From aa375dca9fbeff03904cd7b7dcc5014bfa19b0fb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Szymon=20O=C5=BC=C3=B3g?=
 <58388001+SzymonOzog@users.noreply.github.com>
Date: Fri, 7 Feb 2025 06:35:09 +0100
Subject: [PATCH] [Bugfix] Missing quant_config in deepseek embedding layer
 (#12836)

---
 vllm/model_executor/models/deepseek_v2.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/vllm/model_executor/models/deepseek_v2.py b/vllm/model_executor/models/deepseek_v2.py
index 0c6f07ce7b112..fd0e58fa1458d 100644
--- a/vllm/model_executor/models/deepseek_v2.py
+++ b/vllm/model_executor/models/deepseek_v2.py
@@ -581,7 +581,8 @@ class DeepseekV2Model(nn.Module):
             self.embed_tokens = VocabParallelEmbedding(
                 config.vocab_size,
                 config.hidden_size,
-            )
+                quant_config=quant_config,
+                prefix=f"{prefix}.embed_tokens")
         else:
             self.embed_tokens = PPMissingLayer()