bionhoward

Qwen is pretty good and in a variety of sizes. I’d suggest this one Qwen/Qwen3-14B-GGUF Q4_K_M for you given your vram and to run it using llama-server or lm studio (might be alternatives to lm studio but generally these are nice uis for llama server), it’ll use around 7-8GB for weights, leaving room for incidentals

Llama 3.3 could work for you

Devstral is too big but could run a quantized model

Gemma is good, tends to refuse a lot. Medgemma is a nice thing to have in case

“Uncensored” Dolphin models from Eric Hartford and “abliterated” models are what you want if you don’t want them refusing requests, it’s mostly not necessary for routine use, but sometimes you ask em to write a joke and they won’t do it, or if you wanted to do some business which involves defense contracting or security research, that kind of thing, could be handy.

Generally it’s bf16 dtype so you multiply the number of billions of parameters by two to get the model size unquantized

Then to get a model that fits on your rig, generally you want a quantized model, typically I go for “Q4_K_M” which means 4bits per param, so you divide the number of billions of params by two to calculate the vram for the weights.

Not sure the overhead for activations but might be a good idea to leave wiggle room and experiment with sizes well below 16GB

Llama server is a good way to run AI and has a gui on the index route and -hf to download models

LM Studio is a good gui and installs llama server for you and can help with managing models

Make sure you run some server that loads the model once. You definitely don’t want to load many gigabytes of weights into vram every question if you want fast realtime answers

- load mmCIF / PDB - zoom camera out to show everything - show spheres - remove solvents - color by chain or element - take screenshots - rotate molecules - translate atoms

Rust would be wayyy better than pymol, I’m trying to minimize my Python

What kind of smooth brain hears, “they train AI on your ideas and code, humans read your ideas and code, and you agree not to compete back against this thing a multi trillion dollar company just said can do everything, which competes with you,” and says yes? Oh, the smooth brain who doesn’t even realize, because it’s all buried in “additional” legal CYA documents.

ChatGPT still dominates the reach test since I can at least opt out of model training without losing chat logs, even though I have to agree not to compete with the thing that competes with me. Google is like a corporate version of a gross nerd you tolerate because they’re smart, even though they stalk you weirdly.

What a disgrace, we all ought to be sick about creating a legalese-infused black mirror dystopia, racing to replace our own minds with the latest slop-o-matic 9000 and benefit the overlords for the brief instant they count their money while the whole ecosphere is replaced by data centers

It’s like somehow the most magical tech in history (LLMs) comes along and gets completely shafted by elite grifter-tier scumbaggery. Chat bot speech is more protected than human speech these days, Google doesn’t give a crap about privacy, it’s all about the slimy hidden-opt-out of getting everything fed into the slop machine, and break everything if you do.

“Gee, should the app break if the user doesn’t want us to read their shit?” “Yeah, that sounds good, ain’t nobody got time to categorize data into two whole buckets, ship it!”

“How about we make a free better app that doesn’t even have the option of us not reading their shit?” “Oh, yeah, that’ll really help the KPIs!”

we could have our chat histories with Gemini apps activity turned off (decouple the training from the storage) — including no googler eyeballs on our work chats (call this “confidential mode”)

Jules had a confidential mode for these folks (no googler eyeballs on our work code tasks)

Deep research connected to GitHub and was also confidential (don’t train on the codebase, don’t look at the codebase)

The other stuff (videos etc) are obviously valuable but not a big draw for me personally right now…

The biggest draw for me is trusting I can work on private code projects without compromise of security

You can tell because they advertise “Pro” and “Pro+” but then the FAQ reads,

> Does GitHub use Copilot Business or Enterprise data to train GitHub’s model? > No. GitHub does not use either Copilot Business or Enterprise data to train its models.

Aka, even paid individuals plans are getting brain raped

Can somebody please tell me what software product or service doesn’t compete with general intelligence?

Imagine selling intelligence with a legal term that, under strict interpretation, says you’re not allowed to use it for anything.

Is it so vague it’s unenforceable?

How do we own the output if we can’t use it to compete with a general intelligence?

Is it just a “lol nerd no one cares about the legal terms” thing? If no one cares then why would they have a blanket prohibition on using the service ?

We’re supposed to accept liability to lose a lawsuit just to accept their slop? So many questions

In Gemini you can turn off Gemini Apps Activity (warning: deletes your chat log, you need to copy paste everything into notes)

Highly recommended.

What about using it for AI / developing models that compete with our new overlords?

Seems like using this is just asking to get rug pulled for competing with em when they release something that competes with your thing. Am I just an old who’s crowing about nothing? It’s ok for them to tell us we own outputs we can’t use to compete with em?

