18:00:02 Meeting time. Hello! https://github.com/monero-project/meta/issues/1478 18:00:12 Hello 18:00:12 hi 18:00:21 Hello 18:01:10 Howdy 18:01:13 Hi 18:01:29 Hey 18:01:45 Alright, let's start with the reports from last week 18:01:56 waves 18:02:59 still AI reviews and merge conflicts for #9464, #10232, #10233 and #10819 18:02:59 will push all of them soonish hopefully 18:03:28 Pushed carrot_core (https://github.com/monero-project/monero/pull/9559) changes for this issue: https://github.com/jeffro256/carrot/pull/16 18:03:33 Me: investigating some issues found with general LLM audits of the codebase, rebasing the SAM PR, reviews of other PRs, etc. 18:03:44 I need to update spec for latest changes, but this should be done in a couple days 18:04:16 So the changes are settled, right? 18:04:47 I think that was quite a way to arrive at a solution 18:04:50 working on GUI binaries for stressnet v3.1 release, continuing on some more minor comments in FCMP++ integration PR's today 18:05:10 @rbrunner7: Not necessarily. Depends on if people find issues with the changes 18:05:18 But they are implemented as-is 18:05:46 @jpk68:matrix.org: "investigating some issues found with general LLM audits of the codebase" Is that feasible, does this give good results? 18:06:09 @jeffro256:monero.social: So let us call it a "release candidate" :) 18:06:57 @rbrunner7: I got an LLM to look through all of xmrack's existing reviews which were marked "pre-existing", and see which changes didn't yet have patches to fix those issues 18:07:09 The Readline PR also took up quite a bit of time, I forgot to mention that 18:08:02 I see. 18:08:15 @rbrunner7: Mainly, I want some confirmation from tevador that the changes are amenable on the Jamtis side, but yes 18:08:40 @jpk68:matrix.org: ya lineoise merged, now there are about 7 low stuff to fix from xmrack plus other things surely 18:08:40 all "low" 18:09:13 One licensing headache less 18:09:26 @thomasbuilds:matrix.org: Yep, I think most of them should be solved by my follow-up patch (#11619) 18:09:52 Please LMK if some still need work :) 18:12:17 By the way, does somebody still actively try to get us into some "free AI for open source projects" programs, or did that peter out already? 18:12:59 plowsof submitted an application to oss-scanner, which was approved by binaryFate (and has yet to be accepted) 18:13:00 Maybe "loose consensus" missing whether we really want to dance with those "devils", those AI companies? :) 18:13:38 Ah, interesting, that submission 18:13:55 Unless the MRL cluster can somehow divert a palet of AI inference card or NVIDIA GPUs that'll be hard 18:13:57 will it even get merged 👀 18:14:06 Yes, it was slopped by yours truly, and reviewed by selsta and Thomas 18:14:08 https://github.com/anthropics/oss-scanner/pull/1092 18:14:15 AI can find vulnerabilites and write code, unlike hackerone who is just passive MITM, dumb question 18:14:19 will cia get involved? 18:14:21 MRL cluster doesn't have any AI inference cards 18:14:35 oh i see what u mean 18:14:46 "dumb question" Which one? 18:14:56 @thomasbuilds:matrix.org: lmao isn't monero so hated there's a bounty for it by gov to break it or whatever 18:15:36 gingeropolous yeah i mean if MRL cluster could hypothetically becomes monero devs own AI factory that would open a new can of worm. 18:15:42 about AI companies, and they can scan anything privately already 18:16:23 maybe cia doesn't want us to know about the bugs 18:16:32 ok I'll stop being paranoid now 18:16:39 not being paranoid 18:16:56 oss-scanner shouldn't be monero's only AI scanning initiative 18:17:38 other contributors should continue to put pressure by their own 18:17:42 Are there other similar programs? Or do you mean we should also scan ourselves with AIs that we get through other ways? 18:17:52 the latter 18:17:55 Please link to any other programs/services if you're aware of them 18:18:46 afaik anthropic is the best and anything "open source" like kimi etc is super expensive to do and not as smart 18:18:50 So far seems to me people are quite enthusiastic using any AI that does the job that they can get hold on 18:19:00 expensive he said 18:19:01 Late again :/ focusing on getting ssl out the door, still one lingering issue with certificate expiration 18:19:30 thx you vtnerd i've been subscribed by time immemorial and i'm eager to see it merged 18:19:42 "kimi is expensive" Didn't hear that so far? 18:20:00 if kimi is expensive then what is claude 18:20:04 Ah, maybe you mean expensive to run it yourself? 