From 22be45c78cd0059988edb594b6778a9a45bae6cf Mon Sep 17 00:00:00 2001 From: Kingsley Date: Sun, 30 Nov 2025 09:36:36 +0800 Subject: [PATCH] [misc] fix omni thinker load (#9552) Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> --- src/llamafactory/model/model_utils/moe.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/llamafactory/model/model_utils/moe.py b/src/llamafactory/model/model_utils/moe.py index 8a66621d..121c5488 100644 --- a/src/llamafactory/model/model_utils/moe.py +++ b/src/llamafactory/model/model_utils/moe.py @@ -127,7 +127,7 @@ def add_z3_leaf_module(model: "PreTrainedModel") -> None: _set_z3_leaf_modules(model, [Qwen3VLMoeTextSparseMoeBlock]) - if model_type == "qwen3_omni_moe": + if model_type in ("qwen3_omni_moe", "qwen3_omni_moe_thinker"): from transformers.models.qwen3_omni_moe.modeling_qwen3_omni_moe import Qwen3OmniMoeThinkerTextSparseMoeBlock _set_z3_leaf_modules(model, [Qwen3OmniMoeThinkerTextSparseMoeBlock])