Mealie isn’t bad, but the UI makes it a huge pain to enter recipes. Luckily, the web import often works well, but some websites don’t support the JSON recipe format it requires. I recently setup llama.cpp on my server for the “AI import” feature that closes the gap well enough.
Yeah, I use 3.8 27B and it’s like Sonnet 5 for a lot of tasks. If they’re keeping up with Sonnet, 5.5 is similar to Fable, and it would be pretty nuts if Qwen 4 27B kept pace.
Yeah, don’t buy an overpriced Raspberry Pi; instead, you should buy a more reasonably priced…oh.
I know a lot of people love RoR, but I don’t. I have a long list of reasons, including it being slow, the syntax is weird, and while it’s possible to do compile-time type checking, it feels clunky and bolted on. RSpec especially annoys me with its use of dynamic language features to “cheat” when writing tests instead of creating code that is loosely coupled and testable. Don’t get me started on ORMs.
At work, I’m forced to work in legacy RoR codebases, and honestly, AI coming along and letting me work at a layer of abstraction above RoR was a welcome change. I used to spend the a whole day getting pissed off whenever I had to work in these RoR codebases, and letting a LLM generate the code lets me disconnect, keep my emotions out of it’d and pump out whatever drivel is needed to earn my paycheck for the day.
Yep, I assume that’s the plan, though HuggingFace isn’t the only game in town. Hugging Bay, ModelScope, and I assume a horde of others will fill the void if HF goes downhill.
I run Qwen 3.8 27B, which seems pretty comparable to Sonnet 5 for coding and other tasks. Qwen 3.6 was in the neighborhood of Sonnet 4.6. Qwen 4 27B is supposed to drop in the next couple months and we’ll see how it compares to Sonnet 5.5, which benchmarks at Fable / Mythos level.
If the trend keeps up, not sure how useful those data centers will be when a gaming PC is a few months behind the frontier.
Anthropic and OpenAI trained their models on pirated data and have no scruples about hoovering up data by any means necessary. There is no privacy when using their models. My employer has an enterprise contract with Anthropic where they are not supposed to be collecting my employer’s IP and training models on their data. Duck.ai supposedly has a similar agreement. Given Anthropic’s history, I would not put it past them to ignore the terms of the contract and do what they want with the data anyway.