zeknife

7 ms·

31.
by zeknife 3y ago
Whatever it is, it runs at like 5fps in firefox on my 4k monitor
32.
by zeknife 3y ago
I seriously doubt that adding reddit posts is going to increase the output quality of any LLM
33.
by zeknife 3y ago
You could do certainly do it with much less, it would just take proportionally more time
34.
by zeknife 3y ago
Having used VR, I actually think I would rather move around for 3 hours than stand still for 3 hours
35.
by zeknife 3y ago
Probably discord
36.
by zeknife 3y ago
It's not like it prevents you from blinking
37.
by zeknife 3y ago
If I had a fully functional self-driving car, I wouldn't be lamenting that it wasn't creative enough
38.
by zeknife 3y ago
Wouldn't you just be facing backwards?
39.
by zeknife 3y ago
This creeps me out. Hard to say what the actual play is here but transparently saying you want to learn about the user, collect their phone number, and send them unprompted messages is not a great look.
40.
by zeknife 4y ago
Given my experience with debloating w10 you're probably setting yourself up for misery when a mandatory update inevitably conflicts with whatever workarounds you've been using
41.
by zeknife 4y ago
What were these ideas?
42.
by zeknife 4y ago
Thinking an LLM is sentient because it can depict emotions is like thinking stable diffusion is sentient because it can produce an image of a sad face. Lemoine seems like a very superstitious person.
43.
by zeknife 4y ago
Rather, it is a concentration practice, and concentration facilitates mindfulness
44.
by zeknife 4y ago
Look up knowledge distillation
45.
by zeknife 4y ago
There are GPT-2 checkpoints small enough to run on basically any modern computer
46.
by zeknife 4y ago
Why would you trust chatgpt to truthfully explain how chatgpt works?
47.
by zeknife 4y ago
You can avoid post-hoc reasoning by making sure chatgpt works step by step to arrive at the answer instead of first presenting the answer and then the steps. This also makes correct answers more likely, especially if it involves math or som
48.
by zeknife 4y ago
It's not "emulating" anything, there is not an increase in complexity or a change in the type of computation going on. It's just approximating the distribution of naturally occurring text, as always.
49.
by zeknife 4y ago
Looks like this was published shortly after GGPO was made open source in October 2019, along with a big wave of similar rollback hype. It has indeed become more commonplace in the last 3 years.
50.
by zeknife 4y ago
It is not available to run locally, but like the other comments said, an equivalent model would require an inordinate amount of VRAM.
51.
by zeknife 4y ago
It's credited to Jason Lyon
52.
by zeknife 4y ago
If you're DIYing an ergonomic keyboard I don't see why you'd go for a trackpoint over a trackpad or trackball. Personally it's one of the least comfortable and precise input devices I've ever used
53.
by zeknife 6y ago
I was linked here by a friend. There's a lot of thoughts in the discussion about how and if input prediction could be made more sophisticated, maybe using machine learning. This is actually the topic of a degree project I've been