zby

13 ms·

181.
by zby 3y ago
I tried - but they wanted me to give them the data in Excel. Once I did the parsing, summing it up was trivial (and it would not fit in Excel anyway). That was quite a few years ago.
182.
by zby 3y ago
Maybe - but why they forget all reasoning and try to jump directly to the answer?
183.
by zby 3y ago
I wrote my own program to parse the logs I get from exchanges and compute the taxes.
184.
by zby 3y ago
> RAG is quite the hack. All you start with is the question, but why should the documents you need for the answer be similar to the question? I have always thought that a better RAG system would be to let the LLM use the search in the sa
185.
by zby 3y ago
So is it like https://en.wikipedia.org/wiki/Google_Wave - but with limited scope and with new algos and finally usable?
186.
Simplicity theory: teaching relevance to artificial intelligences(imtech.imt.fr)
1 points by zby 3y ago|0 comments
187.
by zby 3y ago
What is your theory? With the current high rates a lot of central banks are probably in the same situation as was Silicon Valley Bank.
188.
by zby 3y ago
That is just the first step in the reasoning. The interesting parts are always further along - see https://en.wikipedia.org/wiki/Five_whys .
189.
by zby 3y ago
I think this is premature optimisation. LLMs are the general tool here - in principle we should try first to adjust LLMs to search instead of doing it the other way around. But really I think that LLMs should use search as just one of their
190.
by zby 3y ago
I am convinced that we should teach the LLMs to use search as a tool instead of creating special search that is useful for LLMs. We now have a lot of search systems and LLMs can in theory use all kind of text interface, the only problem is
191.
by zby 3y ago
A great article - just a technical nitpick for the author: "in-context learning" is not what you do with RAG. "In context learning" is a really confusing name for reasoning by analogy. In RAG you provide source informati
192.
by zby 3y ago
If 400 microbes that contact your body makes you sick - then what if one virus gets into a cell and produces 400 copies? It is certainly possible: "For example, SIV, a cousin and model for the HIV virus, is released from infected T ce
193.
by zby 3y ago
I am waiting for social websites that don't allow any input on mobiles. Reading allowed - but replies only on a proper device. We don't need more low quality content.
194.
by zby 3y ago
So what are the alternatives to LangChain that the HN crowd uses? I see two contenders: https://github.com/minimaxir/simpleaichat/tree/main/simpleai... https://github.com/griptape-ai&#x2F
195.
by zby 3y ago
Sorry - myself I don't have much experience yet, I am at the research phase, but from what I read it makes sense to finetune the model to better understand the format used for calling the external tools including a search engine.
196.
by zby 3y ago
There is an article at the original site about that: https://www.anyscale.com/blog/fine-tuning-is-for-form-not-fa... Everybody new to this field thinks that he needs finetuning to teach the LLM of new facts. I made the
197.
by zby 3y ago
Adding to this: mimetic desire works on a subconscious level, imitation in buying attractive stocks is entirely conscious.
198.
by zby 3y ago
But there are programs that do inferences - and programs work on Turing machines which are just lookup tables (with an infinite tape). Lookup tables can do inferences.
199.
by zby 3y ago
So I guess after all those discussions we are still stuck with LangChain for everything to do with LLMs.
200.
by zby 3y ago
I have the feeling that we are at the MRP stage ( https://en.wikipedia.org/wiki/Material_requirements_planning ) when companies started using computers but writing software to handle production processes was so new that
201.
by zby 3y ago
If intelligence is knowledge and the ability to apply it then maybe we should try to split AI into two systems: https://zzbbyy.substack.com/p/reasoning-machines
202.
by zby 3y ago
This sounds like a paradox - but it is not. You don't give the bot the answer you expect - but only the ground facts, it generates the answer based on these facts by itself. This is a RAG system - and you need to treat it as a whole -
203.
You Need Rag, Not Finetuning(zzbbyy.substack.com)
1 points by zby 3y ago|0 comments
204.
by zby 3y ago
What mostly worked for me was two simple things: splitting the big tasks and constantly evaluating evaluations. With more smaller evaluations negative and positive errors anihilate each other. With adjusting the evaluations you can reach th
205.
by zby 3y ago
Is ERD part of UML? I find ERDs useful representations of databases. They can be generated automatically and are very good for quickly locating the right tables for a query. So I wanted to add a comment stating this to the parent - but the
206.
by zby 3y ago
Nice catch about Bacillus atrophaeus ! Can you quote something from the article about Serratia marcescens ? I cannot find it in the article.
207.
by zby 3y ago
(edited) This is such a clickbait. Everybody reading that title imagines that they tested spreading of some contagious disease (and put it in the same category as the Tuskegee Syphilis Study - https://news.ycombinator.com/i
208.
by zby 4y ago
Input is one thing - but also how much of a program can you see on a smartphone screen?
209.
by zby 4y ago
It is useful for users to learn about the alternatives.
210.
by zby 4y ago
Here is a quote I found in the article: "The decision dictates whether the securities are designated as “held-to-maturity” (HTM) assets " - HTM was an internal designation - an accounting convention. It does not have anything to d

More ›