AraMix is a SOTA Arabic pretraining dataset
Sultan Alrashed PRO
SultanR
AI & ML interests
Smol language modelling and Arabic!
Recent Activity
liked a dataset 3 days ago
HuggingFaceCode/stack-v3-train liked a model 4 days ago
HuggingFaceTB/nanowhale-100m liked a dataset 5 days ago
scholarweave/arxiv-latexOrganizations
A collection of models that use SmolLM2 as the pretrained base in conjunction with AllenAI's Tulu 3 post training pipeline.
-
SmolTulu: Higher Learning Rate to Batch Size Ratios Can Lead to Better Reasoning in SLMs
Paper • 2412.08347 • Published • 4 -
SultanR/SmolTulu-1.7b-Reinforced
Text Generation • 2B • Updated • 10 • 5 -
SultanR/SmolTulu-1.7b-Instruct
Text Generation • 2B • Updated • 81 • 13 -
SultanR/SmolTulu-1.7b-RM
Text Classification • 2B • Updated • 6 • 2
AraMix is a SOTA Arabic pretraining dataset
Largest (as of 2024) machine translated Arabic educational corpus
A collection of models that use SmolLM2 as the pretrained base in conjunction with AllenAI's Tulu 3 post training pipeline.
-
SmolTulu: Higher Learning Rate to Batch Size Ratios Can Lead to Better Reasoning in SLMs
Paper • 2412.08347 • Published • 4 -
SultanR/SmolTulu-1.7b-Reinforced
Text Generation • 2B • Updated • 10 • 5 -
SultanR/SmolTulu-1.7b-Instruct
Text Generation • 2B • Updated • 81 • 13 -
SultanR/SmolTulu-1.7b-RM
Text Classification • 2B • Updated • 6 • 2