📰 AI Updates: TikTok Group Chats and LIVE Chatbot Security Webinar

TikTok Introduces Group Chats to Keep Conversations Within the App

TikTok has added a new group chat feature that lets up to 32 people watch and react to videos together.

However, to keep young users safe, kids aged 13-15 can't use this feature, and 16-17-year-olds can only join if they have at least one friend in the chat.

TikTok also added stickers to direct messages, making it more fun to chat.

TII Unveils Falcon Mamba 7B: The First Open-Source State Space Language Model (SSLM)

In benchmark tests, Falcon Mamba 7B outperformed leading transformer-based models, including Meta's Llama 3.1 8B and Mistral's 7B.

It also surpasses all other open-source SSLMs in previous benchmarks and is expected to lead Hugging Face’s new leaderboard.

Falcon Mamba 7B overcomes the sequence scaling limitations of traditional transformer models while maintaining high performance.

SSLMs like Falcon Mamba 7B represent a newer approach to natural language processing.

Unlike transformer models, they use control theory techniques to handle sequential data, allowing them to efficiently process longer sequences with lower memory requirements.

Alibaba's Qwen2-Math Outperforms GPT-4o, Claude 3.5, and Google's Math-Gemini in Math Problem Solving

Alibaba has released the Qwen2-Math series of Qwen, with its flagship model outperforming all others globally — including OpenAI's GPT-4o, Anthropic's Claude 3.5 Sonnet, and even Google’s Math-Gemini Specialized 1.5 Pro.

Qwen2-Math is a language model built on the Qwen2 LLM and is specialized in solving mathematical problems.

While the model currently primarily supports English, the Qwen team plans to expand Qwen2-Math’s capabilities to include bilingual and multilingual models in the future.

A Powerful Open-Source Tool for Realistic Voice Customization

The new open-source Parler-TTS library is making a significant impact in the text-to-speech (TTS) world.

It features two powerful models—Parler-TTS Large v1 and Mini v1—that can generate natural-sounding speech from text.

These models have been trained on 45,000 hours of audio, resulting in speech that sounds remarkably realistic.

One of the best features of Parler-TTS is the level of control it offers. You can adjust the voice's gender, speed, pitch, and even add background noise—all with a simple text command. Whether you want a slow, deep voice or a fast, clear one, you can easily customize it.

What's even better is that Parler-TTS is open-source, meaning all the tools, data, and code are available for everyone to use and improve.

LIVE Webinar Alert: Gen AI Chatbot Security

Discover how to secure your enterprise’s generative AI chatbots against emerging threats. Join industry experts, including our founder Steve Nouri and Rohit Valia, CEO of Tumeryk.com, as they explore cutting-edge AI security solutions.

Key Highlights:

• Understanding the evolving threat landscape

• Best practices for AI security

• Insights into Tumeryk's AI Guard tools

🗓️ Date: August 14th, 10 AM EDT

Space is limited!

New Merch Drop: Beyond The Stars (#2 in trilogy of Earth To Bella)

Show your support for the AI movement with our GenAI merch.

Get your Tee, post a Photo, Tag us…

We’ll feature you on our social channels.

Check out our GenAI merch!

🤖5 AI Landing Pages Builders

​🏆 Leadpages

📱 Phonesites

🌐 Mixo

🚀 Landingi

🎯 Unbounce