DEJAN (@dejanseo) 's Twitter Profile
DEJAN

@dejanseo

Australian SEO for medium to large brands and e-commerce websites. I use machine learning to build previously unimaginable SEO workflows.

ID: 16647573

linkhttps://dejanmarketing.com/ calendar_today08-10-2008 11:50:41

7,7K Tweet

16,16K Followers

867 Following

DEJAN (@dejanseo) 's Twitter Profile Photo

You know how some treats are chewy but tasty? Well, my next article is a bit like that. Enjoy!🍬 dejanmarketing.com/query-found-to…

DEJAN (@dejanseo) 's Twitter Profile Photo

Google's experimental Gemini 1.5 model is by far the most intelligent model suitable for very complex tasks and reasoning. Miles ahead of the competition.

Google's experimental Gemini 1.5 model is by far the most intelligent model suitable for very complex tasks and reasoning. Miles ahead of the competition.
DEJAN (@dejanseo) 's Twitter Profile Photo

What did Ilya see? Now a rhetorical question synonymous with the broader alignment problem, comes with a straightforward interpretation. If any one of us would interact with the raw unaligned GPT-4 or Gemini-Level model with no safeguards, it would pass the Turing test. Why?

What did Ilya see?

Now a rhetorical question synonymous with the broader alignment problem, comes with a straightforward interpretation.

If any one of us would interact with the raw unaligned GPT-4 or Gemini-Level model with no safeguards, it would pass the Turing test. 

Why?
DEJAN (@dejanseo) 's Twitter Profile Photo

Look at this blatant and deliberate removal of my article as a canonical source of information about link inversion. Watch till the end to see the clumsy attempt at hiding.

DEJAN (@dejanseo) 's Twitter Profile Photo

This is a very difficult read, goes waaaaay beyond the core message and needs to be partitioned into, like... 20 articles. My friend Andrea Volpini you just set the bar for content too high for the rest of us to follow. Wow! 🙌 searchenginejournal.com/introducing-se…

DEJAN (@dejanseo) 's Twitter Profile Photo

We build on the work by Manaal Faruqui and Dipanjan Das from Google AI Language team to train a search query classifier of well-formed search queries. Our model offers a 10% improvement over Google’s classifier by utilising ALBERT architecture instead of LSTM. With accuracy of

DEJAN (@dejanseo) 's Twitter Profile Photo

Most search queries are ambiguous making it difficult to classify intent and make decisions on how to optimise for them. Query expansion helps, but only only if you know which queries to expand. This is where our model comes in. Take it for a spin here and try proper questions vs

DEJAN (@dejanseo) 's Twitter Profile Photo

To those of you praising Claude / Sonnet 3.5, I just signed up for a paid account and tried it. It's great! But it's not a real product where I can do serious projects. It's a model demo at best. Let me know when it's out of kindergarten. Back to Gemini.

To those of you praising Claude / Sonnet 3.5, I just signed up for a paid account and tried it. It's great! But it's not a real product where I can do serious projects. It's a model demo at best. Let me know when it's out of kindergarten. Back to Gemini.
DEJAN (@dejanseo) 's Twitter Profile Photo

In case this looks familiar, it's a fairly elaborate tweet I decided to turn into a blog post. Putting it out there for indexing and LLMs to feed on. dejan.ai/blog/resource-…

DEJAN (@dejanseo) 's Twitter Profile Photo

Chrome's Built-in Gemini Nano APIs: 1. Summarisation 2. Writing 3. Re-Writing Vide from Chrome Canary. Dev also has the prompt API. #prompt-api-for-gemini-nanoPrompt API for Gemini Nano Enables the exploratory Prompt API, allowing you to send natural language instructions to a