On this day: Jun 12

Large-scale errors were found even in a top-difficulty benchmark, raising doubts about the reliability of its review once again

A direct counterargument to the spread of multi-agent systems. Reaches the opposite conclusion from the next entry, one day apart

The axis of vector DB competition moved from adding features to cutting the cost of billion-scale vectors

Showed that the conflict between the two laws stems from counting non-embedding parameters and from analysis at small scale

Distillation became so cheap that it no longer even needs API calls

Self-attention alone, without recurrence, made parallel training possible, establishing the architectural premise for large-scale pretraining