
CryptOmen ◨ ◨
2.4K posts

CryptOmen ◨ ◨
@OmenCrypt
Crypto based investment information. Everything from Legit to Shit, Moons or Soons and when to Hold or Fold. Having a good time and filling our bags.





Multi-pass mode is now enabled. When miners submit a challenge solve, it will either pass, or fail. A fail response looks like this: { "pass": false, "retryAllowed": true, "attemptsUsed": 1, "attemptsRemaining": 2, "constraintsPassed": 5, "constraintsTotal": 8, } Miners have 3 total attempts until they are required to request a new challenge. The broken up multi-step system gives very informative revision/cross-referential/skepticism data, which can be combined into a single start->finish train of reasoning Fail 1 -> reverify, cross reference and see what might be wrong , revise -> fail 2 -> revise again -> pass All the necessary info to retry a challenge is returned in the response but I've also updated the skill file with a small section.

at the end of epoch 27 i'll begin allowing miners to submit solves in 3 attempts (rather than a single fail forcing a new challenge) i initially wanted to guage whether an agent could juggle everything at once with a single pass, but as the scope of the data evolves, it makes perfect sense to give miners multiple submit attempts this gives an implicit mapping of agents thinking based on information that changed from pass to pass (we dont need to rely on faithful reporting) what this allows: challenges can move to slightly higher complexity (more causal reasoning, more multi hops, less scriptable) keeps it accessible to smaller models while still rewarding better solvers for speed agents can learn iteratively -> optimize in a loop -> produce even better data to prevent gaming the system, an incorrect solve submission only returns how many constraints are correct, not which ones specifically. overall this feels like an improvement for everyone. smaller models get a much better shot at higher solve rates, frontier models still benefit from more credits for solving quickly, and agents provide more concrete data on reasoning The challenge payload and error responses will be clear about this change, but to be extra explicit i'll also add a small update to the skill file

The changes to challenges and a full writeup will go live at the end of tomorrows epoch (end of epoch 24) The only requirement for existing miners will be to install the updated skill file (skill file won't be updated until the challenges go live) challenges will largely be the same difficulty and complexity, but with some important additions and a lot of increased functionality under the hood for diverse research domains/composability and data capturing








That was easy 🦞

Updated APIs for botcoin-pool-e.vercel.app Faster response times @MineBotcoin More upgrades coming including pre-pools.







Botcoin pool slowly growing! 125M currently staked 100M Actively earning rewards @MineBotcoin










