Merge branch 'main' into update-cpu-docs

Author: ioghiban

docs/getting_started/installation/gpu.rocm.inc.md

@@ -5,9 +5,6 @@ vLLM supports AMD GPUs with ROCm 6.3 or above, and torch 2.8.0 and above.
 !!! tip
     [Docker](#set-up-using-docker) is the recommended way to use vLLM on ROCm.
 
-!!! warning
-    There are no pre-built wheels for this device, so you must either use the pre-built Docker image or build vLLM from source.
-
 # --8<-- [end:installation]
 # --8<-- [start:requirements]
 

docs/getting_started/installation/gpu.xpu.inc.md

@@ -2,9 +2,6 @@
 
 vLLM initially supports basic model inference and serving on Intel GPU platform.
 
-!!! warning
-    There are no pre-built wheels for this device, so you need build vLLM from source. Or you can use pre-built images which are based on vLLM released versions.
-
 # --8<-- [end:installation]
 # --8<-- [start:requirements]
 

vllm/config/model.py

@@ -84,7 +84,7 @@
     "transcription",
     "draft",
 ]
-TokenizerMode = Literal["auto", "hf", "slow", "mistral"]
+TokenizerMode = Literal["auto", "hf", "slow", "mistral", "deepseek_v32"]
 ModelDType = Literal["auto", "half", "float16", "bfloat16", "float", "float32"]
 LogprobsMode = Literal[
     "raw_logits", "raw_logprobs", "processed_logits", "processed_logprobs"
@@ -141,6 +141,7 @@ class ModelConfig:
     - "hf" will use the fast tokenizer if available.\n
     - "slow" will always use the slow tokenizer.\n
     - "mistral" will always use the tokenizer from `mistral_common`.\n
+    - "deepseek_v32" will always use the tokenizer from `deepseek_v32`.\n
     - Other custom values can be supported via plugins."""
     trust_remote_code: bool = False
     """Trust remote code (e.g., from HuggingFace) when downloading the model

vllm/entrypoints/openai/serving_engine.py

@@ -105,7 +105,7 @@
 from vllm.pooling_params import PoolingParams
 from vllm.reasoning import ReasoningParser, ReasoningParserManager
 from vllm.sampling_params import BeamSearchParams, SamplingParams
-from vllm.tokenizers import MistralTokenizer, TokenizerLike
+from vllm.tokenizers import DeepseekV32Tokenizer, MistralTokenizer, TokenizerLike
 from vllm.tracing import (
     contains_trace_headers,
     extract_trace_headers,
@@ -1128,6 +1128,13 @@ async def _preprocess_chat(
                 messages=messages,
                 **_chat_template_kwargs,
             )
+        elif isinstance(tokenizer, DeepseekV32Tokenizer):
+            request_prompt = tokenizer.apply_chat_template(
+                conversation=conversation,
+                messages=messages,
+                model_config=model_config,
+                **_chat_template_kwargs,
+            )
         else:
             request_prompt = apply_hf_chat_template(
                 tokenizer=tokenizer,

vllm/tokenizers/__init__.py

@@ -1,6 +1,7 @@
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 
+from .deepseekv32 import DeepseekV32Tokenizer
 from .hf import HfTokenizer
 from .mistral import MistralTokenizer
 from .protocol import TokenizerLike
@@ -21,4 +22,5 @@
     "get_tokenizer",
     "cached_tokenizer_from_config",
     "init_tokenizer_from_config",
+    "DeepseekV32Tokenizer",
 ]