Wow, gpt-5.6-Luna was already a fairly unbeatable bargain and now gpt-6-luna is both cheaper and better. And they did a phenomenal job getting gpt-6-sol to max out right where Astra begins; funny how they just so happened to avoid cannibalizing their best model while still being quite cost-competitive near the frontier.
At least it sounds good on paper, the the graphed results do give me pause as it seems the lower cost might come from a slightly nerfed base model combined with more thinking, going by the more erratic scoring curves and the lower no thinking baseline score. I’ll have to try it out but I really hope they haven’t nerfed Luna/Sol to make this price point possible!
Funny. Look up errata AMD-SB-7055: RDSEED Failure on AMD “Zen 5” Processors.
Zen 5 rdrand16/32 return zero with CF=1 on entropy exhaustion and their recommended approach directly leads to the issue you observed: treat all-zero result of rdseed as if cf=0 (failure) and re-roll the dice, effectively recreating the zen 1/zen 2 issue all over again!
They say this might be addressed by a future microcode update… meaning there’s a chance they’ll just patch it to do just that in software. Maybe that’s how they got into this mess in the first place?
Also, am I a complete idiot or is asserting the relative distribution of a mere 64k possible results a rather easy black box validation test that I would’ve assumed they’d be doing? When I used to write cycle-accurate emulators in the past, that would have been an obvious test to include. This isn’t some arcane instruction no one uses or a really complicated case with deep dependency and/or timing issues; it’s like getting rdtsc wrong.
I was going to suggest exactly that, if you're got an RNG, or pretty much anything else for that matter, you need the ability to return some sort of things-went-wrong-somewhere indicator value, and presumably AMD is using 0 to do this. Yes, there's also the CF, but the caller may not be checking that, particularly if it's being done from a HLL.
Has anyone checked whether it can return ~0, (signed) -1, the traditional error-return value?
That’s if you have disparate prompts and don’t want to actively select a model. If you’re developing a pipeline, it’s a terrible idea. You want to choose a model, validate it, then stick to it.
How do your path-only rules handle the many approaches for loading a file but making it appear to have a different path, such as bind mounts for one example?
Nice work! I wonder if you could (or did?) choose a more median shade for each color, especially blue? It seems, going by the comparison page, that most flags actually call for a less royal, more sky shade of blue, so you could use the median (on a linear RGB scale, or XYZ) shade instead as the designated representative for that color?
I noticed Rwanda seems to have the wrong aspect ratio, though you are able to encode that arbitrarily, no?
If you're on Windows (where Microsoft famously chose they would not get involved with flags and show country codes instead), execute this in dev tools or save it as a javascript bookmarklet to make the emoji flags show up:
It sounds like you lossily re-encoded lossy JPEGs to AVIF, am I reading that right? If so, you're automatically suffering from generational loss there, it can't be the basis of any coherent argument.
At least it sounds good on paper, the the graphed results do give me pause as it seems the lower cost might come from a slightly nerfed base model combined with more thinking, going by the more erratic scoring curves and the lower no thinking baseline score. I’ll have to try it out but I really hope they haven’t nerfed Luna/Sol to make this price point possible!
reply