scenario-KD-PO-CDF-ALL-D2_data-cardiffnlp_tweet_sentiment_multilingual_all_alpha

This model is a fine-tuned version of haryoaw/scenario-TCR_data-cardiffnlp_tweet_sentiment_multilingual_all_a on the tweet_sentiment_multilingual dataset. It achieves the following results on the evaluation set:

• Loss: 3.4750
• Accuracy: 0.5637
• F1: 0.5640

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
4.905	1.09	500	4.5461	0.4363	0.4113
4.0059	2.17	1000	3.6093	0.5058	0.5040
3.4109	3.26	1500	3.4190	0.5208	0.5131
3.0676	4.35	2000	3.2675	0.5490	0.5477
2.7673	5.43	2500	3.2746	0.5467	0.5412
2.5421	6.52	3000	3.1951	0.5475	0.5367
2.3609	7.61	3500	3.3137	0.5432	0.5410
2.1176	8.7	4000	3.5963	0.5451	0.5303
1.9583	9.78	4500	3.5109	0.5571	0.5583
1.8268	10.87	5000	3.3664	0.5471	0.5477
1.7388	11.96	5500	3.3858	0.5517	0.5528
1.5976	13.04	6000	3.4404	0.5617	0.5577
1.4912	14.13	6500	3.3307	0.5586	0.5585
1.4157	15.22	7000	3.5579	0.5432	0.5355
1.3536	16.3	7500	3.3542	0.5617	0.5603
1.2883	17.39	8000	3.6026	0.5571	0.5543
1.2443	18.48	8500	3.6866	0.5478	0.5458
1.1637	19.57	9000	3.6125	0.5536	0.5547
1.1391	20.65	9500	3.5456	0.5613	0.5574
1.1029	21.74	10000	3.4366	0.5513	0.5526
1.0417	22.83	10500	3.6791	0.5586	0.5585
1.0169	23.91	11000	3.6637	0.5656	0.5607
1.0107	25.0	11500	3.5452	0.5575	0.5578
0.9502	26.09	12000	3.4362	0.5748	0.5742
0.9455	27.17	12500	3.4865	0.5694	0.5703
0.9194	28.26	13000	3.4523	0.5737	0.5716
0.9053	29.35	13500	3.5411	0.5586	0.5572
0.8737	30.43	14000	3.6550	0.5586	0.5586
0.865	31.52	14500	3.5079	0.5594	0.5611
0.8444	32.61	15000	3.4885	0.5509	0.5526
0.8343	33.7	15500	3.5705	0.5710	0.5698
0.8122	34.78	16000	3.4910	0.5521	0.5519
0.8161	35.87	16500	3.5302	0.5559	0.5563
0.7923	36.96	17000	3.5031	0.5656	0.5632
0.7824	38.04	17500	3.4182	0.5594	0.5592
0.7658	39.13	18000	3.5265	0.5594	0.5586
0.7588	40.22	18500	3.4465	0.5706	0.5711
0.7541	41.3	19000	3.4879	0.5540	0.5534
0.7488	42.39	19500	3.4246	0.5687	0.5693
0.7412	43.48	20000	3.4806	0.5745	0.5750
0.7314	44.57	20500	3.5638	0.5590	0.5586
0.7283	45.65	21000	3.4212	0.5664	0.5667
0.7179	46.74	21500	3.4444	0.5556	0.5560
0.7168	47.83	22000	3.4104	0.5602	0.5606
0.7161	48.91	22500	3.3766	0.5667	0.5676
0.7052	50.0	23000	3.4750	0.5637	0.5640

Framework versions

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