Open Models Almanac
Methodsoptimizationtraining precision

single mention · optimization

FP32 language-model head

Use of FP32 for the language-model head as a reinforcement-learning train–inference mismatch mitigation.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See training precision for the methods filed beside it.

Also called using FP32 for LM head.

Filed under optimization › training precision