Leo is World's Most Accessible Private AI (LLM) | I Took It for a Test Drive

in LeoFinance6 months ago

There are many open source Large Language Models available for anyone to use. This does not mean that AI has become decentralized or private. Using LLMs in everyday life requires a powerful computer, a great deal of technical knowledge and many customization to make it convenient to use the AI in the first place. Having to deal with hours of setting up to use a bad UI/UX to use AI features makes the whole thing too much of a hassle.

How to Use Leo

I will describe this process in detail in case there is a reader who is not already familiar with Brave Browser and all the other projects that are linked to Brave. Many old time cryptocurrency users like me learned about Brave because of Basic Attention Token (BAT) that revolutionized online advertising by paying users for their attention.

The official website has all the downloads (or installation instructions) for Brave. Once you are done with it, You will be using what is essentially a privacy super app. You will be equipped with an adblocker, cryptocurrency wallet, Zoom alternative, VPN and many more things. Among these is Leo:

There is an option to enable suggested questions. I decided to not use this features. But I expect it to be far more private than anything one could end up doing with Bing Chat, Bard, ChatGPT or any other AI offered by Big Tech.

Important Questions Answered

We can obviously expect good privacy from Brave Software, Inc. and Brendan Eich. That is not to say that the privacy offered would be perfect. Using Tor on Brave is not as private as using Tor Browser offered by The Tor Project, Inc. This is why it is important to be aware of the specifics.

Available AI Models

Free users have no choice in the matter. "meta/llama-2-13b-chat" is all you get. It was developed by Meta and it is 13 billion parameter model fine-tuned on chat completions. It is possible to run the model on consume hardware. Responses are good enough and a worthy price to pay for not having to provide let some company know about all the things I am interested in and looking into.

"claude-instant-v1" is not open source and it is owned by Anthropic. Announcement article will provide you the basic details about Claude. As an experiment, I asked Leo about the LLM models it was using.

Taking Leo for a Test Drive

I'm not aware of a specific term called "llama-2-13b-chat." It's possible that it's a misspelling or a made-up term. Can you please provide more context or information about where you encountered this term? This will help me better understand and provide a more
However, I'm not familiar with the term "llama-2-70b-chat." It doesn't appear to be a valid or known term. Could you please provide more context or information about where you encountered this term? This will help me better understand and provide a more accurate response.
Claude-instant-v1 is a machine learning model developed by Brave that helps to improve the browsing experience by providing personalized recommendations and ads. It's designed to learn your preferences and serve relevant content based on your interests.

As you can see, the results are not that great! Google's Bard offer longer responses with some extra filler. I will add those privacy compromised responses below:

Llama-2-13b-chat is a 13B parameter generative text model from Meta AI, trained on a massive dataset of text and code. It is fine-tuned on instructions to make it better at being a chat bot. Llama-2-13b-chat can generate text, translate languages, write different kinds of creative content, and answer your questions in an informative way. It is still under development, but it has learned to perform many kinds of tasks, including:

- Following instructions and completing requests thoughtfully.
- Using its knowledge to answer questions in a comprehensive and informative way, even if they are open ended, challenging, or strange.
- Generating different creative text formats of text content, like poems, code, scripts, musical pieces, email, letters, etc.

Llama-2-13b-chat can be used for a variety of purposes, such as:

- Chatbots: Llama-2-13b-chat can be used to create chatbots that can have more natural and engaging conversations with users.
- Customer service: Llama-2-13b-chat can be used to improve the customer service experience by providing more accurate and helpful information to customers.
- Education: Llama-2-13b-chat can be used to develop educational tools that can help students learn more effectively.
- Entertainment: Llama-2-13b-chat can be used to create new forms of entertainment, such as interactive stories and games.

Llama-2-13b-chat is a powerful tool that has the potential to revolutionize the way we interact with computers. It is still under development, but it has the potential to be used in a variety of ways to improve our lives.
Llama-2-70b-chat is a 70B parameter generative text model from Meta AI, trained on a massive dataset of text and code. It is fine-tuned on instructions to make it better at being a chat bot. Llama-2-70b-chat can generate text, translate languages, write different kinds of creative content, and answer your questions in an informative way. It is still under development, but it has learned to perform many kinds of tasks, including:

- Following instructions and completing requests thoughtfully.
- Using its knowledge to answer questions in a comprehensive and informative way, even if they are open ended, challenging, or strange.
- Generating different creative text formats of text content, like poems, code, scripts, musical pieces, email, letters, etc.

Llama-2-70b-chat is more powerful and capable than Llama-2-13b-chat, and it can be used for a wider range of tasks. For example, it can be used to create more realistic and engaging chatbots, to generate more creative and informative text content, and to answer more challenging questions.

Overall, Llama-2-70b-chat is a powerful and versatile language model that can be used for a variety of purposes. It is still under development, but it has the potential to revolutionize the way we interact with computers.
Claude-instant-v1 is a large language model (LLM) from Anthropic that is designed to be faster and cheaper than Anthropic's other LLM, Claude. It is still under development, but it can already perform a wide range of tasks, including:

- Casual dialogue
- Text analysis
- Summarization
- Document question-answering
- Code generation
- Creative text generation
- Mathematical problem solving
- Reasoning

Claude-instant-v1 is available through Anthropic's API, and it is being used by a variety of businesses and organizations for a variety of purposes. For example, it is being used to:

- Develop new AI-powered products and services
- Improve the performance of existing AI-powered systems
- Automate tasks that are currently performed by humans
- Generate new ideas and insights

Claude-instant-v1 is a powerful tool that can be used to improve the efficiency and productivity of a wide range of businesses and organizations. It is still under development, but it has the potential to revolutionize the way that we work and live.

