Affine tackles model queue backlog, adjusts economics

Affine reduced sampling speed from 1 day to 30 minutes but faces continued bottlenecks with 100+ models in queue and 2+ day evaluation times. Team increased minimum burn to 0.5 and lowered adjustment alpha to 0.5 to reduce registrations and combat spam. Environment changes (system prompt updates, Docker image updates) are under review for impact on evaluation scores.

• •Sampling speed improved: 1 day → 30 minutes; 100+ models still queued
• •Min burn increased to 0.5, adjustment alpha lowered to 0.5 to reduce spam registrations
• •Docker env images updated; system prompt changes may impact evaluation scoring