ambiguous · filed under optimization
Scaling-law fit
A scaling-law fit is mentioned, but the evidence does not specify its fitting method or how it selects learning rate and batch size.
Also called new scaling-law fit.
- source
- 1
- model
- 1
- lab adopt it
- 1
- strongest
- used
How sources treat it
One count per evidence span, weakest treatment to strongest.
used 1
Documented in
Evidence
1 span quoted from the sources, strongest treatment first.
Downstream benchmark results demonstrate the robust convergence achieved by our new scaling-law fit
usedoptimizationin Qwen3.8-Flash-NextQwen
Filed alongside
Other methods under optimization :: learning-rate schedule.
Cosine decayWarmup-Stable-DecayBatch-size warmupTraining without batch-size warmupBatch-size warmup with adjusted peak learning rateBatch-size warmup with constant-batch optimum peak learning rateEngram learning-rate scalingFixed elevated learning rateLearning-rate annealingLinear warmup followed by cosine decayScheduled batch-size growthWSD-specific scaling law