Arabic NLU and generation tasks, Summarization, Interpretability analyses
Limitations:
Not suited for generalization across all languages
Considerations:
Cultural adaptation is required for use cases in different regions.
Additional Notes
The model implements various cultural adaptations to improve performance in Arabic-speaking contexts.
Supported Languages
Arabic (High Proficiency), English (Strong Capabilities)
Training Details
Data Sources:
web pages, wikipedia articles, news articles, social network content, code in various languages, books, scientific papers, synthetic data using machine translation
Data Volume:
1.6T tokens
Methodology:
From scratch and adapted pre-training
Context Length:
2048
Hardware Used:
64 Cerebras CS-2 Wafer-Scale Engines (WSE-2)
Model Architecture:
Auto-regressive transformer-based, decoder-only (GPT-3) architecture with SwiGLU activation and ALiBi position encoding
Note: green Score (e.g. "73.2") means that the model is better than inceptionai/jais-family-1p3b.
Rank the Jais Family 1p3b Capabilities
๐ Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! ๐
Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation
What open-source LLMs or SLMs are you in search of? 52721 in total.