update chat template of dsv32

vllm/renderers/hf.py

@@ -35,6 +35,7 @@ from vllm.transformers_utils.processor import cached_get_processor
 from vllm.utils.func_utils import supports_kw
 
 from .protocol import RendererLike
+import vllm.envs as envs
 
 logger = init_logger(__name__)
 
@@ -448,11 +449,17 @@ def safe_apply_chat_template(
         model_config=model_config,
     )
     if chat_template is None:
-        raise ChatTemplateResolutionError(
-            "As of transformers v4.44, default chat template is no longer "
-            "allowed, so you must provide a chat template if the tokenizer "
-            "does not define one."
-        )
+        if envs.VLLM_USE_V32_ENCODE:
+            from vllm.entrypoints.encoding_dsv32 import encode_messages
+            encode_config = dict(thinking_mode="thinking", drop_thinking=True, add_default_bos_token=True)
+            prompt = encode_messages(conversation, **encode_config)
+            return tokenizer.encode(prompt)
+        else:
+            raise ChatTemplateResolutionError(
+                "As of transformers v4.44, default chat template is no longer "
+                "allowed, so you must provide a chat template if the tokenizer "
+                "does not define one."
+            )
 
     resolved_kwargs = resolve_chat_template_kwargs(
         tokenizer=tokenizer,