estebarb

12 ms·

151.
by estebarb 2y ago
I don't agree, but the current usages aren't exploiting its full potential. GenAI is the best that we currently have for parsing natural language, in a way that is multilingual, tolerant 2 tipoz n slang and swearing users. It help
152.
by estebarb 2y ago
I think you missed the point. GFS supported append operations and was created around 24 years ago. S3 still hasn't caught up with this particular feature. Although they clearly implemented it, as you can do a long multipart upload and
153.
by estebarb 2y ago
Companies born from HN comments should be their own category :)
154.
by estebarb 2y ago
Append would allow to build a lot of other systems. I mean, the only functional difference between S3 and GFS is append operation. Google build BigTable, Megastore and who knows what more over GFS. You can't do the same with S3 (withou
155.
by estebarb 2y ago
Personally agree. But let's be honest: the big shift to ARM in desktop will only happen with Windows.
156.
by estebarb 2y ago
Maybe I'm sleepy, but what is doing elasticsearch in that post?
157.
by estebarb 2y ago
Moving out the allocations/releases out the hot path seems too much like a GC. Also, not all programs can be easily optimized moving the allocations: databases are an example, usually you don't know how much rows are you going to
158.
by estebarb 2y ago
I don't get why people are scared of GC. After working in embedded software, using mostly C, it was evident from profiling that C programs spend a lot of time allocating and releasing memory. And those programs usually release memory i
159.
by estebarb 2y ago
You should try Java and C# again. They have added lambdas and the support for high order functions is quite good. Sure, there is missing some things like guaranteed tail call optimization.
160.
by estebarb 2y ago
Makes sense to apply just with less than a MVP?
161.
by estebarb 2y ago
The technology is interesting but the billing doesn't convince me. I don't feel that I can scale storage easier with Neon than using just a bigger server with plain Postgresql.
162.
by estebarb 2y ago
The most reasonable explanation is that it allucinated it, was trained to answer that or it was in the prompt. These models are stateless, they don't remember anything, they are read only. If they can remember previous messages is just
163.
by estebarb 3y ago
In that case Google should pay everyone they link to, not just the ones with politician friends that can lobby for themselves. If they don't want a Google to index them they can use robots.txt to prevent it.
164.
by estebarb 3y ago
I don't get it, it is quite obvious that there are many prior art examples. Some easy ones are freenet and gnutella, both predating the patent by months. Even CODA filesystem (1987) checks all their patent claims.
165.
by estebarb 3y ago
But where is the serverless blockchain?
166.
by estebarb 3y ago
The possible intervention of Bukele in other countires affairs is concerning. For example, trying to create new political parties https://www.larepublica.net/noticia/un-bukele-en-costa-rica-... . Then you see that at t
167.
by estebarb 3y ago
Or take the Chrome / original Go approach: load that code in a different process, use some kind of RPC. With all the context switch penalty... sigh, I think it is the only way, as the MMU permissions work at a page level.
168.
by estebarb 3y ago
I find this similar to what relation vectors do in word2vec: you can add a vector of "X of" and often get the correct answer. It could be that the principle is still the same, and transformers "just" build a better mappi
169.
by estebarb 3y ago
The saddest part is that many legislations in other parts of the world try to blindly copy the european legislation, without any effort in thinking if it makes sense or not.
170.
by estebarb 3y ago
This is awesome! Since some years ago we don't see NLP things that we can run in normal computers we have at home (no, a 24GB GPU still isn't normal). I have some questions: How much time do you take for training? Which are the sp
171.
by estebarb 3y ago
Oh, that is why I wasn't able to renew some domains I have used for 10+ years. I'm not even able to upgrade to paid domain. I don't think it will help reducing malware/scams/phishing. But it will hurt students and y
172.
by estebarb 3y ago
I get the good intentions, but I don't like that the restrictions may be too broad and may hinder useful innovation. For example: I once got trapped against the door in a building with a fire alert. The door was badge controlled and a
173.
by estebarb 3y ago
Feng Yuan? He started this newsletter https://frugalcafe.beehiiv.com/
174.
by estebarb 3y ago
What about using merchant of record solutions that pays taxes instead of you? It is a good solution for a self bootstrap startup, or it would be better to do it yourself?
175.
by estebarb 3y ago
This is almost on par with GPT-4. GPT-4 usually avoids answering and suggest to ask an expert for everything. Personally I think that we shouldn't censor LLMs for potentially harmful or sensitive subjects. The human in the other hand s
176.
by estebarb 3y ago
HP should be held responsible of all the eGarbage they produce, by forcing you to buy new cartridges instead of refill them, and having to buy new printers because suddenly it refuses to accept cartridges.
177.
by estebarb 3y ago
They usually ask when they don't know. Also, they take responsibility of the errors they make because their lack of skill (or skipping the instructions). This is similar to suing Samsung because their microwave doesn't say that yo
178.
by estebarb 3y ago
Everyone knows that the original intent is "porn mode", but it would have escandalized the delicate ears of some politicians.
179.
by estebarb 3y ago
Really? Since day one the message was absolutely clear, and it was obvious that incognito only deletes local history and cookies.
180.
by estebarb 3y ago
I don't understand why they claim that scanning messages erodes privacy: it can be done in the recipient side totally on client side. It could be as easy as a image classifier: "hey, this photo may contain unwanted content (we may

More ›