XLM-EusBERTa-sentiment-classification

This model is a fine-tuned version of ClassCat/roberta-small-basque on the basque_glue dataset. It achieves the following results on the evaluation set:

• Loss: 4.0012
• Accuracy: 0.6290
• F1: 0.6291
• Precision: 0.6303
• Recall: 0.6290

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• 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	Precision	Recall
No log	1.0	380	0.7366	0.6736	0.6589	0.6711	0.6736
0.7679	2.0	760	0.7654	0.6767	0.6692	0.6726	0.6767
0.4846	3.0	1140	0.9844	0.6621	0.6599	0.6681	0.6621
0.2952	4.0	1520	1.1162	0.6375	0.6371	0.6473	0.6375
0.2952	5.0	1900	1.4234	0.6329	0.6343	0.6425	0.6329
0.192	6.0	2280	1.8570	0.6413	0.6362	0.6424	0.6413
0.159	7.0	2660	2.1968	0.6152	0.6086	0.6152	0.6152
0.1265	8.0	3040	2.1853	0.6283	0.6267	0.6267	0.6283
0.1265	9.0	3420	2.1953	0.6467	0.6441	0.6435	0.6467
0.0807	10.0	3800	2.2806	0.6367	0.6381	0.6480	0.6367
0.0688	11.0	4180	2.7982	0.6175	0.6167	0.6356	0.6175
0.0675	12.0	4560	2.5182	0.6605	0.6587	0.6584	0.6605
0.0675	13.0	4940	2.6544	0.6413	0.6315	0.6391	0.6413
0.0451	14.0	5320	2.5889	0.6459	0.6427	0.6424	0.6459
0.0432	15.0	5700	2.8100	0.6290	0.6299	0.6359	0.6290
0.0297	16.0	6080	2.9983	0.6275	0.6262	0.6263	0.6275
0.0297	17.0	6460	2.7803	0.6313	0.6289	0.6311	0.6313
0.0369	18.0	6840	2.9602	0.6283	0.6287	0.6353	0.6283
0.0289	19.0	7220	2.9911	0.6298	0.6309	0.6356	0.6298
0.0251	20.0	7600	2.8634	0.6344	0.6350	0.6364	0.6344
0.0251	21.0	7980	2.7171	0.6406	0.6378	0.6375	0.6406
0.0332	22.0	8360	3.0386	0.6275	0.6215	0.6245	0.6275
0.0212	23.0	8740	2.9876	0.6313	0.6319	0.6344	0.6313
0.0218	24.0	9120	2.9776	0.6283	0.6267	0.6348	0.6283
0.0189	25.0	9500	2.9596	0.6329	0.6340	0.6381	0.6329
0.0189	26.0	9880	3.0420	0.6329	0.6324	0.6380	0.6329
0.0172	27.0	10260	3.3335	0.6336	0.6348	0.6369	0.6336
0.0054	28.0	10640	3.2843	0.6429	0.6442	0.6466	0.6429
0.0065	29.0	11020	3.4868	0.6275	0.6291	0.6399	0.6275
0.0065	30.0	11400	3.8241	0.6175	0.6174	0.6209	0.6175
0.0108	31.0	11780	3.5833	0.6260	0.6275	0.6317	0.6260
0.0127	32.0	12160	3.5452	0.6183	0.6203	0.6283	0.6183
0.0092	33.0	12540	3.8349	0.6167	0.6167	0.6389	0.6167
0.0092	34.0	12920	3.6464	0.6244	0.6260	0.6313	0.6244
0.0069	35.0	13300	3.7538	0.6352	0.6352	0.6359	0.6352
0.0028	36.0	13680	3.8862	0.6221	0.6243	0.6350	0.6221
0.0001	37.0	14060	3.9846	0.6229	0.6206	0.6252	0.6229
0.0001	38.0	14440	3.7743	0.6275	0.6287	0.6309	0.6275
0.0057	39.0	14820	3.9002	0.6290	0.6300	0.6319	0.6290
0.0004	40.0	15200	3.9651	0.6306	0.6315	0.6333	0.6306
0.0032	41.0	15580	4.0279	0.6206	0.6213	0.6365	0.6206
0.0032	42.0	15960	3.8244	0.6344	0.6342	0.6344	0.6344
0.0033	43.0	16340	3.9036	0.6198	0.6205	0.6237	0.6198
0.003	44.0	16720	4.0028	0.6198	0.6214	0.6263	0.6198
0.0005	45.0	17100	3.9621	0.6306	0.6315	0.6361	0.6306
0.0005	46.0	17480	3.9682	0.6306	0.6297	0.6298	0.6306
0.0003	47.0	17860	4.0103	0.6321	0.6310	0.6305	0.6321
0.0003	48.0	18240	3.9968	0.6321	0.6316	0.6317	0.6321
0.003	49.0	18620	3.9835	0.6298	0.6297	0.6304	0.6298
0.0005	50.0	19000	4.0012	0.6290	0.6291	0.6303	0.6290

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.0
• Tokenizers 0.15.0