scenario-KD-PR-CDF-ALL-D2_data-cardiffnlp_tweet_sentiment_multilingual_all_alpha

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the tweet_sentiment_multilingual dataset. It achieves the following results on the evaluation set:

• Loss: 3.4722
• Accuracy: 0.5613
• F1: 0.5612

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: 2222
• 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
4.8519	1.09	500	4.2733	0.4707	0.4645
3.9632	2.17	1000	3.7091	0.4869	0.4826
3.3756	3.26	1500	3.3036	0.5440	0.5454
3.0456	4.35	2000	3.3103	0.5428	0.5360
2.7208	5.43	2500	3.3838	0.5394	0.5336
2.5109	6.52	3000	3.1774	0.5502	0.5438
2.3054	7.61	3500	3.2425	0.5548	0.5537
2.1185	8.7	4000	3.4164	0.5509	0.5430
1.9485	9.78	4500	3.4932	0.5552	0.5539
1.8216	10.87	5000	3.3822	0.5440	0.5449
1.7309	11.96	5500	3.5055	0.5486	0.5469
1.5951	13.04	6000	3.3815	0.5594	0.5526
1.4855	14.13	6500	3.2450	0.5463	0.5480
1.4173	15.22	7000	3.3808	0.5525	0.5491
1.3622	16.3	7500	3.3682	0.5602	0.5592
1.2813	17.39	8000	3.5649	0.5633	0.5601
1.2153	18.48	8500	3.8051	0.5525	0.5466
1.1744	19.57	9000	3.6361	0.5455	0.5467
1.1395	20.65	9500	3.6349	0.5648	0.5627
1.0783	21.74	10000	3.5260	0.5482	0.5480
1.0348	22.83	10500	3.5517	0.5544	0.5552
1.0165	23.91	11000	3.5108	0.5702	0.5663
0.9988	25.0	11500	3.5595	0.5559	0.5551
0.95	26.09	12000	3.4064	0.5621	0.5627
0.9334	27.17	12500	3.5069	0.5544	0.5542
0.9119	28.26	13000	3.3631	0.5637	0.5634
0.8913	29.35	13500	3.4203	0.5475	0.5467
0.863	30.43	14000	3.5737	0.5536	0.5548
0.858	31.52	14500	3.4680	0.5617	0.5618
0.8341	32.61	15000	3.4596	0.5513	0.5523
0.8295	33.7	15500	3.5880	0.5633	0.5637
0.8	34.78	16000	3.5100	0.5590	0.5600
0.8107	35.87	16500	3.4800	0.5633	0.5631
0.779	36.96	17000	3.5637	0.5648	0.5620
0.7795	38.04	17500	3.3500	0.5563	0.5573
0.7595	39.13	18000	3.4628	0.5671	0.5675
0.7501	40.22	18500	3.4355	0.5594	0.5590
0.7404	41.3	19000	3.4201	0.5590	0.5592
0.7375	42.39	19500	3.4007	0.5698	0.5709
0.7345	43.48	20000	3.4599	0.5779	0.5778
0.7227	44.57	20500	3.4411	0.5683	0.5690
0.7221	45.65	21000	3.4918	0.5660	0.5662
0.7116	46.74	21500	3.3447	0.5606	0.5612
0.7094	47.83	22000	3.3950	0.5656	0.5661
0.7097	48.91	22500	3.3911	0.5691	0.5699
0.6983	50.0	23000	3.4722	0.5613	0.5612

Framework versions

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