scenario-KD-PR-CDF-ALL-D2_data-cardiffnlp_tweet_sentiment_multilingual_all_gamma

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: 3.4838
• Accuracy: 0.5505
• F1: 0.5508

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: 88458
• 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
4.8911	1.09	500	4.3088	0.4047	0.3809
4.106	2.17	1000	3.7773	0.5058	0.4920
3.4954	3.26	1500	3.3608	0.5347	0.5357
3.1472	4.35	2000	3.4225	0.5343	0.5327
2.8094	5.43	2500	3.4088	0.5436	0.5399
2.5414	6.52	3000	3.3362	0.5552	0.5562
2.331	7.61	3500	3.3218	0.5459	0.5389
2.1295	8.7	4000	3.6107	0.5525	0.5532
1.9843	9.78	4500	3.4533	0.5575	0.5578
1.8472	10.87	5000	3.2933	0.5482	0.5469
1.7227	11.96	5500	3.3387	0.5513	0.5521
1.6067	13.04	6000	3.2725	0.5444	0.5454
1.5328	14.13	6500	3.3817	0.5513	0.5528
1.4166	15.22	7000	3.5382	0.5421	0.5437
1.346	16.3	7500	3.4353	0.5567	0.5574
1.3038	17.39	8000	3.5873	0.5478	0.5462
1.2285	18.48	8500	3.7322	0.5525	0.5516
1.1916	19.57	9000	3.5055	0.5486	0.5488
1.1143	20.65	9500	3.4413	0.5575	0.5589
1.0749	21.74	10000	3.7211	0.5559	0.5572
1.0668	22.83	10500	3.5802	0.5575	0.5576
1.0111	23.91	11000	3.5038	0.5606	0.5598
0.9837	25.0	11500	3.6704	0.5521	0.5517
0.9643	26.09	12000	3.5238	0.5598	0.5609
0.9311	27.17	12500	3.5195	0.5559	0.5558
0.902	28.26	13000	3.3760	0.5679	0.5653
0.8935	29.35	13500	3.6155	0.5536	0.5539
0.8745	30.43	14000	3.5108	0.5667	0.5662
0.8444	31.52	14500	3.6231	0.5606	0.5597
0.8327	32.61	15000	3.5783	0.5552	0.5508
0.8237	33.7	15500	3.5527	0.5556	0.5548
0.8035	34.78	16000	3.4553	0.5660	0.5657
0.7948	35.87	16500	3.4230	0.5490	0.5503
0.7886	36.96	17000	3.5010	0.5482	0.5494
0.7711	38.04	17500	3.4771	0.5644	0.5648
0.76	39.13	18000	3.5514	0.5563	0.5570
0.7509	40.22	18500	3.4726	0.5586	0.5585
0.7522	41.3	19000	3.5237	0.5606	0.5586
0.7368	42.39	19500	3.4514	0.5532	0.5516
0.7377	43.48	20000	3.5320	0.5633	0.5636
0.7142	44.57	20500	3.4685	0.5613	0.5608
0.7255	45.65	21000	3.4919	0.5652	0.5635
0.7139	46.74	21500	3.4869	0.5556	0.5551
0.7124	47.83	22000	3.4748	0.5644	0.5642
0.7065	48.91	22500	3.4405	0.5602	0.5601
0.7038	50.0	23000	3.4838	0.5505	0.5508

Framework versions

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