scenario-kd-from-scratch-silver-data-hate_speech_filipino-model-xlm-roberta-base

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

• Loss: 1.1354
• Accuracy: 0.7665
• F1: 0.7412

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: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6969

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	0.32	100	2.4557	0.6656	0.6952
No log	0.64	200	2.0812	0.7063	0.7186
No log	0.96	300	1.9137	0.7079	0.7300
No log	1.28	400	1.8171	0.7172	0.7401
2.5353	1.6	500	1.7305	0.7462	0.6959
2.5353	1.92	600	2.3251	0.6645	0.7221
2.5353	2.24	700	1.5004	0.7571	0.7299
2.5353	2.56	800	1.7161	0.7431	0.6752
2.5353	2.88	900	1.3750	0.7519	0.7400
1.5143	3.19	1000	1.6104	0.7561	0.6968
1.5143	3.51	1100	1.4419	0.7561	0.7104
1.5143	3.83	1200	1.3306	0.7450	0.7496
1.5143	4.15	1300	1.4285	0.7668	0.7352
1.5143	4.47	1400	1.3335	0.7576	0.7552
1.1029	4.79	1500	1.3649	0.7394	0.7487
1.1029	5.11	1600	1.5830	0.7224	0.7434
1.1029	5.43	1700	1.2794	0.7592	0.7560
1.1029	5.75	1800	1.2877	0.7547	0.7165
1.1029	6.07	1900	1.2428	0.7637	0.7325
0.8948	6.39	2000	1.2774	0.7387	0.7494
0.8948	6.71	2100	1.2324	0.7628	0.7354
0.8948	7.03	2200	1.3675	0.7387	0.7505
0.8948	7.35	2300	1.2021	0.7670	0.7490
0.8948	7.67	2400	1.3012	0.7682	0.7348
0.7714	7.99	2500	1.2338	0.7580	0.7210
0.7714	8.31	2600	1.2189	0.7628	0.7519
0.7714	8.63	2700	1.2962	0.7410	0.7526
0.7714	8.95	2800	1.3151	0.7675	0.7416
0.7714	9.27	2900	1.1539	0.7616	0.7528
0.7096	9.58	3000	1.3696	0.7561	0.7523
0.7096	9.9	3100	1.2055	0.7514	0.7533
0.7096	10.22	3200	1.1354	0.7665	0.7412

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.1
• Datasets 2.14.5
• Tokenizers 0.13.3