Moved

This repository was renamed on 2026-09-20 and holds no weights.

→ mp-juuuns/qwen3.5-4l-vocab40k-en-ko-headless

from transformers import AutoModel, AutoTokenizer

model_id = "mp-juuuns/qwen3.5-4l-vocab40k-en-ko-headless"   # was qwen35-standalone4l-classification-base
tokenizer = AutoTokenizer.from_pretrained(model_id)
backbone  = AutoModel.from_pretrained(model_id)

The artifact is unchanged: the same headless four-layer Qwen3.5 backbone with a 39,866-entry English/Korean vocabulary, its history, its benchmarks and its figures all moved with it. Only the name changed.

Why this stub exists. Hugging Face redirects the old URL, so links still work, but from_pretrained on an old id is reported not to follow that redirect (transformers#17582). This page is here so that anything pinned to the old id fails with an explanation instead of a bare 404.

Why the rename. A Hub search for qwen3.5 returns results containing no qwen35 names at all, and the reverse search is equally disjoint — the old id could not be found by the most obvious query for it. The old name also said classification-base, which named a head the weights do not contain, and it had no slot for the vocabulary size, which is the axis the grid members differ on.