5️⃣ Run Docker image with the following command: docker run -p 8501:8501 full-page
6️⃣ Open this URL in your browser: http://localhost:850
7️⃣ Use the App to take full page Screenshot
🧵 3/3
1️⃣ Download the .tar file from buff.ly/48xWPY4
2️⃣ Install Docker (varies by OS) Run Docker. Ensure it is working
3️⃣ Open Terminal and navigate to where you saved it.
4️⃣ Load Docker image using the following command: docker load -i full-page.tar
🧵 2/3
If you want to see how well your page loads without scrolling (essential for search engines), you can use this little full-page Python app I built (with the aid of GPT-4o).
I have rolled it into a Docker image - so you can easily install and work with it. Follow this:
🧵 1/3
Holy wow 🤯
Watching @thetafferboy reveal SEO findings from his Google exploit = one of the best talks this year
Summarizing can't do it justice (site_quality!) but if you don't already, you'd be wise to follow this smart gentleman
but if you want to run your own model or change the system prompts, you can use them productively. Imagine providing brand TOV information or a list of do/don't say items when generating the output. It can make you more effective when producing content—take it for a spin!
🧵 4/4
The system prompt is VERY important to most chatbots, and you can imagine what you can do with this - for good or ill - to change the outputs from an LLM.
Most commercial LLMs have tried to get ahead of people making changes like this to stop exploitation...
🧵3/4
is software that lets you download LLMs from Hugging Face and run them locally.
In this example, I have Meta's Llama 2.0 Chat loaded, and I show how you can change the system/pre-prompt, which essentially provides the rules for a chatbot and how it behaves.
🧵 2/4
One of the best ways to start getting to grips with Large Language Models and how they can be changed/utilised/manipulated is to work with them yourself.
The barrier to entry is high, but there are some ways around that. LM Studio, for example...
🧵 1/4
Canonical tags - some of the most simple markup, but assuming that all Content Management Systems can tell which URL is “canonical” is, well… naive - but I wish it was that easy.
🧵 4/5
Breadcrumbs - Slightly more subjective, but if I can predict a battleground for any website it is likely to be how the breadcrumbs are displayed, how it works or whether one is present in the first place!
🧵 3/5
Spider traps, rendering issues, canonicalization, hreflang, etc.
Areas I’d be happy never to talk about again:
Semantic headings - h1,h2,h3 etc, such a simple, fundamental concept that's frequently wrong.
🧵 2/5
What SEO problems are you sick of solving?
There's often the thought that much of SEO is constantly in a whirlwind of change - which, in some respects is true - but so much of it frankly isn’t! Technical SEO sees us confronting similar crawl issues with similar outcomes.
🧵 1/5
Head over to buff.ly/4fjnZ7d
Create a copy and save it. Then look for the docID variable on line 3 and replace it with document ID from your Google Doc - it's between /d/ and /edit in the url.
🧵 3/4
Converting this into slide format later is usually a massive pain —or it just feels like it should be easier!
With the assistance of GPT-4o, I now have a Google App Script that does the heavy lifting for me! The process at work is shown in the video; it's very quick.
🧵 2/4
If you're anything like me, you rough out any talks or presentations in list form first. Inspiration usually strikes when I'm on the train or sitting with my phone, and I like to jot everything down quickly while it is fresh 📲
🧵 1/4
If I gave you one of these last night, you then entered a legally-binding* commitment to send me an SEO question that could be answered “it depends…”
tiktok.com/@chrisgreenseo
* not really!