scenario-non-kd-from-post-finetune-div-2-data-hate_speech_filipino-model-haryoaw

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

• Loss: 1.3495
• Accuracy: 0.7779
• F1: 0.7601

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	0.5751	0.7339	0.7510
No log	0.64	200	0.5171	0.7717	0.7450
No log	0.96	300	0.4969	0.7784	0.7539
No log	1.28	400	0.5408	0.7819	0.7548
0.4175	1.6	500	0.5410	0.7635	0.7597
0.4175	1.92	600	0.5166	0.7767	0.7324
0.4175	2.24	700	0.5823	0.7651	0.7450
0.4175	2.56	800	0.5731	0.7672	0.7293
0.4175	2.88	900	0.6860	0.7769	0.7458
0.2936	3.19	1000	0.7409	0.7684	0.7659
0.2936	3.51	1100	0.6544	0.7772	0.7487
0.2936	3.83	1200	0.6719	0.7604	0.7613
0.2936	4.15	1300	0.8242	0.7781	0.7471
0.2936	4.47	1400	0.8741	0.7838	0.7472
0.199	4.79	1500	0.7415	0.7755	0.7509
0.199	5.11	1600	0.9389	0.7897	0.7615
0.199	5.43	1700	0.7985	0.7840	0.7693
0.199	5.75	1800	0.9223	0.7741	0.7600
0.199	6.07	1900	1.0076	0.7727	0.7667
0.1553	6.39	2000	0.8541	0.7800	0.7682
0.1553	6.71	2100	0.9460	0.7810	0.7600
0.1553	7.03	2200	1.0575	0.7791	0.7571
0.1553	7.35	2300	1.0487	0.7687	0.7657
0.1553	7.67	2400	0.8495	0.7732	0.7568
0.1316	7.99	2500	0.9467	0.7812	0.7658
0.1316	8.31	2600	1.0491	0.7722	0.7611
0.1316	8.63	2700	1.0363	0.7691	0.7275
0.1316	8.95	2800	0.9130	0.7758	0.7448
0.1316	9.27	2900	1.4607	0.7717	0.7562
0.1137	9.58	3000	1.1874	0.7727	0.7567
0.1137	9.9	3100	1.1752	0.7836	0.7672
0.1137	10.22	3200	1.3495	0.7779	0.7601

Framework versions

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