NAMAA Community's 6 fine-tuned seq2seq backbones for DialectSentEval 2026 Subtask 2 (Arabic Dialect Sentiment Swap). 96.78 dev / 0.7597 test.
🏗️ Building on HF
Hassan
HassanB4
AI & ML interests
AI, ML, NLP, RL, Gen AI.
Recent Activity
updated a model about 7 hours ago
HassanB4/warfarisk-baselines updated a model about 7 hours ago
HassanB4/warfarisk-autogluon-1780 updated a model about 7 hours ago
HassanB4/warfarisk-autogluon-6256Organizations
-
HassanB4/halluscoring-camelbert-qa
Text Classification • 0.1B • Updated • 8 -
HassanB4/halluscoring-camelbert-nli
Text Classification • 0.1B • Updated • 10 -
HassanB4/halluscoring-marbert-nli
Text Classification • 0.2B • Updated • 14 -
HassanB4/halluscoring-arbert-nli
Text Classification • 0.2B • Updated • 9
-
HassanB4/Ketaba-OCR-LoRA
Image-to-Text • Updated • 249 • 7 -
HassanB4/Qari-OCR-LoRA
Image-Text-to-Text • Updated • 8 • 1 -
sherif1313/Arabic-English-handwritten-OCR-v3
Image-Text-to-Text • 4B • Updated • 1.65k • 16 -
sherif1313/Arabic-handwritten-OCR-4bit-Qwen2.5-VL-3B-v2
Image-to-Text • 4B • Updated • 13.8k • 13
Reproducible ML pipeline for warfarin dosing. Best MAE 8.55mg/wk, R2=0.475. Reveals a per-ancestry calibration gap masked in aggregate.
-
HassanB4/t1_s1_marbert_text_target
Text Classification • 0.2B • Updated • 8 -
HassanB4/t1_s2_bert-large-arabertv02-twitter_text_target
Text Classification • 0.4B • Updated • 12 -
HassanB4/t1_s3_arabert_twitter_text_target
Text Classification • 0.1B • Updated • 10 -
HassanB4/t1_s4_camelbert_text_target
Text Classification • 0.1B • Updated • 9
9 models + 1 dataset for detecting cultural hallucinations in Arabic LLM outputs. ICAIRE 2026 Track 3.
NAMAA Community's 6 fine-tuned seq2seq backbones for DialectSentEval 2026 Subtask 2 (Arabic Dialect Sentiment Swap). 96.78 dev / 0.7597 test.
Reproducible ML pipeline for warfarin dosing. Best MAE 8.55mg/wk, R2=0.475. Reveals a per-ancestry calibration gap masked in aggregate.
-
HassanB4/t1_s1_marbert_text_target
Text Classification • 0.2B • Updated • 8 -
HassanB4/t1_s2_bert-large-arabertv02-twitter_text_target
Text Classification • 0.4B • Updated • 12 -
HassanB4/t1_s3_arabert_twitter_text_target
Text Classification • 0.1B • Updated • 10 -
HassanB4/t1_s4_camelbert_text_target
Text Classification • 0.1B • Updated • 9
-
HassanB4/halluscoring-camelbert-qa
Text Classification • 0.1B • Updated • 8 -
HassanB4/halluscoring-camelbert-nli
Text Classification • 0.1B • Updated • 10 -
HassanB4/halluscoring-marbert-nli
Text Classification • 0.2B • Updated • 14 -
HassanB4/halluscoring-arbert-nli
Text Classification • 0.2B • Updated • 9
9 models + 1 dataset for detecting cultural hallucinations in Arabic LLM outputs. ICAIRE 2026 Track 3.
-
HassanB4/Ketaba-OCR-LoRA
Image-to-Text • Updated • 249 • 7 -
HassanB4/Qari-OCR-LoRA
Image-Text-to-Text • Updated • 8 • 1 -
sherif1313/Arabic-English-handwritten-OCR-v3
Image-Text-to-Text • 4B • Updated • 1.65k • 16 -
sherif1313/Arabic-handwritten-OCR-4bit-Qwen2.5-VL-3B-v2
Image-to-Text • 4B • Updated • 13.8k • 13