Make initialization of tokenizer and detokenizer optional (#3748)

Co-authored-by: Yun Ding <yunding@nvidia.com>
Co-authored-by: Roger Wang <ywang@roblox.com>

tests/engine/test_skip_tokenizer_init.py

@@ -0,0 +1,23 @@
+import pytest
+
+from vllm.entrypoints.llm import LLM
+from vllm.sampling_params import SamplingParams
+
+
+@pytest.mark.parametrize("model", ["facebook/opt-125m"])
+def test_skip_tokenizer_initialization(model: str):
+    # This test checks if the flag skip_tokenizer_init skips the initialization
+    # of tokenizer and detokenizer. The generated output is expected to contain
+    # token ids.
+    llm = LLM(model=model, skip_tokenizer_init=True)
+    sampling_params = SamplingParams(prompt_logprobs=True, detokenize=True)
+    with pytest.raises(ValueError) as err:
+        llm.generate("abc", sampling_params)
+    assert "prompts must be None if" in str(err.value)
+    outputs = llm.generate(prompt_token_ids=[[1, 2, 3]],
+                           sampling_params=sampling_params)
+    assert len(outputs) > 0
+    completions = outputs[0].outputs
+    assert len(completions) > 0
+    assert completions[0].text == ""
+    assert completions[0].token_ids

vllm/config.py

@@ -66,6 +66,8 @@ class ModelConfig:
         max_context_len_to_capture: Maximum context len covered by CUDA graphs.
             When a sequence has context length larger than this, we fall back
             to eager mode.
+        skip_tokenizer_init: If true, skip initialization of tokenizer and
+            detokenizer.
     """
 
     def __init__(
@@ -85,6 +87,7 @@ class ModelConfig:
         enforce_eager: bool = False,
         max_context_len_to_capture: Optional[int] = None,
         max_logprobs: int = 5,
+        skip_tokenizer_init: bool = False,
     ) -> None:
         self.model = model
         self.tokenizer = tokenizer
@@ -99,6 +102,7 @@ class ModelConfig:
         self.enforce_eager = enforce_eager
         self.max_context_len_to_capture = max_context_len_to_capture
         self.max_logprobs = max_logprobs
+        self.skip_tokenizer_init = skip_tokenizer_init
 
         self.hf_config = get_config(self.model, trust_remote_code, revision,
                                     code_revision)
@@ -106,7 +110,8 @@ class ModelConfig:
         self.dtype = _get_and_verify_dtype(self.hf_text_config, dtype)
         self.max_model_len = _get_and_verify_max_len(self.hf_text_config,
                                                      max_model_len)
-        self._verify_tokenizer_mode()
+        if not self.skip_tokenizer_init:
+            self._verify_tokenizer_mode()
         self._verify_quantization()
         self._verify_cuda_graph()
 

vllm/engine/arg_utils.py

@@ -16,6 +16,7 @@ class EngineArgs:
     """Arguments for vLLM engine."""
     model: str
     tokenizer: Optional[str] = None
+    skip_tokenizer_init: bool = False
     tokenizer_mode: str = 'auto'
     trust_remote_code: bool = False
     download_dir: Optional[str] = None
@@ -93,6 +94,10 @@ class EngineArgs:
             type=str,
             default=EngineArgs.tokenizer,
             help='Name or path of the huggingface tokenizer to use.')
+        parser.add_argument(
+            '--skip-tokenizer-init',
+            action='store_true',
+            help='Skip initialization of tokenizer and detokenizer')
         parser.add_argument(
             '--revision',
             type=str,
@@ -453,7 +458,7 @@ class EngineArgs:
             self.code_revision, self.tokenizer_revision, self.max_model_len,
             self.quantization, self.quantization_param_path,
             self.enforce_eager, self.max_context_len_to_capture,
-            self.max_logprobs)
+            self.max_logprobs, self.skip_tokenizer_init)
         cache_config = CacheConfig(self.block_size,
                                    self.gpu_memory_utilization,
                                    self.swap_space, self.kv_cache_dtype,

vllm/engine/llm_engine.py

@@ -100,6 +100,7 @@ class LLMEngine:
             f"model={model_config.model!r}, "
             f"speculative_config={speculative_config!r}, "
             f"tokenizer={model_config.tokenizer!r}, "
+            f"skip_tokenizer_init={model_config.skip_tokenizer_init}, "
             f"tokenizer_mode={model_config.tokenizer_mode}, "
             f"revision={model_config.revision}, "
             f"tokenizer_revision={model_config.tokenizer_revision}, "
@@ -132,8 +133,14 @@ class LLMEngine:
         self.decoding_config = decoding_config or DecodingConfig()
         self.log_stats = log_stats
 
