bert-base-multilingual-uncased-sentiment_v3

This model is a fine-tuned version of nlptown/bert-base-multilingual-uncased-sentiment on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4987
• Accuracy: 0.9286
• Precision Macro: 0.8226
• Recall Macro: 0.7931
• F1 Macro: 0.8061
• F1 Weighted: 0.9269

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: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	F1 Weighted
0.3933	1.0	90	0.2349	0.9292	0.8484	0.7197	0.7474	0.9202
0.2051	2.0	180	0.2166	0.9236	0.8134	0.7619	0.7811	0.9199
0.1494	3.0	270	0.2369	0.9362	0.8619	0.7775	0.8072	0.9321
0.1233	4.0	360	0.2290	0.9343	0.8660	0.7894	0.8176	0.9309
0.0838	5.0	450	0.2490	0.9375	0.8610	0.8200	0.8378	0.9358
0.0799	6.0	540	0.2579	0.9343	0.8528	0.7977	0.8197	0.9317
0.0481	7.0	630	0.3494	0.9223	0.7926	0.8252	0.8064	0.9247
0.0406	8.0	720	0.3154	0.9368	0.8591	0.7986	0.8227	0.9341
0.032	9.0	810	0.3219	0.9305	0.8238	0.8153	0.8194	0.9301
0.0333	10.0	900	0.3787	0.9286	0.8387	0.8048	0.8198	0.9270
0.0278	11.0	990	0.3914	0.9311	0.8432	0.7948	0.8148	0.9288
0.0165	12.0	1080	0.4155	0.9318	0.8627	0.7830	0.8120	0.9282
0.0126	13.0	1170	0.4029	0.9368	0.8550	0.8161	0.8328	0.9352
0.0133	14.0	1260	0.4398	0.9324	0.8460	0.7915	0.8134	0.9297
0.01	15.0	1350	0.4571	0.9318	0.8347	0.7913	0.8094	0.9294
0.008	16.0	1440	0.4685	0.9299	0.8303	0.7899	0.8070	0.9276
0.0058	17.0	1530	0.4846	0.9318	0.8403	0.7954	0.8142	0.9295
0.0022	18.0	1620	0.4905	0.9280	0.8249	0.7928	0.8068	0.9262
0.0038	19.0	1710	0.5043	0.9299	0.8272	0.7897	0.8057	0.9277
0.0015	20.0	1800	0.4987	0.9286	0.8226	0.7931	0.8061	0.9269

Framework versions

• Transformers 4.55.0
• Pytorch 2.7.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4