redman25

6 ms·

61.
by redman25 1y ago
I’m American. Just giving some background to the feeling. There’s some discontent with some western communities (localllama) that Chinese developers have been open weighting all of their models while most western models have been closed wei
62.
by redman25 1y ago
Arroyo uses it for their streaming database. Think streaming stateful aggregations. It’s pretty cool tech. Looks like they were recently acquired by cloudflare.
63.
by redman25 1y ago
Solutions like this can help with complex transformations that rely on intermediate state like streaming aggregations. They often have watermarking, windowing, and all that good stuff built in whereas with redis you would have to build all
64.
by redman25 1y ago
It can be good to know that SSDs are fast until you exhaust the cache by copying gigs of files around. It doesn’t hurt to be aware of the limitations even if for the common case things are much better.
65.
by redman25 1y ago
Sad to say but it would have put US companies at a major disadvantage if they were not allowed to.
66.
Apple open weights code diffusion model(huggingface.co)
1 points by redman25 1y ago|0 comments
67.
by redman25 1y ago
Qwen is great but for creative writing I think Gemma is a good choice. It has better EQ than Qwen IMO.
68.
by redman25 1y ago
The point is the language forces you to explicitly handle the null. You have to make the choice. Using `expect`, the programmer would understand that that is equivalent to a panic. Why would they intentionally panic?
69.
by redman25 1y ago
Crazy, they could call it the "internet" or something like that... kind of rolls off the tongue.
70.
by redman25 1y ago
That's true typed languages seem to handle the slop better. One thing I've noticed specifically with rust is that agents tend to overcomplicate things though. They tend to start digging into the gnarlier bits of the language much
71.
by redman25 1y ago
Not the parent but I would say bad defaults or naming. There are countless posts from newbies wondering why a model doesn’t work as well as it should. It’s usually either because the context size is set very low by default or they didn’t re
72.
by redman25 1y ago
It may not have been intentionally misleading. Some benchmarks can take a lot of horsepower and time to run. Their preparation for release likely was done well in advance of the model release before the new deepseek r1 model had even been a
73.
by redman25 1y ago
The majority of successful attacks are business email compromise. In that context take9 makes sense in my mind because there _is_ responsibility for users to be mindful of clicks. Not every phishing email can be caught before it reaches use
74.
by redman25 1y ago
Gemma-3-12b-qat https://huggingface.co/google/gemma-3-12b-it-qat-q4_0-gguf Qwen_Qwen3-14B-IQ4_XS.gguf https://huggingface.co/bartowski/Qwen_Qwen3-14B-GGUF Gemma3 is a good conversationalist but te
75.
by redman25 1y ago
They can if they've been post trained on what they know and don't know. The LLM can first been given questions to test its knowledge and if the model returns a wrong answer, it can be given a new training example with an "I d
76.
Vision Now Available in Llama.cpp(github.com)
550 points by redman25 1y ago|104 comments
77.
by redman25 1y ago
It's not that simple how would the model know when it knows? Removing hallucination has to be a post-training thing because you need to test the model against what it actually knows first in order to provide training examples of what i
78.
by redman25 1y ago
I think only 2/3 of ram is allocated to be available to the gpu, so like 14gb which is probably not enough to run even Q4 quant.
79.
by redman25 1y ago
You could always run your own server locally if you have a decent gpu. Some of the smaller LLMs are getting pretty good.
80.
LocalScore – LLM benchmark from the Mozilla Builders project(localscore.ai)
1 points by redman25 2y ago|1 comments
81.
by redman25 2y ago
I spent a couple weekends trying to reimplement microsoft's inferencing for phi4 multimodal in rust. I had zero experience messing with ONNX before. Claude produced a believably good first pass but it ended up being too much work in th
82.
by redman25 2y ago
Doesn't 8B need at least 16gb of ram? Otherwise, your swapping I would imagine...
83.
by redman25 2y ago
Not the exact same comparison but I have an M1 mac with 16gb ram and can get about 10 t/s with a 3B model. The same model on my 3060ti gets more than 100 t/s. Needless to say, ram isn't everything.
84.
by redman25 2y ago
Minneapolis, Chicago, a lot of less temperate cities have protected walking tunnels, either underground or protected by buildings.
85.
by redman25 2y ago
I do web dev and prefer safari. It’s less power hungry on a laptop. If you develop for the lowest common denominator, every other browser just works. There’s something to be said for adding _less_ features to web browsers. A simple web is a
86.
by redman25 2y ago
What is the metric for LLMs? Shouldn't more than just accuracy be measured? If something has high accuracy but low recall, won't it be overfit and fail to generalize? Your metrics would give you false confidence in how effective y
87.
by redman25 2y ago
If you consider that evolution has taken millions of years to produce intelligent humans--that LLM training completed in a manner of months can produce parrots of humans is impressive by itself. Talking with the parrot is almost indistingui
88.
by redman25 2y ago
Why does AI have to be smarter than the collective of hummanity in order to be considered intelligent? It seems like we keep raising the bar on what intelligence means ¯\_(ツ)_/¯
89.
by redman25 2y ago
Ah, so this is a failure on nintendo's part for not protecting their email domain properly.
90.
by redman25 2y ago
I had no idea spoofing email sender was so easy. Does anyone know of a good way to defend against this? I've always taken it for granted that if the sender was a domain that I trusted that I could trust the email itself.

More ›