Quote from: ABCbits on Today at 08:49:30 AM

Quote from: Paulfontrahel on July 07, 2025, 07:23:35 AM

I do not understand anything about building a code and turned to gpt chat plus. In the end, he made me some code on sagemath, but for 3 weeks now I have not been able to launch it, all the time the babai error, this damn gpt has already shaken all my nerves, it changes some parameters in a circle, but the result is the same, the error still does not disappear.

Quote from: Paulfontrahel on July 07, 2025, 11:30:33 AM

As for bias, I asked gpt chat to analyze all my signatures from the blockchain, and eventually got an answer (so as not to write too much, I'll publish channels and windows from the code that gpt made for me).

--snip code--

gpt has conducted various analyses and the results are almost always the same, the top 100 signatures have a leak of 3.2 bits per signature, this is on the condition that we did not take channels that give less than 0.3 bits per signature, respectively, for 100 signatures the total leak volume is decent and if you read the work of Biased Nonce Sense, then this is enough for success with a good margin. I really want to believe that this is so, otherwise this is my last lead.

Have you verified that technical statement from ChatGPT is correct/accurate? After all, you can't even launch code it generate.

Quote from: Paulfontrahel on July 07, 2025, 12:57:56 PM

If I really have this vulnerability, then I don't need to look for my old PC in the trash to restore access. I asked the forum for help.

IMO trying to recover data on your old PC is more sensible option. After all, there are already known working tool to make raw copy of your storage drive and known working tool to recover delete file (assuming it's not mostly/fully overwritten).

No, I haven't checked it through any third-party channels. Only the results that gpt itself gave me and that's it. The code doesn't run, that's not quite right, I didn't go into details. The code runs, parses signatures from the file correctly, correctly, exactly as gpt wrote in the analyses, counts the number of bias bits, runs the preliminary LLL, but when it goes to the BKZ block, it immediately falls into the babai loop error. I've already tried the code on sagemath and compiled docker and on pure python, no, the same problem is everywhere. Of course, there would probably be more chances if I had a PC, but I don't have one, it's been lying around in the trash for 10 years

. I'm currently writing messages to the authors of the work I refer to, maybe they'll tell me. Thank you very much for at least some responses, that you don't ignore me.