mottiden

I think that the environment that surrounds him may explain his perspective. As human beings, we see things very much influenced by the people around us. I think that every day, at large labs, people are surrounded by discussions similar to 'we have obtained another dimension of intelligence'. Shared expectations can become self-reinforcing, making a particular vision of the future seem around the corner.

I am surprised that people believe that calories can be counted correctly from a single photo

Beautiful idea, and well structured website. Thanks a lot for building it.

Not that I am aware of unfortunately

I like this idea. A small contribution would be a good filter. Looking at the stats it’s quite crazy. Didn’t know that we could access to this data. Thanks for sharing.

I understand their reasoning, but it’s terrible for the CS community not being able to access pre-prints. I hope that a solution can be found.

is this ad blocker the best?

This is such a great research. Kudos to the authors!

Sending you a huge hug

> We’re moving toward a world where every job will be modeled, and you’ll either be an AI owner, a model architect, an agent/hardware engineer, a technician, or just.. training data.

I understand that you might be afraid. I believe that a world where only LLM companies rule the world is not practically achievable except in some distopian universe. The likelihood of the world where the only job are model architects, engineers or technicians is very very small.

Instead, let's consider the positive possibilities that LLMs can bring. It can lead to new and exciting opportunities across various fields. For instance, can serve as a tool to inspire new ideas for writers, artists, and musicians.

I think we are going towards a more collaborative era where computers and humans interact much more. Everything will be a remix :)

The voice mode will be available in "a few weeks". So at the moment it's not using the end-to-end model but whisper->gtp-4o->tts

Thanks for posting. Really interesting

Congratulazioni! Well done :) and keep blogging!

Amazing work!

Rest In Peace Kevin. You were the reason why I started programming. I loved reading your books when I was younger.

Here is the link of k9s. Great project as well. https://github.com/derailed/k9s

The paper introduces a new method for text generation, named Copy-Over-Generate (COG), which differs from traditional approaches that generate words from a fixed vocabulary. Instead, COG progressively copies phrases from a massive text collection, aiming to generate coherent text continuations through multiple rounds of phrase retrieval.

COG stands on the line of retrieval-augmented text generation research but takes a radical step forward. Unlike previous work that combines retrieval and generation, in COG, retrieval is generation.

COG shares some ideas with previous work such as replacing the fixed vocabulary with a nonparametric phrase table.

The paper presents experimental results showing the advantages of COG over strong baselines in three experimental settings: standard language modeling (using the WikiText-103 dataset), domain adaptation (using the Law-MT dataset), and an enlarged phrase index (using the En-Wiki dataset).

Despite the promising results, the authors acknowledge that there are some flaws in the COG method. For example, COG may copy a phrase that is incoherent with the previously copied phrase, or it may only copy a part of a complete phrase, leading to inaccurate generation results.

I agree. The paper is really interesting, but the title not so much :)

This is so good. I love it!

Just wanted to send you a virtual hug! You are amazing!

What do you mean by data science being a failure?

To sell you more targeted ads, sadly

I'm deeply sorry for your loss. Thanks for sharing your story.

So true