9056 39.8474 5848 23.9056 0.0 39.8474 0.0
2.5895 1.6596 13750 2.3518 23.4952 39.1746 5848 23.4952 39.1746 5848 23.4952 0.0 39.1746 0.0
2.3354 1.6898 14000 2.3872 23.9398 39.7158 5848 23.9398 39.7158 5848 23.9398 0.0 39.7158 0.0
2.4015 1.7200 14250 2.3248 23.4439 39.5145 5848 23.4439 39.5145 5848 23.4439 0.0 39.5145 0.0
2.4879 1.7502 14500 2.3493 24.1279 40.3504 5848 24.1279 40.3504 5848 24.1279 0.0 40.3504 0.0
2.1585 1.7803 14750 2.3772 24.0424 40.2189 5848 24.0424 40.2189 5848 24.0424 0.0 40.2189 0.0
2.307 1.8105 15000 2.3749 23.8543 40.0715 5848 23.8543 40.0715 5848 23.8543 0.0 40.0715 0.0
2.2712 1.8407 15250 2.3785 24.1279 40.3913 5848 24.1279 40.3913 5848 24.1279 0.0 40.3913 0.0
2.2997 1.8709 15500 2.3398 23.6320 39.4552 5848 23.6320 39.4552 5848 23.6320 0.0 39.4552 0.0
2.7463 1.9010 15750 2.3623 23.9227 39.9884 5848 23.9227 39.9884 5848 23.9227 0.0 39.9884 0.0
2.6714 1.9312 16000 2.3396 24.0424 40.0280 5848 24.0424 40.0280 5848 24.0424 0.0 40.0280 0.0
2.4979 1.9614 16250 2.3225 23.7004 39.5895 5848 23.7004 39.5895 5848 23.7004 0.0 39.5895 0.0
2.4623 1.9916 16500 2.3458 23.9056 40.0870 5848 23.9056 40.0870 5848 23.9056 0.0 40.0870 0.0

Framework versions

  • PEFT 0.14.0
  • Transformers 4.48.2
  • Pytorch 2.5.1+cu121
  • Datasets 3.2.0
  • Tokenizers 0.21.0
Downloads last month
2
Inference Providers NEW
This model is not currently available via any of the supported third-party Inference Providers, and HF Inference API was unable to determine this model’s pipeline type.

Model tree for Kobi-01/tamil-qa-distilled

Base model

FacebookAI/xlm-roberta-base
Adapter
(21)
this model

Evaluation results

Metadata error: specify a dataset to view leaderboard