HN user

flakiness

1,847 karma
Posts8
Comments593
View on HN
Back to Kagi 54 minutes ago

This is such a you-must-be-an-iphone-user comment. Enjoy the good life.

It's a bit shocking to see them showing off blatantly disinformation-al examples. See the Japanese manga one. It says "原作 監修 三浦建太郎" which says its original is written by, and it is supervised the by, the famous author (who died a few years ago so who can he possibly supervise?) and "共同制作 白泉社" saying: collaborated by a (famous Japanese) publisher.

I hope these stakeholders have good enough layers.

Codex Micro 7 days ago

Maybe it's worth getting one just to cherish this crazy moment 10 years later. This does cost money but won't cost much of your drawer space.

How do I actually use flashcards? My software of choice is Anki. I am not completely satisfied with it. The UI looks dated, the WYSIWYG HTML editor is clunky, and the undocumented file format makes potential porting and interoperability tricky.

And we still love it. I'm on the same page. This phenomena feels oddly satisfying.

Mistral OCR 4 29 days ago

On our internal multilingual evaluation, OCR 4 leads across all eight language groups — English, Western Europe, Eastern Europe, Middle Eastern, Chinese, East Asian, Southeast Asian, and specialized languages (Hindi, Japanese, Georgian, Bengali, Armenian, Hebrew, Greek, Gujarati, Tamil, Malayalam, Kannada, Telugu).

The initial version of this page called these "minor languages" (vs specialized language), which is telling. If you're a speaker of one of these: This is why you need a sovereign set of models. (Japanese government: Are you listening?)

I see more of these conventional commit-style comments recently and it feels like coming from Claude Code etc. It's a bit unsettling that not only training data but also random lines in the default system prompt affects this kind of software development norms in subtle and pervasive ways.

https://www.asahi.com/articles/ASV5N32MVV5NUTFK005M.html

The title reads "PM's Office calls Calbee's response a "stunt"; Emphasizes naphtha sufficiency, including intermediate products".

Asahi Shinbun is one of the established newspapers.

Also at the end (translated by google):

"The government interviewed Calbee about the situation on the 12th. According to a government official, they explained to Calbee that there is a sufficient amount available in terms of total volume. Sources close to the Prime Minister expressed concern over the ripple effects, stating, "Calbee's reaction is an overreaction. Their announcement will cause other companies to become anxious as well." However, Calbee maintains its stance, with a public relations representative stating, "This is a measure to ensure the stable supply of our products."

So it's relatively mild "nudge", if you compare it to the current US administration.

Bad timing. Not only MB Neo, but also the memory price hike. Whole selling point is vanishing, plus other makers are getting momentum reacting to Neo, further shadowing the FW12's existence.

I hope they can come back with some update with newer chipset, either from Intel or Qualcomm. They were picking the worst Intel generation and I think it was mostly bad luck.

Prompts are shorter than SWE-Bench Pro's but still longer than how developers actually message agents. Behavioral verification needs some minimum specificity to know what surface to test against, which puts a floor on how terse a prompt can be before the test becomes ambiguous.

It'll be tricky to automate the verification with a vague prompt. In other words, the SWE's job these days is to be a intelligent verifier.

I'm using pi agent and love to try qwen models (hosted). What are the good options? The official provider doesn't include Alibaba. Is OpenRouter etc. fast enough?

(As a reference, DeepSeek v4 is severely throttled on these proxy services.)

Gemini 3.5 Flash 2 months ago

Give me a link. Cannot wait. One PSA is that they have 75% discount right now so it is already cheaper than the full price.

I suspect it's mostly about setting the expectation. They don't want to give up the control, they don't make it "free" (although it virtually is). Both are possible with open source but it would need a lot of explanation. Being closed makes it more natural.

A few words on DS4 2 months ago

I believe the assumption is: The code is cheap. The collaboration (eg. upstreaming) is expensive.

Is it true? We'll see, in a few years.

Mojo 1.0 Beta 3 months ago

https://mojolang.org/docs/roadmap/#contributing-to-mojo

We're committed to open-sourcing all of Mojo, but the language is still very young and we believe a tight-knit group of engineers with a common vision moves faster than a community-driven effort. So we will continue to plan and prioritize the Mojo roadmap within Modular until more of its internal architecture is fleshed out.

I hope they stick to their original promise. And the 1.0 release would be a great time to deliver this.