scenario-TCR_data-cardiffnlp_tweet_sentiment_multilingual_all_d

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: 2.8506
• Accuracy: 0.6397
• F1: 0.6384

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: 1e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 53
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.9598	1.09	500	0.8321	0.6335	0.6229
0.7983	2.17	1000	0.7922	0.6381	0.6278
0.7031	3.26	1500	0.8300	0.6520	0.6468
0.6192	4.35	2000	0.8659	0.6497	0.6443
0.5472	5.43	2500	0.9646	0.6331	0.6343
0.4664	6.52	3000	0.9555	0.6485	0.6455
0.4025	7.61	3500	1.0121	0.6427	0.6405
0.3568	8.7	4000	1.1016	0.6327	0.6324
0.3069	9.78	4500	1.2521	0.6408	0.6400
0.2701	10.87	5000	1.3727	0.6397	0.6372
0.2398	11.96	5500	1.4539	0.6319	0.6334
0.2004	13.04	6000	1.6097	0.6420	0.6376
0.1864	14.13	6500	1.6302	0.6343	0.6349
0.157	15.22	7000	1.7491	0.6381	0.6339
0.1411	16.3	7500	1.8634	0.6400	0.6392
0.1318	17.39	8000	2.0229	0.6277	0.6275
0.1159	18.48	8500	2.0196	0.6385	0.6359
0.1135	19.57	9000	2.1959	0.6377	0.6368
0.1018	20.65	9500	2.3238	0.6323	0.6320
0.0888	21.74	10000	2.3449	0.6339	0.6341
0.0797	22.83	10500	2.4967	0.6354	0.6338
0.0828	23.91	11000	2.5070	0.6358	0.6362
0.0675	25.0	11500	2.5895	0.6381	0.6393
0.067	26.09	12000	2.6730	0.6370	0.6372
0.0566	27.17	12500	2.7454	0.6377	0.6386
0.0571	28.26	13000	2.7673	0.6420	0.6413
0.048	29.35	13500	2.8506	0.6397	0.6384

Framework versions

• Transformers 4.33.3
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.13.3