docs2.two.ai Open in urlscan Pro
34.132.239.214  Public Scan

Form analysis
0 forms found in the DOM

Text Content

DocumentationAPI Reference
Version 1

PlaygroundTWO.AI
Find Anything

 * Getting Started
   
   

 *  
   
   Updates
   
   

Dark Mode



GETTING STARTED

Start writing your API Documentation with our platform

API Documentation is an easTTy to use platform that helps collaborate on writing
industry leading documentation so you can:

Hello, How are you?



 1. On-board users quicker with our machine learning assistant

 2. Keep your docs on-brand, so it doesn't feel like a 3rd party tool

 3. Let your users quickly interact and test your APIs

No need to take our word on this, we let our amazing customer docs speak for
themselves :)


ONBOARDING COPIED!

Here's an onboarding video from our CEO that showcases our product and machine
learning assistant in action:

/

/








WHAT IS SUTRA? COPIED!


INTRODUCTION COPIED!

SUTRA (SUpervised TRaining and Alignment) is a family of large multi-lingual
language (LMLMs) models pioneered by Two Platforms. They are trained and
fine-tuned with in-house data of 5M+ (and growing) conversations. They undergo
rigorous preference optimization with user feedback to adapt and align with user
behavior to provide state-of-the-art conversational experiences.

SUTRA is a Sanskrit word signifying a thread or aphorism, encapsulating concise
wisdom. Currently, there are three SUTRA model series:

 * SUTRA-Search : Online language models that can use knowledge from the
   internet to provide the most up-to-date and factual responses.

 * SUTRA-Multilingual : Series of language models that can support text
   generation and instruction following in 50+ languages.

 * SUTRA-Avatar : Model that brings the power of generative ML algorithms to
   enable real-time lifelike visual and voice synthesis.


DRAWBACKS OF CURRENT LLMS COPIED!

Leading LLMs like Llama-2, GPT-3/4 and Mistral have gained popularity in recent
years, but they have strong English language focus - from their training data to
applications. The generative capabilities of these models is significantly worse
in non-Romanized languages like Hindi, Korean, Chinese and others. Some of the
core architectural components like Tokenizers are built for English, they are
inefficient and ineffective to capture core sentiments, due to their limited
vocabulary. For applications such as ZAPPY (TWO's consumer AI application),
where user interaction is frequent, relying on standard models like GPT can
result in substantial API usage and associated costs.


WHY CHOOSE SUTRA? COPIED!

To tackle these challenges, we have developed LLMs from ground-up - we train
with a proprietary multi-language dataset, tokenizers, and models that capture
languages like English, Hindi, Tamil, Korean and are supported by our advanced
AI infrastructure.

 1. 

Next Page

Updates



ON THIS PAGE

   
 * Onboarding
 * What is SUTRA?
 * Introduction
 * Drawbacks of Current LLMs
 * Why Choose SUTRA?

Last updated March 01, 2024

PRIVACY TERMS
© 2024 TWO.