DragGAN & LLaMA Models go Wild

DragGAN

A new AI text-to-image model just dropped. And it’s phenomenal.

How so?

Imagine Photoshop, combined with AI, combined with IG filters, combined with…you get the point.

It’s a very clean and sleek tool.

Basically, using the DragGAN model, you can drag any points of an AI generated image to get to specific target points in an intuitive way.

You get full control of an AI image. And really, this is what many AI image tools have been missing- that seamless editing functionality that we all know and love.

But, this can’t be all… you’ve got to be wondering, what else is the Generative AI world cooking up?

Well, ladies and gentlemen, brace yourselves.

We got….

AI sexbots.

Built on the open-source LLaMA model from Meta.

The trajectory of this technology is pretty insane. And what the hell is LLaMA? Sorry, I was getting a little ahead of my skis.

You see, LLaMA is a 65 billion parameter, open-source large language model built by the Facebook team. Specifically, it is meant to help researchers advance AI and NLP systems.

It’s a noble effort, and many in the open-source developer community have been building furiously…

So furiously. With such an imaginative bent. Now, AI sexbots are a thing.

Let’s dive in.

An individual made an AI chatbot with the persona, Allie. This gets wilder and wilders by the day. With open source, developers have the ability to create any intelligent chatbot that one could desire at any moment.

And so, the developer who made this thought- screw customer service chatbots, you know what’d be good? A sex chatbot…pretty wild future on the horizon.

Anyways, moving on.

________________

We’ve got a quick intermission for the nerds here…and we’re going to tell you how to implement a chatbot with LangChain!

1. Open Replit if you don’t have Python installed on your computer, and make a .py file

2. Open the terminal and hit pip install Langchain

3. Have the following import statements:

from langchain.llms import OpenAI

from langchain import LLMChain

from langchain.prompts.prompt import PromptTemplate

from langchain.memory import ConversationBufferMemory

4. Now add in your template:

Example:

template = """

You are a chatbot that is really hardworking

You're goal is to take any customer support questions

Take what the user is saying and make an actionable step they can take

{chat_history}

Human: {human_input}

Chatbot:

"""

5. Now, assign the prompt, memory, and llm_chain variables

prompt = PromptTemplate (

input_variables= ["chat_history", "human_input"],

template= template

)

memory = ConversationBufferMemory (memory_key = "chat_history")

llm_chain = LLMChain (

llm = OpenAI (openai_api_key = openai_api_key),

prompt = prompt,

verbose = True,

memory = memory

)

6. Finally, call the LLM to predict things

___

Anyways, SalesForce has just released its own Large language model called XGen. It’s a pretty badass name if I do say so myself.

It has a similar structure to Facebook’s LLaMA model in terms of context and parameters. But did I mention the cooler name?

Have a good weekend 🙂