Google's Tiny model beats GPT-3.5

PLUS : Create your own AI personal avatar

August 03, 2024 • Estimated Reading Time: 8 minutes

Greetings, AI Enthusiasts!

Google’s new Gemma 2 family of models outperforms GPT-3.5.

Google poached AI startup founders.

Create your own AI personal avatars in two minutes.

With the new launch of “The Bytes” section in our newsletter edition going forward.

Lets explore!

This week’s News and Insights:

  • GitHub launches GitHub models.

  • Google’s new Gemma 2 family of models

  • OpenAI’s new advanced voice mode.

  • Google’s “Gemini AI” AD gone wrong.

  • Create your own AI personal avatar.

  • PLUS more.

GitHub launches GitHub models

Image Credit : Github

GitHub is no stranger to developers. Now with GitHub models, Developers will be empowered with Generative AI models.

The Essentials:

  • Developers can access a catalogue of AI models within the GitHub platform.

  • AI models like Llama 3.1, GPT-4 and Phi-3 are available.

  • With just one account, it reduces friction and integration. Developers can now access and test multiple AI models on just one platform.

  • From experimentation to production deployments via Microsoft Azure is made easy.

  • A dedicated playground to compare different models. Use-Case: The same prompt can be tested across different AI models for quality and performance.

The Takeaway:

GitHub models are just a one-stop place for Enterprise developers to test and play with different AI models with just one account.

Google’s new Gemma 2 family of models

Image Credit : Google

Google has introduced new models to its Gemma 2 family of models. These models are aimed towards performance, safety and transparency

The Essentials:

  • Gemma 2 2B: A tiny model which outperforms GPT-3.5 models in the Chatbot arena.

    • Use-Case : Smartphones can be powered with Gemma 2 2B for an On-device AI chatbot.

  • ShieldGemma: A safety content classifier, which is trained to detect and filter content like toxicity, s*xual content, violence, and hate speech.

    • Use-Case : Social media platforms can integrate ShieldGemma to detect and filter harmful content.

  • Gemma Scope: A model interpretability tool which provides insights into the decision-making processes of Gemma 2 models.

    • Use-Case : Gemma Scope can be used to analyze medical diagnosis AI model. This could help to identify and address potential biases to improve the model's fairness and accuracy.

  • Gemma 2's model weights can be downloaded from Kaggle, Hugging Face and Vertex AI Model Garden and with capabilities that can be tested in Google AI Studio.

The Takeaway:

Gemma 2 new models enhance the capabilities of AI systems. Users can now have access to a powerful set of tools with which AI can be made safer and more transparent.

OpenAI’s new advanced voice mode

Image Credit : OpenAI

OpenAI advanced voice mode for ChatGPT demonstrates extraordinary capabilities. This sets a new benchmark for AI-powered voice interactions.

The Essentials:

  • OpenAI's new advanced voice mode can: Sing, Imitate accents, Language coaching and storytelling.

  • It demonstrates complex tasks like telling a story as like an Airline pilot through an intercom.

  • Currently available to a small group of ChatGPT Plus subscribers.

  • In Fall 2024, OpenAI plans to make this advanced voice mode available to all ChatGPT Plus subscribers ($20/month)

"ChatGPT Advanced Voice Mode counting as fast as it can to 10, then to 50 (this blew my mind—it stopped to catch its breath like a human would)," wrote tech writer Cristiano Giardina on X.

The Takeaway:

The advanced voice mode has taken AI-powered voice interactions to different heights. The interaction is seamless and it is nearly like a perfect conversation with a human.

Interesting News :

Google’s “Gemini AI” AD gone wrong

Image Credit : Google

  • Google's recent AI advertisement for Gemini AI during the Olympics has faced severe backlash. In the ad, a father describes "Gemini AI" as writing a letter to his daughter's Olympic hero in "how inspiring is she for her". Many viewers believe that using AI destroys the child's creativity.

  • Google missed the audience beat with this ad.

Google poached CharacterAI founders

Image Credit : Character AI founders -Photo by Winnie WinterMayer via Getty Images.

  • As part of the licensing deal, Google has hired Character AI founders “Noam Shazeer and Daniel De Freitas”.

  • With this deal, Google plans to use Character AI technology to create chatbots that can mimic real or fictional characters. Despite this poaching, Character AI will operate independently.

Create your own AI personal avatar

Image Credit : Synthesia

  • Synthesia AI has launched AI personal avatars.

  • Users can create their own avatar from a two-minute footage of themselves.

  • These avatars can do realistic lip-syncing with natural backgrounds.

  • The avatars can communicate in 30 different languages.

  • These personal avatars will be efficient for personalized messages, training videos, and social media content.

The Bytes :

Create a logo using Microsoft designer for free :

  1. Visit Microsoft Designer and log in or create a free account.

  2. After logging in, go to the Brand Kit Creator or click "Create a new design."

  3. Input your brand name and a brief description to help generate design ideas.

  4. Click "Generate" to see logo options, then choose your favourite. Customize the selected logo by adjusting colours, fonts, and layouts.

  5. Download the logo in your preferred format (PNG, JPEG, etc.) for various uses. Checkout the generated logo below.

  6. Save your brand kit to keep all branding elements organized for future use.

Magic Prompt :

A captivating 3D render of a majestic Indian elephant, defying all expectations as it soars through the air in a remarkable long jump.Reminiscent of Olympic champion Mike Powell, the elephant's powerful muscles are on full display as its front legs extend forward with incredible force. The backdrop transforms the wild savanna into an Paris 2024 Olympic stadium written "Paris 2024", complete with audience stands and a long jump pit, creating an eerily surreal and intense atmosphere. This striking image showcases the elephant's raw strength and unexpected athleticism, immortalizing a dramatic, record-breaking moment in time., 3d render, photo, vibrantCredit : ciaffeti

FEEDBACK

Your feedback is precious and helps us tailor future newsletters to better suit your needs!

We welcome your feedback and insights. Please reply to this email with any comments or suggestions you'd like to share.

Appreciate your time.

AI-ways forward,Naveen Krishnan,Founder and Writer,The AI Catalyst Solutions Inc.

Thank you for reading The AI Catalyst. This post is public so feel free to share it.