Just because OpenAI isn’t in “extraction mode” yet doesn’t mean it’s not a scary monopoly.

Source, figure 2 in: [1] https://iot-analytics.com/leading-generative-ai-companies/

Gemini (the app, not the API or AI studio) is one of the few places where we can use frontier generative AI without a “customer noncompete” (you know, the one where they compete with us and then say we’re not allowed to compete back) … if you use Claude or OpenAI or Grok, you’re prohibited from training on your chat logs, or even using the thing to develop AI. Not so with Gemini app.

Too bad you have to lose your chat history just to deactivate model training (“Gemini apps activity” conflates opt-out of training with opt-out of storing chat history)

I don’t know much about the ads space but I just hope going after Google doesn’t create a vacuum that gets filled by an even worse monopoly (OpenAI)

  Thank
  you

https://rust-lang.github.io/rust-clippy/master/#indexing_sli...

You can configure your lints in your workspace-level Cargo.toml (the folder of crates)

“””

[workspace.lints.clippy]

pedantic = { level = "warn", priority = -1 }

# arithmetic_side_effects = "deny"

unwrap_used = "deny"

expect_used = "deny"

panic = "deny"

“””

then in your crate Cargo.toml “””

[lints]

workspace = true

“””

Then you can’t even compile the code without proper error handling. Combine that with thiserror or anyhow with the backtrace feature and you can yeet errors with “?” operators or match on em, map_err, map_or_else, ignore them, etc

[1] https://rust-lang.github.io/rust-clippy/master/index.html#un...

ChatGPT is a great breakthrough but it’s wasted if everyone has to worry about a noncompete with it. Seriously, how is it not insane to think we should outsource our thoughts and agree never to use the thoughts to compete with the thinker? Who wants to live in a world where nobody thinks and nobody can make anything competitive with their “Saviour Machine?”

Anybody who would join an org like that for a few billion dollars is a sell out. It’s an AI safety nightmare, too. I’m just flabbergasted millions of noobs accept not to compete with intelligence, wtf is this world, if you can’t use your thoughts to compete with your thinker, what is left for you? lol this is worse than black mirror

Google is alright, but they have similar stupid noncompete vendor lock in rule, and no way to opt out of training, so there’s no real reason to trust Google. Yeah they could ship tool use in reasoning to catch up to o3, but it’ll just be catching up and not passing unless they fix the stupid legal terms.

Claude IDK how to trust, they train on feedback and everything is feedback, and they have the noncompete rule written even more broadly, dumb to use that.

Grok has a noncompete rule but also has a way to opt out of training, so it’s on the same tier of ClosedAI. I use it sometimes for jokey toy image generation crap but there’s no way to use it for anything serious since it has a copypasted closed ai prohibition

Mistral needs better models and simpler legalese, it’s so complicated and impossible to know which of the million legal contracts applies

IMHO meta is the only player, but they shot themselves in the foot by making Llama 4 too big for the local llama community to even use, super dumb, killed their most valuable thing which was the community.

That means the best models we can use for work without needing to worry about a lawsuit, are Qwen, and DeepSeek distills, no American AI is even in the same ballpark, Gemma 3 is refusal king if you even hint at something controversial. basically, America is getting actively stomped by China in AI right now, because their stuff is open and interoperable, and ours is closed and has legal noncompete bullshit, what can we actually build that doesn’t compete with these companies? Nothing

https://www.quantamagazine.org/first-support-for-a-physics-t...

Does that imply using OpenAI/Anthropic/Gemini API/Grok/Copilot for anything but the most trivial jokish personal uses ought to be considered a serious breach of fiduciary duty because it exposes your business to potentially massive legal costs?

THEN they act like user data isn’t used for training by default, but if you look at the training part of their terms, it says they do train on “feedback” and then the decompiled version of Claude code reads “ // By using Claude Code, you agree that all code acceptance or rejection decisions you make, // and the associated conversations in context, constitute Feedback under Anthropic's Commercial Terms, // and may be used to improve Anthropic's products, including training models. //- You are responsible for reviewing any code suggestions before use.“

This just seems like cover your ass bullshit if every single use of Claude code counts as feedback they train on

Terrible for wagies, great for capitalist equity lords

That’s real bad since they also write “ Memex may generate aggregate, deidentified data from your use of the Services and Subscriber Data ("Usage Data") and use it to operate, improve and support the Services”

AKA “we can learn from your codebase and you aren’t allowed to compete with us”

Basically, it’s a brain-rape machine for idiots who don’t read the fine print. Sad