HackerTrans
TopNewTrendsCommentsPastAskShowJobs

joanfihu

no profile record

Submissions

The Platform for the Universal Personal Assistant

joanfihu.com
2 points·by joanfihu·ปีที่แล้ว·1 comments

VQA and Image Chunking for MLLMs (GPT-4V and Gemini)

joanfihu.com
1 points·by joanfihu·2 ปีที่แล้ว·1 comments

The Generative Internet

joanfihu.com
1 points·by joanfihu·2 ปีที่แล้ว·1 comments

Document Chunking for RAG (Retrieval Augmented Generation)

joanfihu.wordpress.com
1 points·by joanfihu·3 ปีที่แล้ว·1 comments

Can LLMs find knowledge gaps in the Internet?

joanfihu.wordpress.com
2 points·by joanfihu·3 ปีที่แล้ว·0 comments

comments

joanfihu
·ปีที่แล้ว·discuss
Should it need to become independent, OR is it going to be forced to become independent?

It seems Google won’t be able to pay to be the default SE in any browser: Safari, Mozilla, etc.

https://askpandi.com/pandipedia/judgement-against-google-pro...

So I guess Mozilla is scrambling to find new revenue streams since 88% of their revenue comes from it.

I won’t fund Mozilla because it’s been forced to operate fairly.

They should build a product that makes me want to pay for it.
joanfihu
·ปีที่แล้ว·discuss
You need a product that people want to get traction…

Then you can accelerate.
joanfihu
·ปีที่แล้ว·discuss
Out of curiosity, why would ad-sellers be interested in this?
joanfihu
·ปีที่แล้ว·discuss
There is no way I'll read all that text from the demos...

AskPandi has a similar feature called "Super Search" that essentially checks more sources and self validates it's own answers.

iT's AgEnTic.

The answers are easier to digest, if you search for products, you'll get a list of products with images, prices and retailers.
joanfihu
·ปีที่แล้ว·discuss
The quest to build the universal personal assistant (UPA) has been reignited, with both OpenAI and Google going head to head. But what would the ultimate personal assistant look like? Would it be an app, a browser, an OS, or even a physical robot?
joanfihu
·2 ปีที่แล้ว·discuss
Check out ColPali and ColQwen for a SOTA open source version.
joanfihu
·2 ปีที่แล้ว·discuss
Does it not break YT ToS?
joanfihu
·2 ปีที่แล้ว·discuss
It’s sad that a product that I used as a North Star has lost so much momentum. Search has only had incremental updates but nothing major. ChatGPT has been the biggest development in search…

It’s one of the reasons why I started AskPandi as a side project. It’s an answer engine that gives answers without fluff. No link sifting, no ads, no popups, no cookie consents.

It’s making browsing the web fun again.
joanfihu
·2 ปีที่แล้ว·discuss
Handling large images with resolutions higher than 2048x2048 remains a challenge for MLLMs.

Image chunking splits large images into multiple smaller images. Then the model processes each image and uses information from all of them to answer a question.

I managed to get it to work with a 3840px X 20294px web page screenshot. It also works with documents that have figures (tables, charts, illustrations, etc).
joanfihu
·2 ปีที่แล้ว·discuss
Answer engines, content creation and UI navigation are evolving the Internet into a new mutation.

Search returns answers. Web pages are generated on the fly and the browser browses for us.

In this blog post, I write about how the generative Internet is unfolding.
joanfihu
·2 ปีที่แล้ว·discuss
The main problem researchers found is that the content quality has deteriorated. There is more spam now. However, Google is doing a better job at filtering low quality content than anyone else.

I've been working on a side project to filter spammy content and get straight to the point answers: Https://AskPandi.com
joanfihu
·3 ปีที่แล้ว·discuss
“In court Thursday, Lehman said his best guess is that search engines will shift largely from relying on user data to relying on machine learning.”

This is a good point. Foundational/Frontier models already pack lots of contextual knowledge that makes user behaviour data collection less relevant.

Whilst building AskPandi, I have only found user’s location to be the only major signal that would greatly improve search results and answers.
joanfihu
·3 ปีที่แล้ว·discuss
It's not about Ilya wanting to slow things down.

Ilya is the technical mastermind behind OAI. The technical breakthroughs needed for AGI are not there yet. Ilya, Yann, Demis and many others are aware of it.

An aggressive push for applied research and commercialisation means less resources for technical breakthroughs.

This is a tricky situation.
joanfihu
·3 ปีที่แล้ว·discuss
Is it possible to get video transcriptions from any video now?

Last time I tried, I couldn’t do it.
joanfihu
·3 ปีที่แล้ว·discuss
I’ve done a review for UI navigation

https://joanfihu.wordpress.com/2023/10/19/evaluating-adepts-...