M4v3R

19 ms·

241.
by M4v3R 3y ago
What you're saying directly contradicts the shared conversation log. They cannot just inject "bits and pieces of newer information" to the old models as far as I know. Also the responses from the model are too detailed and sp
242.
by M4v3R 3y ago
How are you prompting it to extract the data?
243.
by M4v3R 3y ago
While it didn't point that out specifically in the first message it actually did recognize that when I asked it about it, without suggesting anything: https://i.imgur.com/OE9tCX4.png > Should I attack while it'
244.
by M4v3R 3y ago
Yeah, I saw that, and that's a classic jailbreak to which all GPT models are vulnerable to some extent. Although in this specific case apparently just telling the model to not treat the image as instructions helps: https://p
245.
by M4v3R 3y ago
Sure, the tech was out there for quite some time but never before the quality of the output was so good, it's almost (not 100%, there still are mistakes and hallucinations ocassionally) on par with a human, which to me is really stun
246.
by M4v3R 3y ago
Good call, but I did make sure to strip any metadata from the photos before I uploaded them :)
247.
by M4v3R 3y ago
GPT-4V is mind blowing, it surprising to me that it gets so little attention here on HN, because after playing around with it I get the same sense of excitement I got when I tried the original ChatGPT. The level of understanding of what is
248.
by M4v3R 3y ago
It's actually a pretty cool feature (when it works as intended), I once saw this dialog when visiting a friend who also had a pair of Airpods and we mixed them by accident.
249.
by M4v3R 3y ago
They do, but I think the OP didn’t use Bun's native serve methods and instead used the same code for both Node and Bun, hence the subpar performance.
250.
by M4v3R 3y ago
> Yes, there are some extreme ... fanatics(?), who can live with a phone that doesn't have YouTube, Gmail, Google Maps, Google Drive or Play Store in it, but the Joe/Jill Regular will never ever go for that today. I would fit y
251.
by M4v3R 3y ago
> My understanding is that complex things like statues were not supposed to be possible 12,000 years ago. I think it’s supposed to be more like 6000 years ago I think your understanding was just incorrect. Here's a video [0] from Ku
252.
by M4v3R 3y ago
This lowers the barrier to take unwanted photos/videos significantly. Yes, surveillance cameras are everywhere but your everyday person does not have access to their feed (and the publicly accessible ones are not that common and usuall
253.
by M4v3R 3y ago
Are we reading the same blog post? The post that parent posted literally reads: > "Raspberry Pi 5 has been designed to handle typical client workloads, uncased, with no active cooling".
254.
by M4v3R 3y ago
Where did you get the 700 gigabytes figure from? I don't think OpenAI even released the model size, although it's considered to be 175B parameters. Given how well quantization works at these sizes you would need less than 200 GB o
255.
by M4v3R 3y ago
I’ve heard that take many times, but is there any actual data to back this claim? I mean think about it - your average thief is not the kind of person that will know what part setialization is or how to tell iPhone models apart to know whic
256.
by M4v3R 3y ago
Because it depends on the person. For me autocorrect is 99% spot on and so it’s a valid trade off for me to deal with correcting the remaining 1%. My wife on the other hand cannot live with it so she turned it off. And every time I type som
257.
by M4v3R 3y ago
I'm annoyed and I don't use Chrome. Do you have any statistics about browser usage of the HN crowd that backs your assertion?
258.
by M4v3R 3y ago
Hey Jarred, congrats on the 1.0 release! I've tried this release on one of my projects and encountered some errors, where is the best place to send reports about compatibility issues?
259.
by M4v3R 3y ago
That's just tech-speak. To the end user these are two separate apps. And Meta definitely has a way of telling which bits of data came from Threads and which from Instragram. They just chose not to make it easy for the user to delete
260.
by M4v3R 3y ago
No mention about whether they use their own in-house LLM or whether they use a 3rd party like OpenAI, does anyone know more about this? Edit: from the linked Privacy Statement: > Contracted service providers may also deliver artificial i
261.
by M4v3R 3y ago
I just asked GPT-4 your first question and got this in response. It's still a correct result, but with unnecessary steps: > Fill the 12-liter jug and pour water into the 6-liter jug until it's full. Empty the 6-liter jug and po
262.
by M4v3R 3y ago
> - install node / npx > - npx create-next-app@latest > - realise your node is too old Said no one ever. I can fully understand you not having Node installed (although having any kind of toolchain installed is a prerequisite f
263.
by M4v3R 3y ago
You literally just run: > npx create-next-app@latest Press Enter a few times for default settings and voila, you have a hello world app up and ready to run.
264.
by M4v3R 3y ago
Because GPT-4 gets this right. Also please look at my 2nd edit which is a concrete programming example that Phind's model gets wrong on several levels.
265.
by M4v3R 3y ago
More specifically, they've beaten an old measurement on one specific benchmark. In my personal experience however GPT-4 is still ways ahead CodeLlama and its derivatives. They are really good for what they are, but you just can't
266.
by M4v3R 3y ago
Thanks for the info!
267.
by M4v3R 3y ago
It's not getting close to GPT-4. It's getting closer according to synthetic benchmarks but spend any non-trivial time with both and you'll quickly realise GPT-4 is still leagues ahead, especially for writing code and more com
268.
by M4v3R 3y ago
A 4060 Ti with 16GB memory costs around $700-800 over here, can I connect two of these to get 32 GB memory to run the bigger models (13B stock or 30B quantized)?
269.
by M4v3R 3y ago
From the OP: > Furthermore, we applied OpenAI's decontamination methodology to our dataset to ensure valid results, and found no contaminated examples.
270.
by M4v3R 3y ago
> You are ignoring the fact that a toddler, once their musculature develops, is able to learn to walk after several tries. It's definitely not several tries. I've actually witnessed that very recently with my daughter. It takes

More ›