README.md

metadata

library_name: transformers
language:
  - mt
license: cc-by-nc-sa-4.0
base_model: MLRS/BERTu
model-index:
  - name: BERTu_taxi1500-mlt
    results:
      - task:
          type: text-classification
          name: Topic Classification
        dataset:
          type: taxi1500-mlt_Latn
          name: taxi1500
          config: mlt_Latn
        metrics:
          - type: f1
            args: macro
            value: 78.83
            name: Macro-averaged F1
        source:
          name: MELABench Leaderboard
          url: https://huggingface.co/spaces/MLRS/MELABench
extra_gated_fields:
  Name: text
  Surname: text
  Date of Birth: date_picker
  Organisation: text
  Country: country
  I agree to use this model in accordance to the license and for non-commercial use ONLY: checkbox

BERTu (Taxi1500 Maltese)

This model is a fine-tuned version of MLRS/BERTu on the Taxi1500 dataset. It achieves the following results on the test set:

• Loss: 0.9589
• F1: 0.7883

Intended uses & limitations

The model is fine-tuned on a specific task and it should be used on the same or similar task. Any limitations present in the base model are inherited.

Training procedure

The model was fine-tuned using a customised script.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 2
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: inverse_sqrt
• lr_scheduler_warmup_ratio: 0.005
• num_epochs: 200.0
• early_stopping_patience: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1
No log	1.0	54	1.3509	0.3101
No log	2.0	108	0.9772	0.5176
No log	3.0	162	0.8594	0.5698
No log	4.0	216	0.8265	0.5955
No log	5.0	270	0.8582	0.6333
No log	6.0	324	0.9635	0.6319
No log	7.0	378	1.0398	0.6322
No log	8.0	432	1.1267	0.6378
No log	9.0	486	1.2027	0.6437
0.4687	10.0	540	1.2412	0.6258
0.4687	11.0	594	1.2330	0.6517
0.4687	12.0	648	1.2771	0.6376
0.4687	13.0	702	1.3023	0.6315
0.4687	14.0	756	1.3134	0.6432
0.4687	15.0	810	1.3096	0.6498
0.4687	16.0	864	1.3191	0.6517
0.4687	17.0	918	1.3278	0.6725
0.4687	18.0	972	1.3554	0.6791
0.004	19.0	1026	1.3726	0.6641
0.004	20.0	1080	1.3710	0.6791
0.004	21.0	1134	1.3812	0.7017
0.004	22.0	1188	1.4026	0.6967
0.004	23.0	1242	1.4099	0.6950
0.004	24.0	1296	1.4162	0.6858
0.004	25.0	1350	1.4240	0.6932
0.004	26.0	1404	1.4322	0.6932
0.004	27.0	1458	1.3757	0.6903
0.0017	28.0	1512	1.4508	0.7084
0.0017	29.0	1566	1.4679	0.7084
0.0017	30.0	1620	1.4591	0.7084
0.0017	31.0	1674	1.4587	0.7084
0.0017	32.0	1728	1.4830	0.7084
0.0017	33.0	1782	1.4778	0.7084
0.0017	34.0	1836	1.4931	0.7084
0.0017	35.0	1890	1.5002	0.7084
0.0017	36.0	1944	1.5073	0.7084
0.0017	37.0	1998	1.5027	0.7084
0.0011	38.0	2052	1.4485	0.7004
0.0011	39.0	2106	1.5335	0.6895
0.0011	40.0	2160	1.5393	0.7084
0.0011	41.0	2214	1.5426	0.7084
0.0011	42.0	2268	1.5430	0.7084
0.0011	43.0	2322	1.5405	0.7017
0.0011	44.0	2376	1.5581	0.7084
0.0011	45.0	2430	1.5534	0.7011
0.0011	46.0	2484	1.5674	0.7084
0.0008	47.0	2538	1.5647	0.7084
0.0008	48.0	2592	1.5741	0.7084

Framework versions

• Transformers 4.51.1
• Pytorch 2.7.0+cu126
• Datasets 3.2.0
• Tokenizers 0.21.1

License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. Permissions beyond the scope of this license may be available at https://mlrs.research.um.edu.mt/.

Citation

This work was first presented in MELABenchv1: Benchmarking Large Language Models against Smaller Fine-Tuned Models for Low-Resource Maltese NLP. Cite it as follows:

@inproceedings{micallef-borg-2025-melabenchv1,
    title = "{MELAB}enchv1: Benchmarking Large Language Models against Smaller Fine-Tuned Models for Low-Resource {M}altese {NLP}",
    author = "Micallef, Kurt  and
      Borg, Claudia",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Findings of the Association for Computational Linguistics: ACL 2025",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.findings-acl.1053/",
    doi = "10.18653/v1/2025.findings-acl.1053",
    pages = "20505--20527",
    ISBN = "979-8-89176-256-5",
}