scenario-TCR_data-cardiffnlp_tweet_sentiment_multilingual_all_a

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: 2.6822
• Accuracy: 0.6439
• F1: 0.6444

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: 1234
• 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
0.9412	1.09	500	0.8062	0.6389	0.6335
0.7943	2.17	1000	0.8448	0.6451	0.6394
0.7026	3.26	1500	0.8509	0.6497	0.6438
0.6019	4.35	2000	0.8999	0.6478	0.6468
0.5379	5.43	2500	0.9424	0.6312	0.6222
0.4635	6.52	3000	1.0401	0.6431	0.6439
0.3985	7.61	3500	1.0584	0.6397	0.6390
0.3506	8.7	4000	1.1607	0.6443	0.6432
0.3105	9.78	4500	1.1806	0.6408	0.6423
0.2712	10.87	5000	1.3112	0.6316	0.6304
0.2361	11.96	5500	1.3772	0.6466	0.6454
0.2111	13.04	6000	1.4492	0.6385	0.6396
0.1885	14.13	6500	1.6604	0.6335	0.6347
0.1658	15.22	7000	1.7153	0.6358	0.6353
0.1501	16.3	7500	1.7849	0.6412	0.6427
0.135	17.39	8000	1.9749	0.6416	0.6394
0.1217	18.48	8500	2.0530	0.6439	0.6431
0.1112	19.57	9000	2.1378	0.6439	0.6448
0.1018	20.65	9500	2.2656	0.6393	0.6390
0.0885	21.74	10000	2.3568	0.6431	0.6438
0.0897	22.83	10500	2.3852	0.6435	0.6446
0.0854	23.91	11000	2.4019	0.6327	0.6329
0.0734	25.0	11500	2.5260	0.6331	0.6333
0.067	26.09	12000	2.5368	0.6470	0.6465
0.0546	27.17	12500	2.6255	0.6431	0.6441
0.0581	28.26	13000	2.6467	0.6458	0.6456
0.0564	29.35	13500	2.6822	0.6439	0.6444

Framework versions

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