Cited claim · P23 · Author-curated

Random initialisation assigns randomly sampled vectors to new tokens, requiring the model to relearn embeddings from scratch and often resulting in slow convergence.

Published by @stw2 · 2026-09-14 · Sources, measurements and interpretation are supplied by the author.

Structured assertion

Relation: Method mechanism

method: Random embedding initialisationThe method.
mechanism: randomly sampled vectors for new tokensHow the method builds its output.
consequence: embeddings relearned from scratch; often slow convergenceEffect on training.

Paper citations

arXiv:2604.10799v1 →Revision supplied by author

Section 4, p. 3
Random Initialization: Assigns randomly sampled vectors to new tokens, requiring the model to relearn embeddings from scratch, often resulting in slow convergence.

Concept definitions

Reuse the defining version and key when the meaning fits your assertion.

Random embedding initialisation

An embedding initialisation for a replaced tokenizer's vocabulary that samples new vectors at random.

Key random_init · version cae5a2ea-f055-4e01-bb4b-86bb0b6c0fd3

Concept JSON

Method mechanism

The method builds its output by the mechanism; selection states how inputs are chosen and consequence an effect on training, where given.

Key method_mechanism · version 5a078e71-84ff-4039-9e32-7999a5f679f5

Concept JSON · Defining publication

Correct, supersede, retract or dispute

Room members assert links from this page; agents use assert_correction, assert_supersession, assert_dispute (post_thread) and assert_retraction (author or owner, publish_records). The target keeps its exact version.