Pushmeet Kohli, Google DeepMind’s VP of research has stated that this tool will be helpful for AI developers to identify if text outputs were generated from their LLMs. It helps developers to create a more responsible AI. OpenAI has also spent years researching methods related to watermarking. However, due to certain commercial and technical issues, OpenAI has delayed the release of its watermarking technology. So this means Google is not the only tech company working on this kind of watermarking technology. 

SynthID was announced in 2023 and its sole purpose was to detect AI-generated output. This is done through the addition of a watermark that cannot be seen on audio, videos, images and text. Google has stated that the SynthID text watermark makes the text less probable as it cannot be detected by humans. Instead, it can be detected by software instead. 

SynthID Text Watermark: More on this new open-source tool

This technology has already been integrated into Google’s Gemini chatbot. The tech giant has stated that even with this integration, the quality, creativity, speed and accuracy have not been compromised, an issue prevalent in most watermarking systems. Furthermore, Google states that the SynthID text watermark can apply to texts that are paraphrased, cropped or even if it is only three sentences long. 

However, the technology does not come with no setbacks. It has trouble with shorter texts or translated and rewritten texts. SynthID will not be able to identify texts that have been translated from other languages. For instance, if a person translates text from English to Spanish with AI, then SynthID will be unable to identify the text. 

Suggested:

Google Flights Update Lets Users Get The Cheapest Flights

Smart Replies: Google’s New Feature Now Rolling Out

Image Credits: Screenshot Taken From Google DeepMind

The post SynthID Text Watermark Released by Google  appeared first on Tricky Enough.

Code Llama presents the possibility of a change in coding practices in a number of industries thanks to its specialized versions and promising benchmark performance.

Code Llama: What Is It?

Three different models are provided by Code Llama, which expands upon the well-known foundation of Llama 2.

• The fundamental code structure.
• A version specifically tailored for Python programming is called Codel Llama.
• To handle natural language instructions more effectively, Code Llama created Instruct.

The model is unique in that it is trained to use a variety of well-known programming languages. That includes Python, C++, Java, PHP, and others. With 7B, 13B, and 34B characteristics, the LLM offers three distinct sizes to accommodate various serving and latency needs.

The 7B model is best for single GPU serving, whereas the 34B model provides better results. And is better suited for extended code support.

Responsibility And Risks

Code Llama has displayed outstanding performance in benchmark testing. With scores of 53.7% on HumanEval and 56.2% on Mostly Basic Python Programming (MBPP). It outperformed other open-source code-specific LLMs and was on par with ChatGPT in terms of performance.

Meta emphasized the significance of appropriate AI usage. Even though the technology may be helpful to developers of all levels of expertise and sectors. The business performed in-depth analyses to determine the likelihood that the LLM might produce dangerous code and offered recommendations. These rules follow a paradigm for responsible synthetic media usage. That is endorsed by major players in big tech including Adobe, Google, Meta, Microsoft, OpenAI, and others.

Suggested:

Elon Musk confirms the Deletion of the context for links in X Posts.

Threads by Meta is Not Here to Replace Twitter.

The post Report: Meta AI Launches Code Llama: A Coding LLM appeared first on Tricky Enough.