scenario-NON-KD-PR-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: 3.2943
• Accuracy: 0.5494
• F1: 0.5481

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
1.0706	1.09	500	1.0260	0.4946	0.4746
0.97	2.17	1000	0.9626	0.5270	0.5111
0.9007	3.26	1500	0.9517	0.5490	0.5506
0.8292	4.35	2000	0.9865	0.5475	0.5339
0.7773	5.43	2500	1.0123	0.5475	0.5388
0.7155	6.52	3000	1.0156	0.5536	0.5509
0.6707	7.61	3500	1.0641	0.5637	0.5629
0.6081	8.7	4000	1.1244	0.5617	0.5614
0.5489	9.78	4500	1.1851	0.5475	0.5404
0.503	10.87	5000	1.3213	0.5586	0.5591
0.4602	11.96	5500	1.4164	0.5563	0.5520
0.4134	13.04	6000	1.4394	0.5482	0.5414
0.3697	14.13	6500	1.5564	0.5529	0.5501
0.3256	15.22	7000	1.5412	0.5478	0.5465
0.3028	16.3	7500	1.6169	0.5374	0.5353
0.2673	17.39	8000	1.7963	0.5436	0.5432
0.2501	18.48	8500	1.7266	0.5517	0.5499
0.2225	19.57	9000	1.9252	0.5532	0.5472
0.2021	20.65	9500	1.9622	0.5563	0.5542
0.1875	21.74	10000	1.9973	0.5471	0.5463
0.1681	22.83	10500	2.1299	0.5386	0.5307
0.1534	23.91	11000	2.0761	0.5463	0.5416
0.1427	25.0	11500	2.2814	0.5475	0.5471
0.1304	26.09	12000	2.4128	0.5544	0.5451
0.1126	27.17	12500	2.4318	0.5370	0.5327
0.1169	28.26	13000	2.5110	0.5451	0.5432
0.1044	29.35	13500	2.5768	0.5467	0.5432
0.1011	30.43	14000	2.6120	0.5486	0.5428
0.0982	31.52	14500	2.5795	0.5544	0.5541
0.0854	32.61	15000	2.7525	0.5482	0.5497
0.0853	33.7	15500	2.7322	0.5575	0.5557
0.0851	34.78	16000	2.7708	0.5417	0.5375
0.0726	35.87	16500	2.8363	0.5451	0.5417
0.0706	36.96	17000	2.8634	0.5505	0.5494
0.0653	38.04	17500	2.9653	0.5444	0.5434
0.0669	39.13	18000	3.0624	0.5432	0.5417
0.0585	40.22	18500	3.1669	0.5432	0.5392
0.059	41.3	19000	3.0692	0.5548	0.5544
0.048	42.39	19500	3.2014	0.5494	0.5482
0.0479	43.48	20000	3.2452	0.5428	0.5409
0.052	44.57	20500	3.2338	0.5478	0.5476
0.0477	45.65	21000	3.2556	0.5444	0.5424
0.0395	46.74	21500	3.2952	0.5444	0.5420
0.0477	47.83	22000	3.2726	0.5509	0.5500
0.0408	48.91	22500	3.2894	0.5471	0.5457
0.0407	50.0	23000	3.2943	0.5494	0.5481

Framework versions

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