epups

11 ms·

61.
by epups 3y ago
They should use more of these smart rockets, because so far their dumb ones have leveled Gaza and killed over 10,000 children.
62.
by epups 3y ago
I think the problem for people who are not in Germany is that manufacturers often release different versions for each market. So, for example, you might find that >50% of the models being tested by Stiftung Warentest exist only in German
63.
by epups 3y ago
> Keep in mind that modern quantitative approaches to LLM evaluation have been effectively co-designed with the rise of OpenAI, and folks like Ravenwolf routinely disagree with the leaderboards. Sorry but you're talking complete non
64.
by epups 3y ago
When do you think this switch has occurred, though? I mean, British debt-to-GDP ratio was 200% in 1815 [1], for example, and there are several other examples that suggest that most countries had their national debts peak before the past 10
65.
by epups 3y ago
What exactly are you referring to?
66.
by epups 3y ago
I appreciate the review, especially in making connections between fascism and economic policy. Historically, fascists had support from powerful economic groups because of their ability to deliver on highly unpopular economic measures. The c
67.
by epups 3y ago
I find it baffling that anyone would take these benchmarks seriously. The methodology is not transparent, and some of the tests completely unfair, like those in German or about specific niche German topics. The author readily acknowledges t
68.
by epups 3y ago
For me it's the opposite. Taking on a personal project like this one would be incredibly frustrating, because some relatively small steps would be huge time sinks. Now with LLM's, I am much faster, I just focus on the aspects I wa
69.
by epups 3y ago
Your reference does not seem to substantiate your claim. While interesting, it seems overly occupied with offering heavy moral judgement based on a modern perspective, to the point where it is exhausting and frankly annoying.
70.
by epups 3y ago
I don't buy this theory that what Musk primarily wanted with the hyperloop was to undermine the high-speed train. Yes I've read TFA and I'm aware of that indirect quote, and its retraction for that matter. Nevertheless, it se
71.
by epups 3y ago
"Somewhat reusable"? One Falcon 9 has already flown a record 18 times!
72.
by epups 3y ago
Google became so much of an ad company that they now confuse advertisement with actual product launches.
73.
by epups 3y ago
I've been driving with other people on the road for years. People might be worse on average but at least they don't get a firmware update that could potentially kill you.
74.
by epups 3y ago
You are right - upon closer inspection, even models that were not previously Mistral finetunes are now using Mistral in their later versions. I wasn't aware of it before as I could not filter results in the leaderboard (it doesn't
75.
by epups 3y ago
Good question. If you believe the results on the HuggingFace leaderboard ( https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderb... ), which I find very hard to navigate, we find that Mistral was not even the
76.
by epups 3y ago
Open source databases are SOTA or very close to it, though. Here the value proposition is to pay 10-50% less for an inferior product. Portability is definitely an advantage, but that's another aspect which I think detracts from their v
77.
by epups 3y ago
I understand how Mistral could end up being the most popular open source LLM model for the foreseeable future. What I cannot understand is who they expect to convince to pay for their API. As long as you are shipping your data to a third-pa
78.
by epups 3y ago
I think comparisons to base LLaMA are not so interesting, as almost no one is using those models. The most informative comparison is between Mistral 7B and 8x7B, provided in this picture: https://mistral.ai/images/news&
79.
by epups 3y ago
It sounds to me like you are trying to say that because technology exists to enforce oppressive stances against basic consumer rights we all had just a few years ago, we should also not make use of technology that exists to circumvent this
80.
by epups 3y ago
Benchmark results look awesome, but so does every new open source release these days - it is quite straightforward to make sure you do well in benchmarks if that is your goal. I hope Google cracked it and this is more than PR.
81.
by epups 3y ago
This is not a good example. One of them is under your control, the other isn't.
82.
by epups 3y ago
Yeah, the comparison to colonization doesn't really makes much sense to me. Using 16th century tech, you could expect to have a profitable trip pretty much every time you managed to come back alive, and you did not have to deal with br
83.
by epups 3y ago
There should be stronger consequences for this behavior. It costs hundreds of millions of dollars and several careers to deal with the outcome of data fabrication. Meanwhile, the worst case scenario for the fraudster is losing their job.
84.
by epups 3y ago
I don't find this argument very compelling. Technical work has been commodified to a great extent, even in bleeding edge applications. Does anybody have any doubt that OpenAI has timetables with clear action items that need to happen s
85.
by epups 3y ago
Thank you, I was looking for a similar tool and was surprised by how difficult it was to find something with no bloat. Will give it a shot
86.
by epups 3y ago
Based on the downvotes I am getting and the links posted in the other comment, I think you are absolutely right. People are acting as if ChatGPT is AGI, or very close to it, therefore we have to solve all these catastrophic scenarios now.
87.
by epups 3y ago
I'm familiar with the potential risks of an out-of-control AGI. Can you summarise in one paragraph which of these risks concern you, or the safety advocates, in regards to a product like ChatGPT?
88.
by epups 3y ago
Why can't these safety advocates just say what they are afraid of? As it currently stands, the only "danger" in ChatGPT is that you can manipulate it into writing something violent or inappropriate. So what? Is this some San
89.
by epups 3y ago
So, Adam D'Angelo is the only board member that remains, and he had also voted against Altman before. How interesting, considering all the theory crafting about him being the one who initiated this coup.
90.
by epups 3y ago
ChatGPT is a CSS deity, I have been using it lately and have been impressed by it. I suppose it's because so many issues were raised about it, so it is now trained on every possible mistake you can make

More ›