Developing for Indic languages | Gemma and Navarasa (Extended Edition)

No video

Developing for Indic languages | Gemma and Navarasa (Extended Edition)

Рет қаралды 4,838

Күн бұрын

While many early large language models were predominantly trained on English language data, the field is rapidly evolving. Newer models are increasingly being trained on multilingual datasets, and there's a growing focus on developing models specifically for the world’s languages. However, challenges remain in ensuring equitable representation and performance across diverse languages, particularly those with less available data and computational resources.
Gemma, Google's family of open models, is designed to address these challenges by enabling the development of projects in non-Germanic languages. Its tokenizer and large token vocabulary make it particularly well-suited for handling diverse languages. Watch how developers in India used Gemma to create Navarasa - a fine-tuned Gemma model for Indic languages.
Subscribe to Google for Developers → goo.gle/develo...
#GoogleIO #GoogleIO2024

Пікірлер: 13

@sn5953 2 ай бұрын

I don't understand how the creators of this video missed including a short clip of Navarasa/Gemma in action in a conversation involving an Indic language.

@glenncameronjr 2 ай бұрын

That is in the video starting at 1:38

@sn5953 2 ай бұрын

@@glenncameronjr I was expecting a short video clip of speech recognition of a spoken query in an Indic language followed by an audio response by Navarasa/Gemma in the same language. We know all these parts are technically feasible individually, just that it would've been nice to see all of these in action together demonstrating what Navarasa/Gemma brings to the table.

@Sonu.Singh.28 2 ай бұрын

It would be interesting to see if you guys pull this off. Because there are some languages spoken by atleast 10 million people but the languages don't even have a written grammar or literature.