danbrooks

That's one heck of a pelican!

Agreed on Opus 5.5 being a great model. It's the first one that I trust for long running (>1 hour) tasks.

This type of project looks extremely useful. There was a lot of buzz around Jev, but having models that run locally and can be fine-tuned is extremely helpful.

Glad my work blocked this. But can we get a NSFW tag? If this is NSFW?

I'm confused.

This seems to imply that they used Claude Opus 5.5 before it was released on September 22nd.

Many people knew this announcement was coming. The betting markets suggested a very high likelihood of Opus dropping today. I was anticipating this quite a bit!

Not sure this is as nefarious as it might look at first glance.

Disney+ supports a variety of content types - some of which naturally have ads.

A big example is the Super Bowl (coming in February), which has always had ads.

Nice! Does anyone know how this compares to the Unsloth quantizations of this model? https://unsloth.ai/docs/models/qwen3.8#run-qwen3.8-guide

An aside - isn't it nice that have been fewer model releases the past week or so? A brief respite!

Just gave it a look. Turns out to be a paid service after a bunch of clicks.

Seems like a well designed policy. Currently for one year and up to the middle school level.

Most LLMs can do a pretty good job of this out of the box.

If you have a coding agent available, ask it to try transcribing a few of the PDFs.

I had the same inspiration!

I spent a days reviewing the lecture notes for CS336: Language Modeling from Scratch - and then trained a nanoGPT-esque model in PyTorch.

I'd recommend trying it for those who are curious. Computational bottlenecks become much more intuitive when you've looked at the overall process.

Typically, companies file these patents to protect themselves against patent trolls.

I’d imagine they could. I’d try Gemini 3.5 flash with high fps.

Super high scores for the community!

I got 83/100 suggesting 60,000.

My SAT reading was 760/800.

Very cool. I’ve been daydreaming about whether LLMs can be used to reason through gaming decisions.

I'd be interested in joining a discord server.

Would be great to have a community to discuss the material - even if folks can't commit to the full course.

Theoretically, the fund is non-partisan. I'd be curious to see what happens if some folks claim to be politically targeted by the current DOJ.

Check which model you're using.

The fast version of composer is the default now (which costs ~x3 as much).

This aligns with my understanding.

Before 2018-2019, the opt-in process for data sharing was hidden on a website somewhere. Around that time, the form became part of the vehicle purchasing process.

The instructions likely mean that the bishop itself is black, not the square that it is on.

This was really fun. I like that holding down space bar gives ~20 clicks per second, which helps speeds things up.

Most hugging face models are implemented in PyTorch, with an architecture specified as a series of layers. This looks like a nice visualization of that.

I tried the VS Code + Copilot sidebar approach a few months ago. It was definitely rough around the edges compared to Cursor/Claude. In our corporate environment, we weren't even able to use frontier models.

Amazing ending. I have been told "no one should be using copilot" - and I agree!

It's not a joke - but it sure feels suspicious :D

This feels like regulatory capture by the tech industry. Academic research needs representation as well.

I work in content/video intelligence. Gemini is great for this type of use case out of the box.

I also prefer the original version - the AI version has a strange vibe.