Context:
The BharatGPT group, headed by IIT Bombay and seven other prestigious Indian engineering institutes, revealed plans to introduce its own ChatGPT-like service.
Hanooman: By Seetha Mahalaxmi Healthcare (SML) and the BharatGPT
- Seetha Mahalaxmi Healthcare (SML) and the BharatGPT recently introduced the ‘Hanooman’ series of Indic language models.
What are LLMs?
- Large language models (LLMs) use deep learning techniques to understand extensive text data.
- Their function involves processing large volumes of text, understanding its structure and significance, and deriving insights from it.
- LLMs undergo training to recognize meanings and correlations among words.
- It enhances their ability to understand and generate text.
- The effectiveness of LLMs improves as they receive more training data.
- Training data for LLMs typically comprises Wikipedia, OpenWebText, and the Common Crawl Corpus.
- These datasets consist of large text information that the models use to understand and produce human-like language.
|
What is Hanooman?
Hanooman is a series of large language models (LLMs).
- It can communicate in 11 Indian languages like Hindi, Tamil, and Marathi.
- The size of these AI models varies from 1.5 billion to a massive 40 billion parameters.
Features and Applications
- Hanooman is not just a chatbot but a multimodal AI tool.
- It can produce text, speech, videos, and more in multiple Indian languages.
- It has been designed to serve four key areas: healthcare, governance, financial services, and education.
Customized Versions:
- One of the customized versions is VizzhyGPT, fine-tuned specifically for healthcare.
- It uses extensive medical data to enhance its performance.
Benefits and Challenges of AI models
| Benefits |
Challenges |
- Boosted Efficiency: LLMs help save time by doing language tasks like translating, summarising, and creating content automatically, which makes work smoother and faster.
- Support for Different Languages: LLMs can understand and work with many languages and dialects which makes it easier for people from different language backgrounds to communicate.
- Ease of discovering Insights: LLMs can look at big amounts of text data and find important information and trends which help researchers and businesses learn more from their data
|
- Quality of Datasets: There are issues of quality datasets in the Indian language as this AI model can communicate in 11 languages.
- High probability of Inaccuracy: The synthetic datasets generated artificially can create inaccurate answers.
- Bias and Fairness: LLMs may unknowingly carry biases from the data they learn from, resulting in unfair treatment towards certain groups or people.
- Security Risks: LLMs might be vulnerable to attacks where bad inputs are used to trick them into giving wrong results.
|
Also Read: Global Partnership On Artificial Intelligence – GPAI
News Source: Indianexpress
February 26, 2024