Fetching the paper…

Improving Low Compute Language Modeling with In-Domain Embedding Initialisation · Around