-        self._init_tokenizer()
-        self.detokenizer = Detokenizer(self.tokenizer)
+        if not self.model_config.skip_tokenizer_init:
+            self.tokenizer: BaseTokenizerGroup
+            self._init_tokenizer()
+            self.detokenizer = Detokenizer(self.tokenizer)
+        else:
+            self.detokenizer = None
+            self.tokenizer = None
+
         self.seq_counter = Counter()
         self.generation_config_fields = _load_generation_config_dict(
             model_config)
@@ -187,9 +194,10 @@ class LLMEngine:
                     parallel_config.disable_custom_all_reduce,
                 })
 
-        # Ping the tokenizer to ensure liveness if it runs in a
-        # different process.
-        self.tokenizer.ping()
+        if self.tokenizer:
+            # Ping the tokenizer to ensure liveness if it runs in a
+            # different process.
+            self.tokenizer.ping()
 
         # Create the scheduler.
         # NOTE: the cache_config here have been updated with the numbers of
@@ -296,7 +304,7 @@ class LLMEngine:
             trust_remote_code=self.model_config.trust_remote_code,
             revision=self.model_config.tokenizer_revision)
         init_kwargs.update(tokenizer_init_kwargs)
-        self.tokenizer: BaseTokenizerGroup = get_tokenizer_group(
+        self.tokenizer = get_tokenizer_group(
             self.parallel_config.tokenizer_pool_config, **init_kwargs)
 
     def _verify_args(self) -> None:
@@ -393,8 +401,13 @@ class LLMEngine:
         # Create the sequences.
         block_size = self.cache_config.block_size
         seq_id = next(self.seq_counter)
-        eos_token_id = self.tokenizer.get_lora_tokenizer(
-            lora_request).eos_token_id
+        eos_token_id = None
+        if self.tokenizer:
+            eos_token_id = self.tokenizer.get_lora_tokenizer(
+                lora_request).eos_token_id
+        else:
+            logger.warning("Use None for EOS token id because tokenizer is "
+                           "not initialized")
         seq = Sequence(seq_id, prompt, prompt_token_ids, block_size,
                        eos_token_id, lora_request)
 

vllm/engine/output_processor/single_step.py

@@ -59,7 +59,8 @@ class SingleStepOutputProcessor(SequenceGroupOutputProcessor):
 
         # Process prompt logprobs
         prompt_logprobs = outputs.prompt_logprobs
-        if prompt_logprobs is not None and seq_group.sampling_params.detokenize:
+        if prompt_logprobs is not None and \
+            seq_group.sampling_params.detokenize and self.detokenizer:
             self.detokenizer.decode_prompt_logprobs_inplace(
                 seq_group, prompt_logprobs)
             seq_group.prompt_logprobs = prompt_logprobs
@@ -105,7 +106,7 @@ class SingleStepOutputProcessor(SequenceGroupOutputProcessor):
             child_seqs.append((parent, parent))
 
         for seq, _ in child_seqs:
-            if seq_group.sampling_params.detokenize:
+            if seq_group.sampling_params.detokenize and self.detokenizer:
                 new_char_count = self.detokenizer.decode_sequence_inplace(
                     seq, seq_group.sampling_params)
             else:

vllm/entrypoints/llm.py

@@ -32,6 +32,9 @@ class LLM:
         tokenizer: The name or path of a HuggingFace Transformers tokenizer.
         tokenizer_mode: The tokenizer mode. "auto" will use the fast tokenizer
             if available, and "slow" will always use the slow tokenizer.
+        skip_tokenizer_init: If true, skip initialization of tokenizer and
+            detokenizer. Expect valid prompt_token_ids and None for prompt
+            from the input.
         trust_remote_code: Trust remote code (e.g., from HuggingFace) when
             downloading the model and tokenizer.
         tensor_parallel_size: The number of GPUs to use for distributed
@@ -76,6 +79,7 @@ class LLM:
         model: str,
         tokenizer: Optional[str] = None,
         tokenizer_mode: str = "auto",
+        skip_tokenizer_init: bool = False,
         trust_remote_code: bool = False,
         tensor_parallel_size: int = 1,
         dtype: str = "auto",
@@ -96,6 +100,7 @@ class LLM:
             model=model,
             tokenizer=tokenizer,
             tokenizer_mode=tokenizer_mode,
+            skip_tokenizer_init=skip_tokenizer_init,
             trust_remote_code=trust_remote_code,
             tensor_parallel_size=tensor_parallel_size,
             dtype=dtype,
@@ -160,6 +165,10 @@ class LLM:
         if prompts is None and prompt_token_ids is None:
             raise ValueError("Either prompts or prompt_token_ids must be "
                              "provided.")
+        if self.llm_engine.model_config.skip_tokenizer_init \
+            and prompts is not None:
+            raise ValueError("prompts must be None if skip_tokenizer_init "
+                             "is True")
         if isinstance(prompts, str):
             # Convert a single prompt to a list.
             prompts = [prompts]