Rollo62

Hi there, I received my OpenAI beta test account recently, after waiting for that more than a year now. Finally I was very much excited about the AI capabilities, as I watched the amazing Youtube series of GPT3 Leta from Dr. Alan Thompson, who is experimenting with the DaVinci model and avatars to provide a real human chat experience. This machine will easily pass the turing test IMHO. Well, but when I first checked the Playground, I found the GPT3 models were not so much evolved as I expected. I also tried out the REST API, with the same disapointing outcome. GPT3 often gives completely wrong answers, mostly doesn't understand anything at all, not even a name or very basic texts, and all in all its far from capabilities of the Leta series I mentioned above. So maybe this is the case because I have not provided enough or wrong training data, but there is also no real explanation howto provide information, and especially what data would be really needed, to enable GPT3 to deliver most useful results. In the meantime during waiting for GPT3 access, I already tried GPT-J and tried others, but they all were not especially "intelligent" neither in understanding normal texts nor in providing real world answers (like: who won the latest nobel price in physics ?). Now I would say GPT3 is on a same level as GPT-J, at least when it comes to a unspecific chat application. What is Dr. Thompson's secret to make it so clever ? I assume that all GPT3 models use the same training data ( it was mentioned that training data was used only from the time until year 2018 ), so that also Dr. Thompson's machine should be using the same data. I wanted to ask if anybody here in DP had also some AI experiences, and maybe could squeeze better results out of GPT3 ? Would be great to exchange some tips and hints. howto handle AI best.

I absolutely agree, I have Alexa, Siri (only Google Assistant is more clever ), but they all are miles away from this Leta series. I am not talking about those, but the GPT3 (Leta) videos in the link above, which is in an experimental, closed beta. Are the Leta videos then a hoax finally, I dont think so ? There are about 24 videos right now, and I think Dr. Thompson is close to many AI research teams, having a deeper insight as we could have. Since GPT3 is only a textual processor, maybe the Q/A is somewhat "beautified" before sending to the avatar and speech output. But many of Leta's answers are really astonishing, and would make AI really useful.

Don't you see any use-case, I wonder ? If the a device chat can give perfect answers, like the Leta project can do, I see a lot. Just as example for assisting people in their daily worksite, in real language, maybe by a special trainned AI which is optimized to the special field of the customers current working tasks. The AI could act as a real useful assistant, giving useful tips. But that is only one of many ideas, for the others I better let you sign an NDA No not really, I think there have been already millions of people thought about the same ideas, like always if a new technology is emerging. While OpenAI, Apple, Facebook and those kind of companies ahve already filed their patents. "... increase in income ..." Not sure how you mean that, but I think: Yes, such AI gadgets in apps and software will sell veeeery well to a lot of company and private customers in the near future, if the AI can really do a good job.

Hi there, I want to wave goodbye to Delphi 10.4.2. This was for me one of the most stable versions in the past, maybe only XE8 was similar stable and less critical for me. Yes, it had its hickups here and there, but nothing too annoying at all. It accompanied me in the last 12 months as true friend, so I wish you luck in the afterlife. Looking forward to see Rx11 as stable as its older brother, and to give me another 12 months cycle of convenient Life. The start looks very promising, lets see ...

Thanks to Darian I've noted that I had missed to update my IDE version in DP for quite some time. Just updated, of course its Rx11 now, and was Rx10.4.2 before. A good time to remind everybody to check their profiles too.

Do you mean something like Anchor ?

@Darian Miller Thanks for such list, that is very helpful. Such a general "Can-I-Use" list would be a big step to more quality, but I'm afraid this will end as fulltime job. I'm not sure if preparing such useful lists could be done with Github or other tools on-the-fly, maybe so. Anyway, I think this would be a perfect community work, to prepare especially something like Delphi polyfills, to add certain "workaround" to each issue. If anybody knows how such work could be "automated", that would be great, maybe this could be even a side-project within DP ? Anyway, my recipe to avoid most issues are the following: - don't use the latest features immediately, just observe and test, but let them "stabilize" - avoid using the latest fancy hot components, like TNumberBox, just take the time to build your own from more basic components ( I highly recommend frames for this ) - try to use only minimal, absolute necessary external components, to avoid too much external bindings - modularize your libraries, so that they can be easier be managed and replaced (if needed ), single responsibility principle is also very much true for units IMHO, so I try to decouple functions into separate spaces. - tweak the Delphi core units, whereever needed, to workaround issues and to receive the desired behaviour you want to have ( record helper or interposer) - UnitTest most of your work, to be able to spot sudden changes early

I think you could bind the Tag property too, with LiveBindings. For me the whole LB thing is also still very much unclear, whats it's BestPractices are, to get best results out of it. Until now: "Handle with care"

Why, it looks OK to me. The index shows the entries from 0 ... nn, the Tag determines what item you have. If you filter, then the index might change, I think that is correct.

@mitzi Thanks a lot. This could make some backed up WinXP VMs with old XP projects obsolete, if we can compile them in current Rx11 too. Lets utilize the free'd disk-space with some new, astonishing and modern FMX project's

Simpler, yes, but the "Ex" receives also more information Especially the ItemObject is quite handy, to select the right sub-component in the TListItem, which you are interested in.