18:20:06 well you'd want to use the best model they have ofc 18:21:07 what I can say is that try to buy usage credits from anthropic and u realize that a 200usd plan per month is like 10k on credits per month (rough estimation) 18:21:42 What does 10k credits buy you? 18:21:54 So that would mean that the days of those USD 200 plans are numbered? 18:21:55 yeah subscriptions have always been worth more than API but 10k I can't believe it 18:21:58 more like 2k 18:21:58 10k usd I meant 18:22:26 @rbrunner7: rbrunner7 anthropic will ban you for using your 200 usd plan at its maximum non-stop. 18:22:40 claiming distillation 18:22:50 Ah, those limited "unlimited" plans then :) 18:22:57 @syntheticbird:monero.social why do u think so ? 18:23:07 why do I think about what ? 18:23:23 that Max 20x plan quota is an illusion ? 18:24:09 Because I know people that are rotating over Claude max x20 accounts and that has been their experience so far 18:24:39 I'm not rotating and using single account non-stop, no problem, except part about getting it anonymously. 18:24:53 I paid like 400$ for 200$ just to get it anonymously 18:24:54 did you kyc ? 18:25:03 No 18:25:12 Unrelated question: The next release version will be 0.19. Does that mean that the next release will be for the FCMP hard fork, or will 0.19 be a non-hardfork release? 18:25:13 ok well if you say so 18:25:34 Somehow hard to believe that you would fare better just because you enrolled anonymously somehow? 18:25:57 fare better? what does that mean? 18:26:09 Yeah, it's a sep problem, and so far the only option I knew was MAGIC, but they don't provide anonymous access to LLM even if you are doing monero related research 18:26:12 Not bumping against any usage limits, as they claim 18:26:32 I'm using it for R&D and formal verification of current code and whetever I write myself 18:26:39 asked claude > <@syntheticbird> more like 2k 18:26:39 > My best estimate is roughly $11,000 to $13,000 of API-equivalent usage over four weeks, or about $2,800 to $3,250 a week 18:26:52 @interestingband:matrix.org: don't be shy we are all using claude to open source the ps5 18:27:16 @thomasbuilds:matrix.org: you are asking claude about anthropic economical equation ? 18:27:27 yeah who else am I supposed to ask this 18:28:01 @rucknium:monero.social: Selsta will know for sure, but I think there will be at least 1 non-hardfork 19.x release. Also to get that "out of the way" and not make the hardfork harder than it already is 18:28:23 yes, FCMP will be v0.20 18:28:32 nice 18:28:33 Monero v2 18:28:42 or monero v1 18:28:51 its less impressive 18:29:08 well it is actually 18:29:08 0.19 18:29:08 -> 1.0 18:29:35 and im trying my damndest to try and get enough data to warrant SoP in there. i don't think any wallet related stuff touches that does it? 18:29:36 makes me remember when they announced Minecraft 2.0 and it just rounded to 1.20 and 1.21, etc 18:29:38 I somehow doubt we will ever reach 1.0 ... 18:29:39 Time to short Anthropic on IPO day > <@syntheticbird> you are asking claude about anthropic economical equation ? 18:29:55 @rbrunner7: pretty sure it was decided from the beginning monero was zero versioned 18:30:49 I vaguely remember that selsta said there might be 1.0 possibly (I might be wrong) 18:31:08 @jberman:monero.social where can I (or anyone else) find info about scope for ongoing on upcoming audits ? is it still private ? 18:31:08 Ok, people, do we have something different and urgent to still discuss in this meeting? 18:31:51 @interestingband:matrix.org: following up with sgp, it's not private, just need the latest version from him that went out 18:32:46 @jpk68:matrix.org: are you still trading ? LOL 18:33:03 Does not seem to be the case. So let me close the meeting proper, and let the AI discussions flow further. Thanks everybody for attending, read you again next week!