From 445321fab445394c38aa485c27a783acc01e1b83 Mon Sep 17 00:00:00 2001 From: Kyle Sayers Date: Mon, 6 Jul 2026 22:28:00 -0400 Subject: [PATCH] [Bugfix] [Quantization] Fix loading for CT DSV2 (#47780) --- vllm/model_executor/models/deepseek_v2.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vllm/model_executor/models/deepseek_v2.py b/vllm/model_executor/models/deepseek_v2.py index d9102cb5c9..134a6d84ed 100644 --- a/vllm/model_executor/models/deepseek_v2.py +++ b/vllm/model_executor/models/deepseek_v2.py @@ -829,7 +829,7 @@ def _try_load_fp8_indexer_wk( if "indexer.wk." not in name or "wk_weights" in name: return False # Weight is not an isolated WK weight for the indexer, ignore. is_weight = name.endswith(".weight") and tensor.dtype == torch.float8_e4m3fn - is_scale = "weight_scale_inv" in name + is_scale = "weight_scale" in name if not is_weight and not is_scale: return False # WK is not in FP8 format, ignore. # Buffer this tensor (weight or scale) until both have arrived.