In ancient past versions I had some issues from time to time, on various platforms, not sure what the root cause of them was. Anyway, I only can say that since I moved completely to ItemClickEx I had no more issues. Maybe everything is OK now, in the last Delphi versions.

I rely on ItemClickEx only, for all purposes, this works best IMHO.

Not sure if you really propose to use that ? The first seems to be a joke only, and will be probably offline very soon. The latter could be self-hosted, but requires node.js environment. When looking for the simplest possible solution, why not looking after PHP ? The sites can be very cheap, and even if you install WordPress (ab)using its REST API, it could not be much simpler. If you set up a REST API only, thats also quite simple with PHP too. What is your main goal (Price, Self-Hosting, 0-Setup, Stability, Safety, Longterm availability, Support, Memory footprint, Performance, ... ) ? Thats a little unclear to me, from your examples above.

Thanks for the info, that looks very interesting. What is unclear to me, does it depend on Bass.dll ?

It has good and bad sides, as always.

Yes, that is what they WANT you believe .... I still prefer to have cash in the pocket. Digital currency, not bound to any real value, allows the FED's and covernments to set its value at will. This opens all doors to very ugly manipulations, control and value transfers, which is controlled from the government. They tell us the story of crime and corruption, so that we forget to give it a second thought. For example, the EU FED already has their hands already in sorts of directly supporting countries, governments, banks, etc. ... WAIT: What was the one and only task of the FED again ? Yes, to ensure price stability (NOTHING ELSE).

In Germany banks silently rised their fees and conditions over the past years. Only recently a judgement was mad, which allows bank customers to set back to their original conditions, when they opened the account. Unfortunately this goes back unly until 2018, but at least this may count up to 200-300 EUR, or maybe more, depending on the use case. I think it was an EU law that triggered the judgement. Yes, I also think the fee's are to high ( for a fully automated process ). The banks will get obsolete anyway in a few years, when all countries managed to launch "digital currenries", and the FED's take over. I think no one needs these greedy imtermediate money grabbers anymore soon ( but then we're dependent from FED's and completely transparent, which opens a whole bunch of other nasty issues too )

Interesting idea, you intend to open PDF in the default browser, right ? What prerequisites do you need for that, does all modern browser have PDF viewer already integrated ? I think Firefox can do that without any AddOn, but I'm not so sure about all the others ( Chrome, Edge ).

This is a class, not an enum. I think the advantage of enum ist, that you can use a "speakable name" to identify a numeric value. If you enhance this class to store also the name of an enum, you get closer, but you still cannot simply use "AVariable := TMyDynamicEnumType.FancyEnumName; ", I think this is what bravesofts is asking for. Ok, you could use specific class properties as names for each enum, but still hard to dynamically enhance that.

You mean to define a NEW type, like that type TFileTypeRange = ftPDF.. ftXLSX; I think you can do that in certain local scopes and places, but does this really make sense ? The real goal of enum's is, when they were shared and used to guarantee correct values between several units, like pre-defined constants.

Not sure if you're concern is the term Set or Enum in the type name. I usually try to use singular, plural nomenclature anyway ( with or without Set/Enum mostly ). TMySpecialValue = (svOne, svTwo, svThree, svFour, svFive); TMySpecialValues = set of TMySpecialValue; or TMySpecialValueSet = set of TMySpecialValue; That could be mixed e.g. with array or list, but in the code it will be directly clear usually.

Thanks for pointing to this topic, I have not heard that term either. After fastly checking, it seems to be somewhat in current discussions. From my experience, building common banking API standards was always massively hindered by the banks and institutes itself in the past. Did somebody or something stimulated the 3 brain cells of banking managers recently, towards 21st century

@angusj Hi Angus, I'm still playing with your nice Image32 library. It behaves really well under all circumstances. What I'm trying to achieve is to open an SVG, and to have the ability to access certain elements and/or layers, to make them visible/invisible, including all their possible sub-elements in the tree. So my goal is to have a complex SVG, with certain layers, where I can control the rendering on-the-fly at runtime, like on/off changing color, etc.. The internal structures are somewhat hidden, and the SvgReader/SvgParser seems to decode sub-elements on demand from text of TAnsi record. I have tried to make the internal SvgTree structures accessible, but that doesn't help much to get easy control over the certain elements. The .text property seems to hold the whole node data as string text, with the text.AsUTF8String; it only returns a certain part, I assume only the <style> part. Since .text is a pointer, I'm not sure if the following text is usuable at all, but its looks that its containing the current node as string. The Id and other identifiers seems all to be hashed, so I cannot retrieve and find nodes easily in the tree. I can see that the internal structures are optimized for best rendering SVG, which it does very well ( even SVG that fails elsewhere ). My question is, if there is a simple way to access and control the SVG-tree on the fly, to control parts of that rendering process ?

@angusj Thanks for the proposal, yes that works well in some tests. I had considered before that this might add too much additional redudancy, since the same data is copied or hashed maybe in other places still, and I have tried to restore the hashed data before. In reality, after a second thought, this additional data doesn't matter much. The reason is that only a few "Id's" will be used in the whole SVG, so thats irrelevant regarding redundancy.