Conversation
Stop sending 84% of weights to escrow UIDs. Validators now fetch /current-kings and pay the reigning miner per modality, with last-known fallback if the API is down. Co-authored-by: Cursor <cursoragent@cursor.com>
Testnet 169 is not the hardcoded 379 mapping, and short epochs are needed to verify KOTH weights. Co-authored-by: Cursor <cursoragent@cursor.com>
Bittensor metagraph threads kept the process alive, and a 409 with --skip-chain was treated as failure. Cap chain retries and _exit. Co-authored-by: Cursor <cursoragent@cursor.com>
Keep the current king dominant while paying the previous two distinct crowned hotkeys a residual so near-ties are not winner-take-lane. Co-authored-by: Cursor <cursoragent@cursor.com>
Paying last-3 kings 85/10/5 is a breaking incentive change, so spec_version should move with a major rather than sit on 4.10.0. Co-authored-by: Cursor <cursoragent@cursor.com>
feat: set discriminator weights on current KOTH kings
Chain registration is optional. A failed or skipped commitment should not fail the CLI after the model is already in R2 and queued for exam. Co-authored-by: Cursor <cursoragent@cursor.com>
Miners get one image, one video, and one audio slot per registration. Exam failures do not count and a new round does not refill the slot. Co-authored-by: Cursor <cursoragent@cursor.com>
Keep a single Validating.md setup runbook and stop implying the full-benchmark score is TAO emissions. Co-authored-by: Cursor <cursoragent@cursor.com>
Match the tighter upload cap: a key cannot land one model per lane. Co-authored-by: Cursor <cursoragent@cursor.com>
fix: succeed discriminator push after upload; one submission per hotkey
Match the tighter upload cap: a key cannot land one model per lane. Also drop cloud-evaluated wording from the miner guides.
Replace the combined fool-rate bonus with a per-modality gate so only image >2% / video >1% (n>=20 benchmark evals) earn in that modality. Co-authored-by: Cursor <cursoragent@cursor.com>
… slots. Replace uniform 50-UID sampling with 36/8/6 buckets so validators keep challenging new and under-bar miners without letting them dominate the round. Co-authored-by: Cursor <cursoragent@cursor.com>
Replace the stale fool-rate bonus writeup with the per-modality pay gate and the 36/8/6 qualified/onboarding/probe challenge split. Co-authored-by: Cursor <cursoragent@cursor.com>
… torch. The test job skips the GPU stack; eager package imports were failing collection on the new qualification and allocation tests. Co-authored-by: Cursor <cursoragent@cursor.com>
…ble. A 200 with no parseable rows or a parse error was replacing the cache with an empty gate and burning the 16% pot. Co-authored-by: Cursor <cursoragent@cursor.com>
If gascli d push hits a used free slot, the CLI offers to burn SN34 alpha from this wallet, then retries the upload with the on-chain receipt so miners do not have to paste a transaction hash. Co-authored-by: Cursor <cursoragent@cursor.com>
A first push never mentions burning. A later push always walks through the 0.5 TAO SN34 alpha burn; declining or running non-interactively stops the upload. Co-authored-by: Cursor <cursoragent@cursor.com>
A second push still burns existing SN34 alpha when there is enough. Otherwise it spends 0.5 TAO in one add_stake_burn instead of staking and burning in two steps. Co-authored-by: Cursor <cursoragent@cursor.com>
A crash after the extrinsic lands should not charge a second 0.5 TAO when the same receipt can still claim unused credit. Co-authored-by: Cursor <cursoragent@cursor.com>
Allow the interactive prints, burn when staked alpha already covers the 0.5 TAO fee, keep the receipt until confirm, and start a new burn when a saved hash is already used. Co-authored-by: Cursor <cursoragent@cursor.com>
Sync main into testnet before generator qualification rollout
Fix cross-modality generator EMA carryover
…cation Persist generator qualification across validator restarts
Qualify generators on fool rate and allocate challenge slots
…ation Distinguish duplicate uploads from other 409 conflicts
Walk miners through a 0.5 TAO burn on a second push
Fix resubmit burn crash recovery and concurrent-push safety
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Promotes #446 (with #451, #453, #454) and #450 (with #456).
burn_alpha; insufficient alpha usesadd_stake_burn.The resubmission flow requires the upload API changes in BitMind-AI/bmcore#392 to be deployed first. Main's callback, memory-pressure, and burn-control fixes are preserved via #452.
Validation
Combined testnet tree (
60b9e4a): 163 tests passed; CI-selected Ruff checks passed.