seemaze

err.. seems like many of the UI elements taken straight from OnShape?

https://www.onshape.com

I was under the impression that the DoD was not a big fan of Claude.. Codex perhaps?

On mac, I just do a quick screenshot and use the builtin OCR in Preview to select and copy text all the time.

I've been using Raspberry Pi Zeros for cheap little linux appliances since they were released. Boot them up with the latest Alpine Linux and run the whole thing from ram. You can also remove a card from a running machine with no ill affect, and they easily survive power cuts. I've never had a card fail.

This is my feeling 100%. If I'm on the phone, it's as a last resort because all the other prescribed pathways have failed.

That sounds like buying a business that owns real estate as an asset.

I stumbled on the most hilarious cross-walk encounter between one of the these delivery bots and a Waymo in downtown phoenix.. it seems that neither was programmed (probably rightfully so) to take the initiative in the situation, so what ensued was a painfully drawn out exchange of agentic deference.

That's interesting. I've been using Qwen3.5-35B for (poorly) structured table extraction based largely on the reports that Qwen had a much better vision implementation.

I have not benchmarked Qwen3.5 vs. Qwen3.6 for the same task, nor trialed Gemma4-26B. Guess it's time for some testing!

I appreciate the objective anecdote, but obviously Coke™.

I recently replied to a thread about La Marzocco espresso machines[0] regarding the value proposition of expensive coffee gear. I think the mistake in magical thinking is the attempt to rationalize the high cost through a comparative cost benefit analysis with retail (non-commercial) competition.

"I just overhauled a la marzocco sitting in my kitchen. People often inquire about whether it's worth buying an espresso machine for the home, or if it's a good investment as a coffee connoisseur.

My reply is always that it was the best money I've ever spent and the worst investment I've ever made. It's a lifestyle choice, and a questionable one at that. But one I'd make again every time given the opportunity."

[0]https://news.ycombinator.com/item?id=47844085#47885805

This makes no sense. It's like asking:

    "Alice is in Denver. Is Alice in (a) Canada or (b) Mexico?"

    - Your boundary between Canada and Mexico is at 40° latitude, more southern than 53% of the population.

I just overhauled a la marzocco sitting in my kitchen. People often inquire about whether it's worth buying an espresso machine for the home, or if it's a good investment as a coffee connoisseur.

My reply is always that it was the best money I've ever spent and the worst investment I've ever made. It's a lifestyle choice, and a questionable one at that. But one I'd make again every time given the opportunity.

Like the idea, but if the web UI stutters anywhere near as much as the landing page, no thanks.

I grew up with the understanding that acetaminophen was the safe choice for fever or aches, and ibuprofen what the more potent compound for inflammation and severe pain. I recall casual anecdotes that "my doctor said 1.5x or 2x ibuprofen dose is ok when warranted" to address major incursions.

I've never once thought about taking more than the recommended dosage of acetaminophen, largely because I had no expectation that it would provide additional benefit..

In reality, I try to consume 1/2 doses of anything or nothing at all, unless it's a serious medical treatment being administered by a professional.

I think you are correct. I’ve mostly been working with plain llama.cpp, but recently started looking into lemonade for the baked-in NPU support.

Strix Halo, I'm seeing performance inline with these results[0].

I'm interested to investigate the claimed gains from the lemonade-sdk port of Apple MLX inference[1].

[0]https://kyuz0.github.io/amd-strix-halo-toolboxes/

[1]https://github.com/lemonade-sdk/lemonade/issues/1642

Check out the officially supported project Lemonade[0] by AMD. It has gfx1151 specific builds of vLLM, llama.cpp, comfy-ui, and even a PR to merge a Strix Halo port of Apple’s MLX[1] with a quick and easy install.

[0] https://www.amd.com/en/developer/resources/technical-article...

[1] https://github.com/lemonade-sdk/lemonade/issues/1642

I squeeze Qwen3.5-122B-A10B at Q6 into 128GB. It's a great model.

Qwen3.5-9B has been extremely useful for local fuzzy table extraction OCR for data that cannot be sent to the cloud.

The documents have subtly different formatting and layout due to source variance. Previously we used a large set of hierarchical heuristics to catch as many edge cases as we could anticipate.

Now with the multi-modal capabilities of these models we can leverage the language capabilities along side vision to extract structured data from a table that has 'roughly this shape' and 'this location'.

Fingers crossed for mid and larger models as well. I'd personally love to see Qwen3.6-122B-A10B.

nice! will check it out

edit: and thanks for the packaging work!

>Wouldn't change a thing..

That's exactly what certification or licensure does; it imposes financial, civil, and criminal penalties for malpractice.

The liability of incurring penalties quickly outweigh the benefit of arbitraging costs with an unqualified practitioner.

Thats funny, it failed my usual ‘hello world’ benchmark for LLM’s:

“Write a single file web page that implements a 1 dimensional bin fitting calculator using the best fit decreasing algorithm. Allow the user to input bin size, item size, and item quantity.”

Qwen3.5, Nematron, Step 3.5, gpt-oss all passed first go..

I realize it does not address the OP security concerns, but I'm having success running rocm containers[0] on alpine linux specifically for llama.cpp. I also got vLLM to run in a rocm container, but I didn't have time to to diagnose perf problems, and llama.cpp is working well for my needs.

[0] https://github.com/kyuz0/amd-strix-halo-toolboxes

I've been thoroughly impressed with tmux control mode[0] in iTerm2. This lets you manage remote terminal windows with your local window management provider. It is currently in the process of being implemented in ghostty[1] as well, can't wait!

[0] https://github.com/tmux/tmux/wiki/Control-Mode

[1] https://github.com/ghostty-org/ghostty/issues/1935#issuecomm...

Exactly this. When I was young I was told to practice the piano with golf balls under my palms to maintain proper wrist position.

There are a handful of modern keysets available with APL legends as well

https://novelkeys.com/products/cherry-olivia?variant=4300136...

https://21kb.com/products/21kb-apl-classic-retro-beige-keyca...

https://omnitype.com/products/gmk-dimensional-cyl?srsltid=Af...

I've always been curious about APL. My grandfather partnered with an APL programmer in the early 70's to develop shallow geodesic dome geometries suitable for home construction and optimized for least number of distinct strut lengths. The one he lived in is still standing[0]

[0] https://maps.app.goo.gl/YgExrSKh3UZu4XKw9

and yet.. https://news.ycombinator.com/item?id=47615490

Ooph, what midlevel SWE at MS did he rub the wrong way..?