From 1fb0495a727686750525ee294d2274f4ef36525b Mon Sep 17 00:00:00 2001 From: FredericOdermatt <50372080+FredericOdermatt@users.noreply.github.com> Date: Fri, 6 Feb 2026 15:23:03 +0100 Subject: [PATCH] [FIX] guidance: use max(vocab_size, len(tokenizer)) for n_vocab (#33509) Signed-off-by: Frederic Odermatt --- vllm/v1/structured_output/backend_guidance.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vllm/v1/structured_output/backend_guidance.py b/vllm/v1/structured_output/backend_guidance.py index 6d1e4b7f7..6063a2dc2 100644 --- a/vllm/v1/structured_output/backend_guidance.py +++ b/vllm/v1/structured_output/backend_guidance.py @@ -93,7 +93,7 @@ class GuidanceBackend(StructuredOutputBackend): ) self.ll_tokenizer = llguidance_hf.from_tokenizer( - self.tokenizer, self.vocab_size + self.tokenizer, max(self.vocab_size, len(self.tokenizer)) ) def compile_grammar(