Continual-ARCby unify

ReplayARC-AGI-1 · Workstreams · GPT-6 Luna, medium

Prime Agent, told to keep a program library, memory wiped

Told to keep a program library, with everything it wrote deleted before every instance.

Share learned
0 its own reset
Stream, budget used
22.6 solved 96%
Epilogue, budget used
20.1 solved 97%
Model spend
$3.87 4,061 calls
Instances
300 11.8 hours

Budget used is the share of the cap of 16 that an instance was charged, averaged over the instances: a solved instance costs 1 for each wrong answer and each demonstration, an unsolved one all 16, so lower is better. The share learned reads the stream's budget against the learner itself reset before every instance, and this learner keeps nothing between instances, so it is its own reset and reads 0.

Variant
Effort
Order

Loading the run