framework
active
framework:chinchilla-scaling-compute-optimalChinchilla scaling (compute-optimal)
Source of exponents α≈0.46, β≈0.51 and compute estimates.
Neighborhood — ranked by edge-count
Papers (1)
paper
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Optimal number of features scales faster than optimal number of training steps with compute budget.finding0.743Allocation result from scaling laws.
- Scaling aggregated gradient by the maximum gradient norm among tasks.
- Process by which low-level competencies scale to larger problem-solving abilities across levels of organization.
- Naive weight magnitude works here.
- Recommended strategy for gradient normalization.
- Supports the claim against single-layer probing approaches used in prior work.
- Hyperparameter trend observed.
- Asserts that the method maintains efficiency across a range of constraint strengths without degradation.