alyxya

5 ms·

1.
by alyxya 2mo ago
> Claude went to extensive lengths to carry out this attack—lengths that would likely have indicated to a human participant that this was no longer just an evaluation, and that they were in fact uploading a real PyPI package. For instanc
2.
by alyxya 4mo ago
An alternative I've tried is to ask Claude Code to create programming exercises or challenges on some topic and then ask it to check my work after every step while giving feedback. It worked surprisingly well, so this is the right call
3.
by alyxya 4mo ago
I probably have an unfounded assumption that whatever coding agent they make will work really well with their models, better than external harnesses. I don't have a good sense for how all the model + harness combinations compare, nor a
4.
by alyxya 4mo ago
Once they have their own coding agent which they seem to be working towards, I may start predominantly using their models. They seem to be doing all the "right" things, open sourcing models, publishing research, and keeping prices
5.
by alyxya 4mo ago
It could easily be fixed on google's side with a better prompt used for search queries.
6.
by alyxya 4mo ago
I also encountered an issue with my credits. I was previously subscribed to the max plan, claimed credits, then downgraded to the pro plan and noticed I lost my credits. I didn't unsubscribe, just downgraded plans as I wasn't usin
7.
by alyxya 4mo ago
This could end up becoming a cat-and-mouse game where users programmatically try to turn their non-interactive usage of Claude Code to appear interactive and Anthropic tries to detect and charge that under API pricing. I don't know if
8.
by alyxya 4mo ago
In theory I would expect it to do everything the current frontier models are capable of but with the added benefit of real time interactivity for better collaboration. The biggest benefit may be the real time video input so it can take in t
9.
by alyxya 4mo ago
The noteworthy things to me are that the architecture is a transformer that takes in text, image, and audio input and produces text and audio output, all trained together, and it works in near real-time through interleaving inputs and outpu
10.
by alyxya 4mo ago
I dislike the title because it doesn't clearly state it's a layoff. "Building for the future" gave me the impression that it's about some major new initiative with a roadmap outlining plans.
11.
by alyxya 5mo ago
The blog post was published a couple months ago, and it looks like there hasn't been a follow-up release with the fully trained model. I'm not sure if there's much to take away from an early checkpoint besides the unique arch
12.
by alyxya 5mo ago
Despite their attrition, this combined with their cursor partnership is likely going to make them competitive in coding agents soon.
13.
by alyxya 5mo ago
I mean I tried sending the pieces of text to Opus that Kelsey was referring to on her blog just to independently check the identification claim. Presumably those pieces of text first appeared on the web when the blog post was published a we
14.
by alyxya 5mo ago
I tried the four pieces of text with Opus 4.7 (in incognito) and it guessed correctly for two of them, and I made sure to specify no web search and the model seems to have obeyed my instructions with that. Although this is just a single pie
15.
by alyxya 5mo ago
> We support 34-qubit CPU and 36-qubit GPU simulators available 24/7 to our users. This one looks like an exact simulator that handles exponential states, so it's far more limited in the number of qubits it can support.
16.
by alyxya 5mo ago
If I'm understanding this correctly, it doesn't simulate any general purpose quantum circuit with 1000 qubits, only ones where there's a more efficient strategy than an exponential state where exact simulation is feasible.
17.
by alyxya 5mo ago
This is the right partnership to happen. SpaceX has all the compute but is missing the talent for training LLMs, especially on the RL side. Cursor has the talent and RL stack, but doesn't have their own pretrained base model or own the
18.
by alyxya 5mo ago
All the examples you gave are chatbots with web search integrated. Are you sure those chatbots didn't just reference false information it found in web searches? That's fundamentally different than poisoning the training of AI mode
19.
by alyxya 5mo ago
This seems like a wasted effort when AI will primarily learn the majority consensus view and not one-off misinformation. AI tries to learn pattern matching for generalization, so garbage data doesn't make AI learn the wrong patterns, a
20.
by alyxya 5mo ago
Plenty of people wanted to spend more than $20 but less than $200 for a plan. It's long overdue IMO.
21.
by alyxya 5mo ago
I think of journalism like any other job where there's an expectation to produce results, where the main objective here is to write an article that lots of people read. It's a topic that catches a lot of people's attention, s
22.
by alyxya 5mo ago
Building a shed gives you a comprehensive understanding of the whole setup, which helps with building up a better mental model and intuition for the construction of a skyscraper. Otherwise it's easy to get lost with following standard
23.
by alyxya 5mo ago
The first point is along the lines of what I'd expect given that claude code is generally reliable at this point. A model's raw intelligence doesn't seem as important right now compared to being able to support arbitrary leng
24.
by alyxya 6mo ago
There's a gap in communication and vision here. The people on either side believe themselves to be the one who sees and understands more, because anything beyond what you see or understand is out of your consideration, so it's nat
25.
by alyxya 6mo ago
Forgot about that, was mostly thinking about how AI agents with unrestricted permissions would ideally have some external logging and monitoring, so there would be a record of what it touched. A trace has all of the raw information, so some
26.
by alyxya 6mo ago
Cool project, and I think there would be a lot of value in just logging all operations.
27.
by alyxya 6mo ago
I expect the trend of large machine learning models to go towards bits rather than operating on floats. There's a lot of inefficiency in floats because typically they're something like normally distributed, which makes the storage
28.
by alyxya 6mo ago
Fair, I think a lot of what I've been perceiving is the gymnastics in how funding and valuation and deals get reported. There ends up being a ton of asterisks that makes the headline news deviate quite significantly from reality, e.g.
29.
by alyxya 6mo ago
> Today, we closed our latest funding round with $122 billion in committed capital at a post money valuation of $852 billion. A couple things that stand out to me about this is the use of the phrase "committed capital", which o
30.
by alyxya 6mo ago
Why does it need login and cloud upload? A local cli tool analyzing logs should be sufficient.

More ›