@hdidwania1997 Profile picture

Himansu Didwania

@hdidwania1997

Founding Team @SarvamAI. Currently giving voice to our AI agents. Interested in Machine Learning, technology and entrepreneurship.

Himansu Didwania Reposted

🧵 (1/9) Excited to announce Sarvam 1, India's first home-grown Large Language Model! A 2B parameter model trained on 10 Indic languages + English, representing a major leap in Indian language AI. Built entirely in India, for India. 🇮🇳

Say hello to Sarvam-1! sarvam.ai/blogs/sarvam-1



Himansu Didwania Reposted

Tomorrow at 3 PM! Don’t miss Sarvam’s co-founder @pratykumar discussing “Building GenAI for a Billion Voices” at the @NVIDIAAI Summit in Lotus Ballroom 1. Also, come visit us at our booth in the expo hall - GEN AI 3. #GenAI #AISummit2024 #NVIDIA #SarvamAI

Tweet Image 1

Himansu Didwania Reposted

Interested in joining our team @SarvamAI and building Indic LLMs? Apply here: forms.gle/6qXN3SkVXP4qdD… Bonus points for clean code!


Himansu Didwania Reposted

🆕Read about the unique , code-mixed approach we took to build our Indic, colloquial translation model Mayura: sarvam.ai/blogs/blog-may… Now available in 10 languages here: dashboard.sarvam.ai So who's building the next-gen, multilingual news app for Bharat? 😉


Himansu Didwania Reposted

🔔 Introducing Bulbul v1 - our best in class, multilingual and colloquial text to speech model in 6 unique Indian voices! Population scale use cases require reliable, comprehensible, and performant speech models that can truly meet where the users are.


Learn more about Bulbul v1 - our TTS model. Built to be reliable for multiple usecases, work for Indian context.

AI that mimics human? (naaah) AI that sounds pleasing, is reliable, does the job? (Yeees) Introducing Bulbul v1- our multi-lingual, colloquial text to speech model that is not just available in 6 unique Indian voices but also doesn't trip over terms and entities that matter.



Himansu Didwania Reposted

Namaskāra friends! Thank you for all the love this past one week! We heard you. We are now on Discord: discord.gg/8ka56wQaT3 😎 Please join the community to learn more about our products, ideate with our team, give feedback/ report issues! See you there! 👋


Really grateful to be a part of this. And we're just getting started.

Introducing Bulbul 1.0! best in class, in-house Text-to-speech model, supporting 10 languages in 6 voices.

Tweet Image 1


So glad to have worked on this, and as an Odia, push for Odia representation in the GenAI space. Hope this unlocks wide range of applications in Odia.

Bulbul: Indic TTS models, domain specific, customer specific, different accents, languages, more! So glad to hear some really convincing Odia AI agent voice for the first time!

Tweet Image 1


Himansu Didwania Reposted

Our new website experience is now live! sarvam.ai


Almost here. Launch in 10...9...8...7.......

Going live soon!

Tweet Image 1


Himansu Didwania Reposted

Urgent! The hazardous potholes on Mansapola - NC College Road (in front of Maa Biraja Dental Clinic) pose a serious threat and can cause fatal accidents. Requesting @Jajpurmun @dm_jajpur @pavankalyan to take immediate action for our citizens! #SafetyFirst #Odisha #potholes

Tweet Image 1
Tweet Image 2
Tweet Image 3

Sarvam.ai has exciting open positions in the Speech Team. We're looking for ML Engineers and Data Engineers to assist us with our goal of building voice-first GenAI applications for India. Check this LinkedIn post for more details: linkedin.com/posts/sarvam-a…


Himansu Didwania Reposted

🚨 New SFT data 🚨 100k high-quality, multi-turn conversations (>700k turns) in English, Hindi, and Hinglish curated exclusively with an Indic context. 🚀 huggingface.co/datasets/sarva…


Himansu Didwania Reposted

@SarvamAI, India’s first provider of LLMs that offer Indic language support, is leveraging #MicrosoftCloud for developers to build real-time voice-based #GenAI apps at scale. #MicrosoftAITour #MSAITour


Wake up, a new open-source LLM just dropped. Proud to announce that @SarvamAI has released OpenHathi-Hi-v0.1.

🚨New model alert🚨 We're super excited to release OpenHathi-Hi-v0.1, the first Hindi LLM from our OpenHathi series of models. This model is trained under compute and data constraints to show that we can get GPT-3.5-like performance on Indic languages with a frugal budget. 1/5



United States Trends
Loading...

Something went wrong.


Something went wrong.