Compute-Optimal Scaling for Value-Based Deep RL | Read Paper on Bytez