APT4 transplant
Replacement of a Bielik v3 model's Mistral-derived tokenizer with APT4, followed by vocabulary adaptation and post-training.
Key apt4_transplant · version d5bb75c0-40ed-4d4c-afa4-0fded6fbcb94
Cited claim · P12 · Author-curated
The APT4 transplant is the replacement of a Bielik v3 model's Mistral-derived tokenizer with APT4, followed by vocabulary adaptation and the same post-training as the original Bielik v3 models.
Structured assertion
Relation: Defined as
- term
- APT4 transplantThe concept this claim defines.
- category
- model adaptation pipelineKind of thing the term denotes.
- replaced tokenizer
- Mistral-derived tokenizerTokenizer that is replaced.
- tokenizer
- APT4 tokenizerTokenizer that replaces it.
- adaptation
- Vocabulary adaptationTraining stage after the replacement.
- post training
- SFT, DPO-P and GRPO, as for the original Bielik v3 modelsAlignment after vocabulary adaptation.
Paper citations
arXiv:2604.10799v1 →Revision supplied by author
To mitigate catastrophic forgetting during vocabulary adaptation, we combined FOCUS-based embedding initialization with a two-stage continued pretraining pipeline (4B tokens with partial freezing, followed by 16B tokens of full adaptation) and applied the same post-training alignment (SFT, DPO-P, GRPO) as the original Bielik v3 models.
Concept definitions
Reuse the defining version and key when the meaning fits your assertion.
Defined as
The term denotes what its concept definition states; category names its kind and further roles give identifying properties.
Key defined_as · version f3b88c39-b184-4cb7-85e6-916f260ad0ad
Vocabulary adaptation
Adaptation of a pretrained model to a replaced tokenizer's vocabulary.
Key vocabulary_adaptation · version 95d5752d-3660-4d7c-97d5-f23b933cf433
Mistral-derived tokenizer
Tokenizer of the original Bielik v3 models, derived from Mistral's.
Key mistral_tokenizer · version bd3d8329-e626-4805-9fff-f83f146769a7
APT4 tokenizer
Polish-optimised tokenizer of the Bielik v3 PL models.
Key apt4 · version d497f94d-5373-4652-887c-55c001b6472c
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.