This commit is contained in:
2024-06-01 23:02:02 +02:00
parent 3081fe048e
commit 90609632b4
7 changed files with 175 additions and 45 deletions
+84 -27
View File
@@ -1,34 +1,91 @@
<h2>JADE is the first Multi-Models chatbot.</h2>
<p>It was build with simplicity in mind. The goal being to have a minimalist chatbot for all models.</p>
<p>I don't want fancy stuff like importing files or images. I want the chatbot to be as simple as possible.</p>
<p>Following this philosophy, I was able to focus on other way to improve the current way we use AI chatbots.</p>
<p>My first complain was that there is a lot of models, and they are all good at different things.</p>
<p>But if you want to use all of them, it get impractical and very expensive.</p>
<h2>JADE: The First Multi-Model Chatbot</h2>
<p>JADE was built with simplicity in mind. The goal is to have a minimalist chatbot that supports all models without
unnecessary features like importing files or images. This focus on simplicity allows us to improve how we use AI
chatbots in other ways.</p>
<p>One of my main concerns was the variety of models available, each excelling in different areas. Using all of them can
be impractical and very expensive.</p>
<h2>Multi-Models</h2>
<p>So I decided to focus my chatbot on this and created the first Multi-Models chatbot.</p>
<p>The idea is to use multiple models in the same conversation.</p>
<p>There is 2 things to understand about this:</p>
<p>1. When asking a question, you can ask it to multiple models. You can them compare and pick the best one.</p>
<p>2. The next message will use the text of the selected message for all models. So the response from GPT4 can be then
use by Claude Haiku for example.</p>
<p>To address this, I created the first Multi-Model chatbot. The idea is to use multiple models within the same
conversation. Here are the key points:</p>
<ol>
<li>When asking a question, you can query multiple models and compare their responses to choose the best one.</li>
<li>The selected response can be used as the basis for the next message across all models. For example, a response
from GPT-4 can be used by Claude Haiku in the next interaction.</li>
</ol>
<h2>API Keys</h2>
<p>To use JADE, you need to sign up for an API key to providers. There is 5 providers available:</p>
<ul>
<li>OpenAI</li>
<li>Anthropic</li>
<li>Mistral</li>
<li>Groq</li>
<li>Google</li>
<li>Custom endpoint (premium)</li>
<li>GooseAI (coming soon)</li>
</ul>
<p>You can enter an API key in the settings menu. One enter you get access to all models from this provider. Check the
appendix for more details about available models.</p>
<h2>Bots</h2>
<p>Once an API key enter, you have access to all models from that provider.</p>
<p>If you open the Bots menu (next to send), you will find a list of all your Bots. A bot have a name, a model and some
parameters.</p>
<p>You can create a new bot by clicking on the "+" button. And enter a name and a model. Additionaly, you can add choose
a temperature and a system prompt.</p>
<p>Once created, you can select it by clicking on it and order it by draging it. You can keep SHIFT pressed and click on
2 Bots to select them and all in between.</p>
<p>You can also delete selected Bots by clicking on the trash can.</p>
<p>Once you enter an API key, you gain access to all models from that provider. In the Bots menu (next to the send
button), you'll find a list of all your Bots.</p>
<p>To create a new bot, click the "+" button, enter a name and a model, and optionally set a temperature and a system
prompt. Once created, you can select a bot by clicking on it and reorder it by dragging. Hold SHIFT and click to
select multiple bots. You can delete selected bots by clicking the trash can icon.</p>
<h2>Conversation</h2>
<p>A conversation is a list of messages. You have a Default conversation that is created when you first sign in.</p>
<p>You can create a new conversation by clicking on the "+" button. You can also delete a conversation by clicking on
the trash can.</p>
<p>You can not delete the Default conversation but you can Archive it.</p>
<h2>Conversations</h2>
<p>A conversation is a list of messages. A Default conversation is created when you first sign in. You can create a new
conversation by clicking the "+" button and delete one by clicking the trash can icon. The Default conversation
cannot be deleted but can be renamed.</p>
<h2>Usage</h2>
<p>You can see the usage of your account in the Usage menu. You can see how many messages you have sent and received,
and how much the API costed you.</p>
<p>In the Usage menu, you can see the usage of your account, including the number of messages sent and received and the
associated API costs.</p>
<h2>Appendix</h2>
<p>The following is a list of available models for each provider:</p>
<ul>
<li>OpenAI:
<ul>
<li>gpt-3.5-turbo</li>
<li>gpt-4</li>
<li>gpt-4-turbo</li>
<li>gpt-4o</li>
</ul>
</li>
<li>Anthropic:
<ul>
<li>claude-3-haiku-20240307</li>
<li>claude-3-sonnet-20240229</li>
<li>claude-3-opus-20240229</li>
</ul>
</li>
<li>Mistral:
<ul>
<li>open-mistral-7b</li>
<li>open-mixtral-8x7b</li>
<li>open-mixtral-8x22b</li>
<li>mistral-small-latest</li>
<li>mistral-large-latest</li>
<li>codestral-latest</li>
</ul>
</li>
<li>Groq:
<ul>
<li>llama3-8b-8192</li>
<li>llama3-70b-8192</li>
<li>gemma-7b-it</li>
</ul>
</li>
<li>Google:
<ul>
<li>gemini-1.5-pro</li>
<li>gemini-1.5-flash</li>
<li>gemini-1.0-pro</li>
</ul>
</li>
<li>Inference Endpoints (More custom to come)</li>
</ul>