feat：Support QvQ-72B-Preview

xorbitsai · Dec 27, 2024 · e7b0406 · e7b0406
1 parent 7ba5660
commit e7b0406
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 2 deletions.
diff --git a/xinference/core/model.py b/xinference/core/model.py
@@ -76,7 +76,7 @@ class _OutOfMemoryError(Exception):
     "glm-4v",
     "MiniCPM-V-2.6",
     "qwen2-vl-instruct",
-    "qvq-72b-preview",
+    "QvQ-72B-Preview",
 ]
 
 XINFERENCE_TEXT_TO_IMAGE_BATCHING_ALLOWED_MODELS = ["FLUX.1-dev", "FLUX.1-schnell"]

diff --git a/xinference/model/llm/transformers/core.py b/xinference/model/llm/transformers/core.py
@@ -69,6 +69,7 @@
     "deepseek-v2.5",
     "deepseek-v2-chat-0628",
     "glm-edge-v",
+    "QvQ-72B-Preview",
 ]
 
 

diff --git a/xinference/model/llm/vllm/core.py b/xinference/model/llm/vllm/core.py
@@ -195,7 +195,7 @@ class VLLMGenerateConfig(TypedDict, total=False):
     VLLM_SUPPORTED_MODELS.append("llama-3.2-vision")
     VLLM_SUPPORTED_VISION_MODEL_LIST.append("llama-3.2-vision-instruct")
     VLLM_SUPPORTED_VISION_MODEL_LIST.append("qwen2-vl-instruct")
-    VLLM_SUPPORTED_VISION_MODEL_LIST.append("qvq-72b-preview")
+    VLLM_SUPPORTED_VISION_MODEL_LIST.append("QvQ-72B-Preview")
 
 
 class VLLMModel(LLM):