Skip to content

[P0] Possible cause of missed inferences #629

Closed @tcharchian opened 2026-01-23 19:47 UTC 4 comments Updated 2026-04-28 18:28 UTC

We're distributing inference requests on the chain based on the total weight of the participant, not the weight of the participant's mlnode for a specific model_id. Seems like it's easy can be the cause of missed inferences (e.g. I have 100 nodes for model_id1 and 3 nodes for model_id2, but I get the amount of requests based on the weight of 103 nodes for model_id2)

💬 Comments (4)

@x0152 commented 2026-01-26 08:06 UTC

642

@tcharchian commented 2026-03-20 23:41 UTC

@tamazgadaev @IgnatovFedor @0xgonka do we want the same for preserved ML Nodes during PoC phase?

@tcharchian commented 2026-03-24 23:12 UTC

This issue is a subset of multi-model support (https://github.com/gonka-ai/gonka/issues/728) We'll figure out when to review and merge this issue during work on multimodels https://github.com/gonka-ai/gonka/issues/728 @x0152 @0xgonka

@x0152 commented 2026-04-28 18:28 UTC

Closing as resolved by the multi-PoC updates

🔄 Auto-synced from Issue #629 every hour.