Course Description
Mastering Multilingual GenAI – Open-Weights for Indic Languages" is a course designed to equip you with the knowledge to develop state-of-the-art multilingual AI models using open-weight architectures. Focusing on low-resource languages, particularly Indic languages, the course covers essential topics like multilingual AI training, instruction fine-tuning, model building, and performance evaluation.
Course curriculum
-
1
Mastering Multilingual GenAI
- Introduction
- Importance of Multilingual
- Training for Multilingual Gen AI
- Instruction Fine-Tuning Data for Multilingual
- Measuring Performance for Multilingual
- Building a Model
- Human Preferences
- Curse of Multilinguality
- Coding Hands-On
Certificate of Completion
Who should Enroll?
-
Those looking to build state-of-the-art multilingual models for low-resource languages.
-
Tech Entrepreneurs and Innovators who want to build scalable, inclusive AI systems that cater to a diverse, multilingual user base.
-
Data Scientists seeking to explore state-of-the-art instruction fine-tuning techniques and work with large multilingual datasets.
-
Students and Professionals aspiring to work in cutting-edge AI fields, with a specific interest in bias mitigation, safety, and ethical considerations in language models.
About the Instructor
Viraat Aryabumi - Research Scholar at Cohere for AI
FAQs
-
What prior knowledge is required for this course?
A basic understanding of AI and machine learning concepts is recommended. Familiarity with natural language processing (NLP) will be helpful but is not mandatory.
-
What tools and technologies will I learn in this course?
You'll work with state-of-the-art open-weight models, instruction fine-tuning techniques, and the Aya dataset.
-
How does this course address low-resource languages?
The course focuses on training models that perform well on low-resource languages, particularly Indic languages, through the use of the Aya dataset, multilingual instruction fine-tuning, and bias mitigation techniques.
-
Will there be hands-on projects?
Yes, the course includes practical coding exercises and real-world projects where you will build and fine-tune multilingual AI models, with a focus on tackling real-world challenges in low-resource language modeling.
Key Takeaways
-
Learn to create AI models for diverse languages, focusing on low-resource ones.
-
Get hands-on experience with cutting-edge models and multilingual data.
-
Implement strategies for safer, unbiased AI models in global applications.