scenario-NON-KD-SCR-COPY-CDF-ALL-D2_data-cardiffnlp_tweet_sentiment_multilingual
This model is a fine-tuned version of xlm-roberta-base on the tweet_sentiment_multilingual dataset. It achieves the following results on the evaluation set:
- Loss: 5.5508
- Accuracy: 0.4985
- F1: 0.4951
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: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 11213
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
---|---|---|---|---|---|
1.1212 | 1.09 | 500 | 1.1080 | 0.3789 | 0.2909 |
0.9761 | 2.17 | 1000 | 1.1799 | 0.4823 | 0.4755 |
0.625 | 3.26 | 1500 | 1.5124 | 0.5015 | 0.4837 |
0.3503 | 4.35 | 2000 | 1.5134 | 0.5220 | 0.5211 |
0.205 | 5.43 | 2500 | 2.4452 | 0.5143 | 0.5135 |
0.1271 | 6.52 | 3000 | 2.9579 | 0.4911 | 0.4721 |
0.1042 | 7.61 | 3500 | 2.7856 | 0.4892 | 0.4895 |
0.079 | 8.7 | 4000 | 3.0258 | 0.5035 | 0.4994 |
0.061 | 9.78 | 4500 | 3.6011 | 0.4884 | 0.4892 |
0.0504 | 10.87 | 5000 | 3.5818 | 0.4819 | 0.4791 |
0.0474 | 11.96 | 5500 | 3.3451 | 0.5031 | 0.4913 |
0.0462 | 13.04 | 6000 | 3.5810 | 0.4988 | 0.4966 |
0.0317 | 14.13 | 6500 | 3.6952 | 0.4988 | 0.4967 |
0.0325 | 15.22 | 7000 | 3.9715 | 0.4919 | 0.4847 |
0.0269 | 16.3 | 7500 | 3.8701 | 0.4942 | 0.4934 |
0.0268 | 17.39 | 8000 | 4.2094 | 0.4826 | 0.4834 |
0.0256 | 18.48 | 8500 | 3.9200 | 0.4934 | 0.4947 |
0.0228 | 19.57 | 9000 | 3.7341 | 0.4946 | 0.4932 |
0.0183 | 20.65 | 9500 | 4.2147 | 0.4915 | 0.4867 |
0.0176 | 21.74 | 10000 | 4.3251 | 0.4985 | 0.4942 |
0.012 | 22.83 | 10500 | 4.2855 | 0.5042 | 0.4920 |
0.011 | 23.91 | 11000 | 4.2243 | 0.4946 | 0.4874 |
0.0106 | 25.0 | 11500 | 4.4153 | 0.4877 | 0.4855 |
0.0085 | 26.09 | 12000 | 4.6839 | 0.4946 | 0.4920 |
0.0105 | 27.17 | 12500 | 4.5992 | 0.4923 | 0.4933 |
0.0095 | 28.26 | 13000 | 4.7752 | 0.4985 | 0.4952 |
0.0083 | 29.35 | 13500 | 4.7973 | 0.4942 | 0.4948 |
0.007 | 30.43 | 14000 | 4.7373 | 0.4969 | 0.4937 |
0.0041 | 31.52 | 14500 | 5.0320 | 0.4954 | 0.4816 |
0.0042 | 32.61 | 15000 | 5.1395 | 0.4934 | 0.4921 |
0.0049 | 33.7 | 15500 | 4.9622 | 0.4958 | 0.4957 |
0.0054 | 34.78 | 16000 | 5.2670 | 0.4846 | 0.4826 |
0.0042 | 35.87 | 16500 | 5.1694 | 0.4958 | 0.4951 |
0.004 | 36.96 | 17000 | 5.2387 | 0.4938 | 0.4867 |
0.002 | 38.04 | 17500 | 5.4227 | 0.4842 | 0.4797 |
0.0025 | 39.13 | 18000 | 5.4860 | 0.4896 | 0.4849 |
0.003 | 40.22 | 18500 | 5.3279 | 0.4923 | 0.4870 |
0.0026 | 41.3 | 19000 | 5.2518 | 0.4923 | 0.4924 |
0.0019 | 42.39 | 19500 | 5.1927 | 0.4996 | 0.4989 |
0.0018 | 43.48 | 20000 | 5.2447 | 0.4992 | 0.4991 |
0.0007 | 44.57 | 20500 | 5.4302 | 0.5015 | 0.4986 |
0.0002 | 45.65 | 21000 | 5.4714 | 0.4950 | 0.4936 |
0.0001 | 46.74 | 21500 | 5.5014 | 0.4969 | 0.4955 |
0.0003 | 47.83 | 22000 | 5.5333 | 0.4973 | 0.4934 |
0.0001 | 48.91 | 22500 | 5.5591 | 0.4969 | 0.4942 |
0.0002 | 50.0 | 23000 | 5.5508 | 0.4985 | 0.4951 |
Framework versions
- Transformers 4.33.3
- Pytorch 2.1.1+cu121
- Datasets 2.14.5
- Tokenizers 0.13.3
- Downloads last month
- 10
Finetuned from
Evaluation results
- Accuracy on tweet_sentiment_multilingualvalidation set self-reported0.498
- F1 on tweet_sentiment_multilingualvalidation set self-reported0.495