Support Alibaba-NLP/gte-Qwen2-7B-instruct embedding Model (#1186)

Co-authored-by: Ying Sheng <sqy1415@gmail.com>

python/sglang/srt/managers/tokenizer_manager.py

@@ -94,7 +94,10 @@ class TokenizerManager:
             trust_remote_code=server_args.trust_remote_code,
             model_overide_args=model_overide_args,
         )
-        self.is_generation = is_generation_model(self.hf_config.architectures)
+
+        self.is_generation = is_generation_model(
+            self.hf_config.architectures, self.server_args.is_embedding
+        )
 
         if server_args.context_length is not None:
             self.context_len = server_args.context_length

python/sglang/srt/managers/tp_worker.py

@@ -94,6 +94,7 @@ class ModelTpServer:
             context_length=server_args.context_length,
             model_overide_args=model_overide_args,
         )
+
         self.model_runner = ModelRunner(
             model_config=self.model_config,
             mem_fraction_static=server_args.mem_fraction_static,