General Guide

General Guide of AI Use Cases

Categories to focus

Text 2 Text (Txt2Txt)
1. Summarization (Sum) for LD;DR
  1. summarize paper
  2. summarize reports
  3. summarize videos (plugins help)
2. Text Generation and Continuation
3. Text Classification
Text 2 Image (Txt2Img)
1. Image 2 Image (Img2Img)
2. Image 2 Text (Img2Txt)
Text 2 Voice (Txt2Voice)
Voice 2 Voice (Voice2Voice)
Text 2 Video (Txt2Video)
Text 2 Code (see Programming Code)

Breakdown

`Txt2Txt`

Use text prompt to generate text

core usage includes:
1. Summarization
2. Text Generation and creative continuation, such as
  1. email
  2. story
  3. manual report
  4. product description (including yourself :upside_down_face:)
  5. song lyrics
  6. Poetry
  7. cosplay or role play
  8. generate prompt for MidJourney/StableDiffusion
3. Read and Write programming code (see Programming Code)
4. Basic Data Analysis (csv files preferred)
the best products rawly based on their models:
1. OpenAI ChatGPT

Google Gemini Advance
Claude
Open Source Meta Llama 2 | Married with MSFT

portal websites to combined models:
1. Poe

ChatGPT

You will love the ChatGPT Plus plan

It defaults to GPT-4 if you are a plus member (USD$20/month)
The Code Interpreter can help you read files, such as json, Java, Perl, csv (for Excel to open), PDF files, etc. Try to ask it for anythiing on top of your mind :smile:, e.g., 'what is the problem with my Perl code?', 'where is the bug in my Java file?' etc.
It's a bunch of fun to play the plugins, for examples:
- WebPilot
- Zapier
- Expedia, Kayak,
- AskYourPDF
- Youtube Summary, VoxScript, Video Insights (or search with keyword Youtube)
Don't forget to check the bottom line, the "ChatGPT xxxxxx version" link, you may find more tricks in the change log to play with.
you can install your own plugins as well.

AI Agents

AutoGPT
BabyAGI
AgentGPT
GPT Engineer | YouTube
MetaGPT | YouTube
Camel
MSFT Jarvis / HuggingGPT | Paper | HuggingFace Playground
Fun game: Westworld

`Txt2Img`

Use text to generate images

the best products - only 2:
- MidJourney - paid plan

open source model Stable Diffusion

prompt and checkpoint sites:
- PromptHero
- famous CiviTai

civitai runmodel

civitai runmodel selection

more online playing websites will be discovered :heart:

a few playing projects or sites:
- SD WebUI open sourced
- DIY and play your own hosting server: Run Diffusion

Adobe Firefly, btw, great transformation, Adobe!!!
OpenAI and Bing(Download Microsoft Edge browser) also provide Dall-E 2 model :thumbsdown: but we are waiting for Dall-E 3

`Txt2Voice`

use text to generate voice, and visa versa

popular open source projects:
- Bark YouTube
Chinese txt2voice services:
- Baidu
- XunFei

`Voice2Voice`

use one voice to generate another voice

popular open source projects:
- so-vits-svc

`Txt2Video`

use text to generate video

Currently Runway Gen2 is providing the best services (few min free)

`Digital Character`

this technology is not mature yet, but it's still fun to play some famous products:

d-id

d-id

HeyGen

heygen

Programming Code

plugins integrated into modern IDEs to help you write code (VsCode, IntelliJ, etc.)

Github Copilot

Open Source Cursor helps you write code in browser

Learning Resources

Popular learning resources to suggested:

Our Prompt Guides. A few hightlights are:
Andrew Ng AI courses

Popular Playgrounds