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: 5.7429
- Accuracy: 0.4973
- F1: 0.4956
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: 222
- 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 |
---|---|---|---|---|---|
1.1068 | 1.09 | 500 | 1.0681 | 0.4352 | 0.4321 |
0.9081 | 2.17 | 1000 | 1.2631 | 0.5046 | 0.5021 |
0.5532 | 3.26 | 1500 | 1.5304 | 0.5108 | 0.5089 |
0.2998 | 4.35 | 2000 | 2.0584 | 0.4884 | 0.4858 |
0.1717 | 5.43 | 2500 | 2.7362 | 0.5 | 0.4939 |
0.1242 | 6.52 | 3000 | 3.0470 | 0.4969 | 0.4938 |
0.0874 | 7.61 | 3500 | 2.7990 | 0.5046 | 0.5037 |
0.0669 | 8.7 | 4000 | 3.2793 | 0.4942 | 0.4940 |
0.056 | 9.78 | 4500 | 3.2094 | 0.5027 | 0.5028 |
0.0487 | 10.87 | 5000 | 3.5054 | 0.4992 | 0.4972 |
0.0539 | 11.96 | 5500 | 3.2798 | 0.5008 | 0.5003 |
0.0317 | 13.04 | 6000 | 3.4251 | 0.5004 | 0.4994 |
0.0449 | 14.13 | 6500 | 4.0353 | 0.4969 | 0.4923 |
0.0303 | 15.22 | 7000 | 4.3157 | 0.4850 | 0.4733 |
0.0285 | 16.3 | 7500 | 3.8740 | 0.4985 | 0.4987 |
0.0214 | 17.39 | 8000 | 4.5553 | 0.4842 | 0.4828 |
0.0228 | 18.48 | 8500 | 4.7444 | 0.4946 | 0.4903 |
0.0177 | 19.57 | 9000 | 4.5373 | 0.4969 | 0.4939 |
0.0167 | 20.65 | 9500 | 4.4792 | 0.4927 | 0.4859 |
0.0144 | 21.74 | 10000 | 4.6491 | 0.4896 | 0.4897 |
0.0164 | 22.83 | 10500 | 4.8310 | 0.4934 | 0.4926 |
0.0116 | 23.91 | 11000 | 4.6267 | 0.4996 | 0.4965 |
0.0102 | 25.0 | 11500 | 5.0420 | 0.4904 | 0.4808 |
0.0053 | 26.09 | 12000 | 5.2202 | 0.4915 | 0.4824 |
0.01 | 27.17 | 12500 | 4.8786 | 0.4900 | 0.4868 |
0.0076 | 28.26 | 13000 | 4.8830 | 0.4919 | 0.4906 |
0.0064 | 29.35 | 13500 | 5.2319 | 0.4934 | 0.4890 |
0.0055 | 30.43 | 14000 | 5.4810 | 0.4973 | 0.4953 |
0.0057 | 31.52 | 14500 | 5.4109 | 0.5035 | 0.5019 |
0.0032 | 32.61 | 15000 | 5.3979 | 0.5054 | 0.5041 |
0.0092 | 33.7 | 15500 | 5.3848 | 0.4942 | 0.4940 |
0.0053 | 34.78 | 16000 | 5.2937 | 0.5066 | 0.5046 |
0.0029 | 35.87 | 16500 | 5.5430 | 0.5012 | 0.4971 |
0.0011 | 36.96 | 17000 | 5.6338 | 0.4919 | 0.4905 |
0.0027 | 38.04 | 17500 | 5.6234 | 0.4958 | 0.4960 |
0.0042 | 39.13 | 18000 | 5.5802 | 0.4988 | 0.4991 |
0.0012 | 40.22 | 18500 | 5.6464 | 0.4988 | 0.4993 |
0.0037 | 41.3 | 19000 | 5.6227 | 0.4965 | 0.4945 |
0.0007 | 42.39 | 19500 | 5.6263 | 0.4958 | 0.4939 |
0.0003 | 43.48 | 20000 | 5.6946 | 0.4934 | 0.4937 |
0.0016 | 44.57 | 20500 | 5.6654 | 0.4973 | 0.4977 |
0.0018 | 45.65 | 21000 | 5.6725 | 0.4965 | 0.4952 |
0.0012 | 46.74 | 21500 | 5.6500 | 0.4873 | 0.4869 |
0.0008 | 47.83 | 22000 | 5.6626 | 0.4992 | 0.4985 |
0.0006 | 48.91 | 22500 | 5.7378 | 0.4985 | 0.4968 |
0.0004 | 50.0 | 23000 | 5.7429 | 0.4973 | 0.4956 |
Framework versions
- Transformers 4.33.3
- Pytorch 2.1.1+cu121
- Datasets 2.14.5
- Tokenizers 0.13.3
- Downloads last month
- 7
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for haryoaw/scenario-NON-KD-SCR-COPY-CDF-ALL-D2_data-cardiffnlp_tweet_sentiment_multilingual
Base model
FacebookAI/xlm-roberta-base
Finetuned
this model
Evaluation results
- Accuracy on tweet_sentiment_multilingualvalidation set self-reported0.497
- F1 on tweet_sentiment_multilingualvalidation set self-reported0.496