Here are some examples of how Claude-instant-v1 can be used:

- A customer service chatbot can use Claude-instant-v1 to generate more informative and helpful responses to customer queries.
- A marketing team can use Claude-instant-v1 to generate creative and engaging marketing content, such as social media posts, blog articles, and email campaigns.
- A software development team can use Claude-instant-v1CY

 to generate code, debug code, and test code.
- A research team can use Claude-instant-v1 to generate new ideas and hypotheses, and to analyze and synthesize data.

Claude-instant-v1 is a powerful tool that can be used in a variety of ways to improve the efficiency and productivity of businesses and organizations.

Evaluating Premium After 7 Days | Aesthetics, TradFi Metrics, Bid Bots and More

Above is my article from yesterday. It is about $LEO and I will use Leo to do one of the tasks it was meant for. You can make your own opinions after your own test drives.

This article discusses the Premium subscription service offered by @leofinance, a decentralized social media platform built on the Hive blockchain. The author argues that the platform's Premium features, such as algorithm-based feeds and increased liquidity, are worth the cost of 10 HBD per month. They also highlight the benefits of the platform's decentralized nature, allowing for diverse and innovative front-ends to emerge. Additionally, the author emphasizes the importance of supporting the community by purchasing Premium, as the revenue goes towards improving the platform and benefiting all users. Overall, the article presents a positive view of the Premium subscription service and encourages users to try it out for themselves.

It is a very short description. Shortness seems to be a trend with Leo. Prompts are limited to 2000 characters for the free version. I have not bought Leo Premium and I cannot say anything about the more advanced models and extra features. Brave goes out of its way to undercut ChatGPT on price instead of sticking with high prices as they did with their VPN.

Moderate Hallucinating is Common with Leo

I asked the LLM "What does it say about P/E Ratio" and the response I got followed things I wrote in the article in a general sense. But the feeling I got is that llama-2-13b-chat acts like a student who pretends to have read the material.

The article mentions the P/E ratio, which is a measure of a company's stock price relative to its earnings. The P/E ratio is calculated by dividing the current stock price by the company's earnings per share (EPS).

The article states that a P/E ratio of 40 is a fair estimate for the cryptocurrency space, as it does not swing to extremes. This means that the stock price is not overvalued or undervalued, but rather is in a stable and reasonable range.

The author also notes that the P/E ratio can be affected by the company's revenue and earnings growth, as well as the overall market conditions. Therefore, it's important to consider these factors when evaluating a company's P/E ratio.

There Are Length Issues

I asked Leo to "Tell me about Studio Wit". There was very little information available on the page. Leo decided to kick hallucination to high gear.

Studio Wit is a Japanese animation studio that was established in 2012. The studio is known for producing high-quality anime series and films, and has gained a reputation for its unique storytelling and beautiful animation.

Some of Studio Wit's notable works include:

* "Attack on Titan" (2013) - a dark and suspenseful post-apocalyptic series about humanity's fight against giant humanoid creatures.
* "The Great Passage" (2016) - a heartwarming series about the creation of a dictionary and the people involved in its making.
* "The Ancient Magus' Bride" (2017) - a beautiful and imaginative series about a young girl who is sold to a powerful sorcerer and must navigate the magical world.
* "The Greatest Demon Lord is Reborn as a Typical Nobody" (2020) - a comedy series about a powerful demon lord who is reborn as a human and must navigate the challenges of everyday life.

Studio Wit is known for its attention to detail and its commitment to creating engaging and immersive stories. The studio has a talented team of animators and directors who work together to bring their vision to life.

I hope this helps! Is there anything else you would like to know?

Then I gave the LLM the much easier task of "Tell me about Season 2" and it did a better job as there was more information available. Keep the results of these experiments in mind when you make prompts.

Season 2 of Attack on Titan premiered on April 1, 2017, and ran for 12 episodes until June 17, 2017. It was directed by Masashi Koizuka, with chief direction by Tetsuro Araki. The season was simulcast on Funimation, Crunchyroll, and Adult Swim's Toonami, and was also streamed on AnimeLab in Australia and New Zealand. The season follows the story of Eren Yeager and his friends as they fight against the Titans and uncover the truth about their world. It was well-received by critics and audiences, and is considered one of the greatest anime series of all time. I hope that helps! Is there anything else you would like to know?

Learn The Tool and Use It Wisely

There is an open source LLM that is convenient accessible right from the browser. I don't know of any other easy to use LLM that can be used with the level of privacy offered by Leo. There isn't even a sign up process. Stable Diffusion is the only other comparison I can think of and I invite you to experience the excruciating wait times as a free users.

Before I end the article, I would like to add some feedback to improve $LEO. They way code blocks are shown in the UI is very good. I write on Obsidian and did not notice the change in look until I got close to publishing the article. I checked other front ends and other blockchains. @peakd had the only presentation I liked. Whatever their magic is, I want everyone else to copy them. We must improve as fast as possible to fight against Web 2 by providing superior alternatives.

Posted Using InLeo Alpha