README.md

metadata

language: ar
license: apache-2.0
library_name: peft
base_model: UBC-NLP/MARBERT
tags:
  - arabic
  - dialect-classification
  - lora

HammaLoRAMarBert

Advanced Arabic Dialect Classification Model with Complete Training Metrics

Full Training History

epoch	train_loss	eval_loss	train_accuracy	eval_accuracy	f1	precision	recall
1	1.01756	1.0054	0.70748	0.717978	0.693725	0.706778	0.70748
2	0.762952	0.75223	0.771853	0.78764	0.771604	0.778861	0.771853
3	0.650689	0.648891	0.796329	0.803371	0.797666	0.801681	0.796329
4	0.622925	0.626332	0.801449	0.811798	0.801765	0.80837	0.801449
5	0.576898	0.588152	0.809815	0.812921	0.810793	0.814344	0.809815
6	0.567929	0.60128	0.814623	0.810674	0.816486	0.823517	0.814623
7	0.556496	0.58585	0.818244	0.820225	0.818915	0.822701	0.818244
8	0.54978	0.592384	0.821054	0.820225	0.82197	0.82844	0.821054
9	0.543711	0.587352	0.824301	0.816854	0.826151	0.83428	0.824301
10	0.51674	0.565089	0.830607	0.818539	0.831944	0.83726	0.830607
11	0.520477	0.580509	0.830669	0.819663	0.832265	0.837997	0.830669
12	0.507471	0.563466	0.833729	0.82809	0.834758	0.839029	0.833729
13	0.498436	0.557207	0.834603	0.825281	0.835891	0.840618	0.834603
14	0.496213	0.551106	0.836289	0.828652	0.837213	0.840592	0.836289
15	0.493182	0.549526	0.836414	0.826404	0.837405	0.840693	0.836414

Label Mapping:

{0: 'Egypt', 1: 'Iraq', 2: 'Lebanon', 3: 'Morocco', 4: 'Saudi_Arabia', 5: 'Sudan', 6: 'Tunisia'}

USAGE Example:

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="Hamma-16/HammaLoRAMarBert",
    device="cuda" if torch.cuda.is_available() else "cpu"
)

sample_text = "شلونك اليوم؟"
result = classifier(sample_text)
print(f"Text: {sample_text}")
print(f"Predicted: {result[0]['label']} (confidence: {result[0]['score']:.1%})")