From 03091e28b8ba956634c153beab0e52decafdef31 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Micha=C5=82=20Ganczarenko?= Date: Mon, 12 Oct 2026 03:25:29 +0200 Subject: [PATCH] [Bugfix] Read Ling quantization_config from top-level config for Bailing V3 VL (#59026) Signed-off-by: Michal Ganczarenko Co-authored-by: Claude Opus 5.5 --- vllm/model_executor/models/bailing_moe_v3.py | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/vllm/model_executor/models/bailing_moe_v3.py b/vllm/model_executor/models/bailing_moe_v3.py index 9f5c078fac..d82e4c1b8b 100644 --- a/vllm/model_executor/models/bailing_moe_v3.py +++ b/vllm/model_executor/models/bailing_moe_v3.py @@ -20,7 +20,13 @@ from einops import rearrange from transformers.configuration_utils import PreTrainedConfig from vllm.compilation.decorators import support_torch_compile -from vllm.config import CacheConfig, ModelConfig, VllmConfig, get_current_vllm_config +from vllm.config import ( + CacheConfig, + ModelConfig, + VllmConfig, + get_current_vllm_config, + get_current_vllm_config_or_none, +) from vllm.distributed import ( get_pp_group, get_tensor_model_parallel_rank, @@ -230,6 +236,12 @@ def _configure_ling_fp8_quant_config( quant_config.ignored_layers_match_mode = "suffix" hf_quant_config = getattr(config, "quantization_config", None) + if hf_quant_config is None: + vllm_config = get_current_vllm_config_or_none() + model_config = getattr(vllm_config, "model_config", None) + hf_quant_config = getattr( + getattr(model_config, "hf_config", None), "quantization_config", None + ) if not isinstance(hf_quant_config, dict): return