site banner
PaperclipPerfector  12d ago (text post)   1991 thread views

Friday Fun Thread for November 8, 2024

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

1
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

If you use llama.cpp, you can load part of the model into VRAM and evaluate it on the GPU, and do the rest of the evaluation on CPU. (The -ngl [number of layers] parameter determines how many layers it tries to push to the GPU.)

In general, I strongly recommend using this over the "industry standard" Python-based setup, as the overheads of 1GB+ of random dependencies and interpreted language do tend to build up in ways beyond what shows up in benchmarks. (You might not lose so much time per token, but you will use more RAM (easy to measure) and put more strain on assorted caches and buffers (harder to attribute) and have more context switches degrading UI interactivity.)

In general, I strongly recommend using this over the "industry standard" Python-based setup, as the overheads of 1GB+ of random dependencies and interpreted language do tend to build up in ways beyond what shows up in benchmarks

Very true. Playing around with stable diffusion has made the impossible - made my windows seize up over memory, actual app crashes & need to restart the system more than every three weeks when the inevitable update happens.

I had to download rammap from sysinternals because it could delete leaked memory a1111 spilled in gigabytes every time a model was being swapped. Every ~50 mb of pics generated it will need restarting.

Maybe I should go to comfyUI.

Thanks! Do you use it?

Yes, though I haven't paid attention to it in about half a year so I couldn't answer what the capabilities of the best models are nowadays. My general sense was that performance of the "reasonably-sized" models (of the kind that you could run on a standard-architecture laptop, perhaps up to 14B?) has stagnated somewhat, as the big research budgets go into higher-spec models and the local model community has structural issues (inadequate understanding of machine learning, inadequate mental model of LLMs, inadequate benchmarks/targets). That is not to say they aren't useful for certain things; I have encountered 7B models that could compete with Google Translate performance on translating some language pairs and were pretty usable as a "soft wiki" for API documentation and geographic trivia and what-not.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats