[data] optimize qwen3 loss computation (#7923)

2025-08-21 21:22:50 +08:00 · 2025-04-30 16:18:00 +08:00 · 2025-04-30 16:18:00 +08:00 · d8295cd601
commit d8295cd601
parent a8430f4244
11 changed files with 205 additions and 39 deletions
--- a/README.md
+++ b/README.md
@ -5,7 +5,7 @@
 [![GitHub contributors](https://img.shields.io/github/contributors/hiyouga/LLaMA-Factory?color=orange)](https://github.com/hiyouga/LLaMA-Factory/graphs/contributors)
 [![GitHub workflow](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml/badge.svg)](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml)
 [![PyPI](https://img.shields.io/pypi/v/llamafactory)](https://pypi.org/project/llamafactory/)
-[![Citation](https://img.shields.io/badge/citation-429-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
+[![Citation](https://img.shields.io/badge/citation-447-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
 [![GitHub pull request](https://img.shields.io/badge/PRs-welcome-blue)](https://github.com/hiyouga/LLaMA-Factory/pulls)

 [![Twitter](https://img.shields.io/twitter/follow/llamafactory_ai)](https://twitter.com/llamafactory_ai)
@ -246,11 +246,11 @@ Compared to ChatGLM's [P-Tuning](https://github.com/THUDM/ChatGLM2-6B/tree/main/
 | [Command R](https://huggingface.co/CohereForAI)                   | 35B/104B                         | cohere              |
 | [DeepSeek (Code/MoE)](https://huggingface.co/deepseek-ai)         | 7B/16B/67B/236B                  | deepseek            |
 | [DeepSeek 2.5/3](https://huggingface.co/deepseek-ai)              | 236B/671B                        | deepseek3           |
-| [DeepSeek R1 (Distill)](https://huggingface.co/deepseek-ai)       | 1.5B/7B/8B/14B/32B/70B/671B      | deepseek3           |
+| [DeepSeek R1 (Distill)](https://huggingface.co/deepseek-ai)       | 1.5B/7B/8B/14B/32B/70B/671B      | deepseekr1          |
 | [Falcon](https://huggingface.co/tiiuae)                           | 7B/11B/40B/180B                  | falcon              |
 | [Gemma/Gemma 2/CodeGemma](https://huggingface.co/google)          | 2B/7B/9B/27B                     | gemma               |
 | [Gemma 3](https://huggingface.co/google)                          | 1B/4B/12B/27B                    | gemma3/gemma (1B)   |
-| [GLM-4/GLM-4-0414/GLM-Z1](https://huggingface.co/THUDM)           | 9B/32B                           | glm4                |
+| [GLM-4/GLM-4-0414/GLM-Z1](https://huggingface.co/THUDM)           | 9B/32B                           | glm4/glmz1          |
 | [GPT-2](https://huggingface.co/openai-community)                  | 0.1B/0.4B/0.8B/1.5B              | -                   |
 | [Granite 3.0-3.3](https://huggingface.co/ibm-granite)             | 1B/2B/3B/8B                      | granite3            |
 | [Hunyuan](https://huggingface.co/tencent/)                        | 7B                               | hunyuan             |
--- a/README_zh.md
+++ b/README_zh.md
@ -5,7 +5,7 @@
 [![GitHub contributors](https://img.shields.io/github/contributors/hiyouga/LLaMA-Factory?color=orange)](https://github.com/hiyouga/LLaMA-Factory/graphs/contributors)
 [![GitHub workflow](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml/badge.svg)](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml)
 [![PyPI](https://img.shields.io/pypi/v/llamafactory)](https://pypi.org/project/llamafactory/)
-[![Citation](https://img.shields.io/badge/citation-429-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
+[![Citation](https://img.shields.io/badge/citation-447-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
 [![GitHub pull request](https://img.shields.io/badge/PRs-welcome-blue)](https://github.com/hiyouga/LLaMA-Factory/pulls)

 [![Twitter](https://img.shields.io/twitter/follow/llamafactory_ai)](https://twitter.com/llamafactory_ai)
@ -249,11 +249,11 @@ https://github.com/user-attachments/assets/43b700c6-a178-41db-b1f8-8190a5d3fcfc
 | [Command R](https://huggingface.co/CohereForAI)                   | 35B/104B                         | cohere              |
 | [DeepSeek (Code/MoE)](https://huggingface.co/deepseek-ai)         | 7B/16B/67B/236B                  | deepseek            |
 | [DeepSeek 2.5/3](https://huggingface.co/deepseek-ai)              | 236B/671B                        | deepseek3           |
-| [DeepSeek R1 (Distill)](https://huggingface.co/deepseek-ai)       | 1.5B/7B/8B/14B/32B/70B/671B      | deepseek3           |
+| [DeepSeek R1 (Distill)](https://huggingface.co/deepseek-ai)       | 1.5B/7B/8B/14B/32B/70B/671B      | deepseekr1          |
 | [Falcon](https://huggingface.co/tiiuae)                           | 7B/11B/40B/180B                  | falcon              |
 | [Gemma/Gemma 2/CodeGemma](https://huggingface.co/google)          | 2B/7B/9B/27B                     | gemma               |
 | [Gemma 3](https://huggingface.co/google)                          | 1B/4B/12B/27B                    | gemma3/gemma (1B)   |
-| [GLM-4/GLM-4-0414/GLM-Z1](https://huggingface.co/THUDM)           | 9B/32B                           | glm4                |
+| [GLM-4/GLM-4-0414/GLM-Z1](https://huggingface.co/THUDM)           | 9B/32B                           | glm4/glmz1          |
 | [GPT-2](https://huggingface.co/openai-community)                  | 0.1B/0.4B/0.8B/1.5B              | -                   |
 | [Granite 3.0-3.3](https://huggingface.co/ibm-granite)             | 1B/2B/3B/8B                      | granite3            |
 | [Hunyuan](https://huggingface.co/tencent/)                        | 7B                               | hunyuan             |
--- a/src/llamafactory/chat/hf_engine.py
+++ b/src/llamafactory/chat/hf_engine.py
@ -103,9 +103,11 @@ class HuggingfaceEngine(BaseEngine):
        messages = template.mm_plugin.process_messages(
            messages, mm_input_dict["images"], mm_input_dict["videos"], mm_input_dict["audios"], processor
        )
-        paired_messages = messages + [{"role": "assistant", "content": ""}]
+        # add thought words to avoid skipping thinking
+        paired_messages = messages + [{"role": "assistant", "content": template.add_thought("")}]
        system = system or generating_args["default_system"]
-        prompt_ids, _ = template.encode_oneturn(tokenizer, paired_messages, system, tools)
+        enable_thinking = input_kwargs.pop("enable_thinking", True)
+        prompt_ids, _ = template.encode_oneturn(tokenizer, paired_messages, system, tools, enable_thinking)
        prompt_ids, _ = template.mm_plugin.process_token_ids(
            prompt_ids,
            None,
--- a/src/llamafactory/chat/sglang_engine.py
+++ b/src/llamafactory/chat/sglang_engine.py
@ -146,9 +146,11 @@ class SGLangEngine(BaseEngine):
        messages = self.template.mm_plugin.process_messages(
            messages, images or [], videos or [], audios or [], self.processor
        )
-        paired_messages = messages + [{"role": "assistant", "content": ""}]
+        # add thought words to avoid skipping thinking
+        paired_messages = messages + [{"role": "assistant", "content": self.template.add_thought("")}]
        system = system or self.generating_args["default_system"]
-        prompt_ids, _ = self.template.encode_oneturn(self.tokenizer, paired_messages, system, tools)
+        enable_thinking = input_kwargs.pop("enable_thinking", True)
+        prompt_ids, _ = self.template.encode_oneturn(self.tokenizer, paired_messages, system, tools, enable_thinking)
        prompt_length = len(prompt_ids)

        temperature: Optional[float] = input_kwargs.pop("temperature", None)
--- a/src/llamafactory/chat/vllm_engine.py
+++ b/src/llamafactory/chat/vllm_engine.py
@ -123,9 +123,11 @@ class VllmEngine(BaseEngine):
        messages = self.template.mm_plugin.process_messages(
            messages, images or [], videos or [], audios or [], self.processor
        )
-        paired_messages = messages + [{"role": "assistant", "content": ""}]
+        # add thought words to avoid skipping thinking
+        paired_messages = messages + [{"role": "assistant", "content": self.template.add_thought("")}]
        system = system or self.generating_args["default_system"]
-        prompt_ids, _ = self.template.encode_oneturn(self.tokenizer, paired_messages, system, tools)
+        enable_thinking = input_kwargs.pop("enable_thinking", True)
+        prompt_ids, _ = self.template.encode_oneturn(self.tokenizer, paired_messages, system, tools, enable_thinking)
        prompt_length = len(prompt_ids)

        temperature: Optional[float] = input_kwargs.pop("temperature", None)
--- a/src/llamafactory/data/template.py
+++ b/src/llamafactory/data/template.py
@ -13,6 +13,7 @@
 # limitations under the License.

 import re
+from copy import deepcopy
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, Optional, Union

@ -59,9 +60,10 @@ class Template:
        messages: list[dict[str, str]],
        system: Optional[str] = None,
        tools: Optional[str] = None,
+        enable_thinking: bool = True,
    ) -> tuple[list[int], list[int]]:
        r"""Return a single pair of token ids representing prompt and response respectively."""
-        encoded_messages = self._encode(tokenizer, messages, system, tools, remove_thought=True)
+        encoded_messages = self._encode(tokenizer, messages, system, tools)
        prompt_ids = []
        for encoded_ids in encoded_messages[:-1]:
            prompt_ids += encoded_ids
@ -77,7 +79,7 @@ class Template:
        tools: Optional[str] = None,
    ) -> list[tuple[list[int], list[int]]]:
        r"""Return multiple pairs of token ids representing prompts and responses respectively."""
-        encoded_messages = self._encode(tokenizer, messages, system, tools, remove_thought=False)
+        encoded_messages = self._encode(tokenizer, messages, system, tools)
        return [(encoded_messages[i], encoded_messages[i + 1]) for i in range(0, len(encoded_messages), 2)]

    def extract_tool(self, content: str) -> Union[str, list["FunctionCall"]]:
@ -92,6 +94,19 @@ class Template:

        return list(stop_token_ids)

+    def add_thought(self, content: str) -> str:
+        r"""Add empty thought to assistant message."""
+        return f"{self.thought_words[0]}\n\n{self.thought_words[1]}\n\n" + content
+
+    def remove_thought(self, content: str) -> str:
+        r"""Remove thought from assistant message."""
+        pattern = re.compile(f"{re.escape(self.thought_words[0])}(.*?){re.escape(self.thought_words[1])}", re.DOTALL)
+        return re.sub(pattern, "", content).lstrip("\n")
+
+    def get_thought_word_ids(self, tokenizer: "PreTrainedTokenizer") -> list[int]:
+        r"""Get the token ids of thought words."""
+        return tokenizer.encode(f"{self.thought_words[0]}\n\n{self.thought_words[1]}\n\n", add_special_tokens=False)
+
    def _convert_elements_to_ids(self, tokenizer: "PreTrainedTokenizer", elements: "SLOTS") -> list[int]:
        r"""Convert elements to token ids."""
        token_ids = []
@ -111,18 +126,12 @@ class Template:

        return token_ids

-    def _remove_thought(self, content: str) -> str:
-        r"""Remove thought from assistant message."""
-        pattern = re.compile(f"{re.escape(self.thought_words[0])}(.*?){re.escape(self.thought_words[1])}", re.DOTALL)
-        return re.sub(pattern, "", content).lstrip("\n")
-
    def _encode(
        self,
        tokenizer: "PreTrainedTokenizer",
        messages: list[dict[str, str]],
        system: Optional[str],
        tools: Optional[str],
-        remove_thought: bool,
    ) -> list[list[int]]:
        r"""Encode formatted inputs to pairs of token ids.

@ -140,18 +149,14 @@ class Template:
                    tool_text = self.format_tools.apply(content=tools)[0] if tools else ""
                    elements += self.format_system.apply(content=(system + tool_text))

-            content = message["content"]
-            if remove_thought and message["role"] == Role.ASSISTANT and (i != len(messages) - 1):
-                content = self._remove_thought(content)
-
            if message["role"] == Role.USER:
-                elements += self.format_user.apply(content=content, idx=str(i // 2))
+                elements += self.format_user.apply(content=message["content"], idx=str(i // 2))
            elif message["role"] == Role.ASSISTANT:
-                elements += self.format_assistant.apply(content=content)
+                elements += self.format_assistant.apply(content=message["content"])
            elif message["role"] == Role.OBSERVATION:
-                elements += self.format_observation.apply(content=content)
+                elements += self.format_observation.apply(content=message["content"])
            elif message["role"] == Role.FUNCTION:
-                elements += self.format_function.apply(content=content)
+                elements += self.format_function.apply(content=message["content"])
            else:
                raise NotImplementedError("Unexpected role: {}".format(message["role"]))

@ -331,7 +336,6 @@ class Llama2Template(Template):
        messages: list[dict[str, str]],
        system: str,
        tools: str,
-        remove_thought: bool,
    ) -> list[list[int]]:
        system = system or self.default_system
        encoded_messages = []
@ -345,18 +349,14 @@ class Llama2Template(Template):
                    tool_text = self.format_tools.apply(content=tools)[0] if tools else ""
                    system_text = self.format_system.apply(content=(system + tool_text))[0]

-            content = message["content"]
-            if remove_thought and message["role"] == Role.ASSISTANT and (i != len(messages) - 1):
-                content = self._remove_thought(content)
-
            if message["role"] == Role.USER:
-                elements += self.format_user.apply(content=system_text + content)
+                elements += self.format_user.apply(content=system_text + message["content"])
            elif message["role"] == Role.ASSISTANT:
-                elements += self.format_assistant.apply(content=content)
+                elements += self.format_assistant.apply(content=message["content"])
            elif message["role"] == Role.OBSERVATION:
-                elements += self.format_observation.apply(content=content)
+                elements += self.format_observation.apply(content=message["content"])
            elif message["role"] == Role.FUNCTION:
-                elements += self.format_function.apply(content=content)
+                elements += self.format_function.apply(content=message["content"])
            else:
                raise NotImplementedError("Unexpected role: {}".format(message["role"]))

@ -395,6 +395,60 @@ class Llama2Template(Template):
        return jinja_template


+@dataclass
+class ReasoningTemplate(Template):
+    r"""A template that add thought to assistant message."""
+
+    @override
+    def encode_oneturn(
+        self,
+        tokenizer: "PreTrainedTokenizer",
+        messages: list[dict[str, str]],
+        system: Optional[str] = None,
+        tools: Optional[str] = None,
+        enable_thinking: bool = True,
+    ) -> tuple[list[int], list[int]]:
+        messages = deepcopy(messages)
+        for i in range(len(messages)):
+            if messages[i]["role"] == Role.ASSISTANT and (i != len(messages) - 1):
+                messages[i]["content"] = self.remove_thought(messages[i]["content"])
+
+        encoded_messages = self._encode(tokenizer, messages, system, tools)
+        prompt_ids = []
+        for encoded_ids in encoded_messages[:-1]:
+            prompt_ids += encoded_ids
+
+        if not enable_thinking or (
+            messages[-1]["role"] == Role.ASSISTANT
+            and self.thought_words[0] not in messages[-1]["content"]
+            and self.thought_words[1] not in messages[-1]["content"]
+        ):
+            prompt_ids += self.get_thought_word_ids(tokenizer)
+
+        response_ids = encoded_messages[-1]
+        return prompt_ids, response_ids
+
+    @override
+    def encode_multiturn(
+        self,
+        tokenizer: "PreTrainedTokenizer",
+        messages: list[dict[str, str]],
+        system: Optional[str] = None,
+        tools: Optional[str] = None,
+    ) -> list[tuple[list[int], list[int]]]:
+        messages = deepcopy(messages)
+        encoded_messages = self._encode(tokenizer, messages, system, tools)
+        for i in range(len(messages) - 1):
+            if (
+                messages[i + 1]["role"] == Role.ASSISTANT
+                and self.thought_words[0] not in messages[i + 1]["content"]
+                and self.thought_words[1] not in messages[i + 1]["content"]
+            ):
+                encoded_messages[i] += self.get_thought_word_ids(tokenizer)
+
+        return [(encoded_messages[i], encoded_messages[i + 1]) for i in range(0, len(encoded_messages), 2)]
+
+
 TEMPLATES: dict[str, "Template"] = {}


@ -778,6 +832,15 @@ register_template(
 )


+# copied from deepseek3 template
+register_template(
+    name="deepseekr1",
+    format_user=StringFormatter(slots=["<｜User｜>{{content}}<｜Assistant｜>"]),
+    format_prefix=EmptyFormatter(slots=[{"bos_token"}]),
+    template_class=ReasoningTemplate,
+)
+
+
 register_template(
    name="deepseekcoder",
    format_user=StringFormatter(slots=["### Instruction:\n{{content}}\n### Response:"]),
@ -878,6 +941,22 @@ register_template(
 )


+# copied from glm4 template
+register_template(
+    name="glmz1",
+    format_user=StringFormatter(slots=["<|user|>\n{{content}}<|assistant|>"]),
+    format_assistant=StringFormatter(slots=["\n{{content}}"]),
+    format_system=StringFormatter(slots=["<|system|>\n{{content}}"]),
+    format_function=FunctionFormatter(slots=["{{content}}"], tool_format="glm4"),
+    format_observation=StringFormatter(slots=["<|observation|>\n{{content}}<|assistant|>"]),
+    format_tools=ToolFormatter(tool_format="glm4"),
+    format_prefix=EmptyFormatter(slots=["[gMASK]<sop>"]),
+    stop_words=["<|user|>", "<|observation|>"],
+    efficient_eos=True,
+    template_class=ReasoningTemplate,
+)
+
+
 register_template(
    name="granite3",
    format_user=StringFormatter(
@ -1458,6 +1537,7 @@ register_template(
    format_tools=ToolFormatter(tool_format="qwen"),
    stop_words=["<|im_end|>"],
    replace_eos=True,
+    template_class=ReasoningTemplate,
 )


--- a/src/llamafactory/extras/constants.py
+++ b/src/llamafactory/extras/constants.py
@ -533,6 +533,17 @@ register_model_group(
            DownloadSource.DEFAULT: "deepseek-ai/DeepSeek-V3",
            DownloadSource.MODELSCOPE: "deepseek-ai/DeepSeek-V3",
        },
+        "DeepSeek-V3-671B-0324-Chat": {
+            DownloadSource.DEFAULT: "deepseek-ai/DeepSeek-V3-0324",
+            DownloadSource.MODELSCOPE: "deepseek-ai/DeepSeek-V3-0324",
+        },
+    },
+    template="deepseek3",
+)
+
+
+register_model_group(
+    models={
        "DeepSeek-R1-1.5B-Distill": {
            DownloadSource.DEFAULT: "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",
            DownloadSource.MODELSCOPE: "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",
@ -566,7 +577,7 @@ register_model_group(
            DownloadSource.MODELSCOPE: "deepseek-ai/DeepSeek-R1",
        },
    },
-    template="deepseek3",
+    template="deepseekr1",
 )


@ -737,6 +748,13 @@ register_model_group(
            DownloadSource.DEFAULT: "THUDM/GLM-4-32B-0414",
            DownloadSource.MODELSCOPE: "ZhipuAI/GLM-4-32B-0414",
        },
+    },
+    template="glm4",
+)
+
+
+register_model_group(
+    models={
        "GLM-Z1-9B-0414-Chat": {
            DownloadSource.DEFAULT: "THUDM/GLM-Z1-9B-0414",
            DownloadSource.MODELSCOPE: "ZhipuAI/GLM-Z1-9B-0414",
@ -746,7 +764,7 @@ register_model_group(
            DownloadSource.MODELSCOPE: "ZhipuAI/GLM-Z1-32B-0414",
        },
    },
-    template="glm4",
+    template="glmz1",
 )


--- a/src/llamafactory/webui/chatter.py
+++ b/src/llamafactory/webui/chatter.py
@ -191,6 +191,7 @@ class WebChatModel(ChatModel):
        temperature: float,
        skip_special_tokens: bool,
        escape_html: bool,
+        enable_thinking: bool,
    ) -> Generator[tuple[list[dict[str, str]], list[dict[str, str]]], None, None]:
        r"""Generate output text in stream.

@ -210,6 +211,7 @@ class WebChatModel(ChatModel):
            top_p=top_p,
            temperature=temperature,
            skip_special_tokens=skip_special_tokens,
+            enable_thinking=enable_thinking,
        ):
            response += new_text
            if tools:
--- a/src/llamafactory/webui/components/chatbot.py
+++ b/src/llamafactory/webui/components/chatbot.py
@ -79,6 +79,7 @@ def create_chat_box(
                temperature = gr.Slider(minimum=0.01, maximum=1.5, value=0.95, step=0.01)
                skip_special_tokens = gr.Checkbox(value=True)
                escape_html = gr.Checkbox(value=True)
+                enable_thinking = gr.Checkbox(value=True)
                clear_btn = gr.Button()

    tools.input(check_json_schema, inputs=[tools, engine.manager.get_elem_by_id("top.lang")])
@ -103,6 +104,7 @@ def create_chat_box(
            temperature,
            skip_special_tokens,
            escape_html,
+            enable_thinking,
        ],
        [chatbot, messages],
    )
@ -127,6 +129,7 @@ def create_chat_box(
            temperature=temperature,
            skip_special_tokens=skip_special_tokens,
            escape_html=escape_html,
+            enable_thinking=enable_thinking,
            clear_btn=clear_btn,
        ),
    )
--- a/src/llamafactory/webui/locales.py
+++ b/src/llamafactory/webui/locales.py
@ -2468,6 +2468,23 @@ LOCALES = {
            "label": "HTML タグをエスケープ",
        },
    },
+    "enable_thinking": {
+        "en": {
+            "label": "Enable thinking",
+        },
+        "ru": {
+            "label": "Включить мышление",
+        },
+        "zh": {
+            "label": "启用思考",
+        },
+        "ko": {
+            "label": "사고를 활성화하다",
+        },
+        "ja": {
+            "label": "思考を可能にする",
+        },
+    },
    "clear_btn": {
        "en": {
            "value": "Clear history",
--- a/tests/data/test_template.py
+++ b/tests/data/test_template.py
@ -125,6 +125,37 @@ def test_encode_multiturn(use_fast: bool):
    )


+@pytest.mark.parametrize("use_fast", [True, False])
+def test_reasoning_encode_oneturn(use_fast: bool):
+    tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-8B", use_fast=use_fast)
+    template = get_template_and_fix_tokenizer(tokenizer, DataArguments(template="qwen3"))
+    prompt_ids, answer_ids = template.encode_oneturn(tokenizer, MESSAGES)
+    prompt_str = (
+        "<|im_start|>user\nHow are you<|im_end|>\n"
+        "<|im_start|>assistant\nI am fine!<|im_end|>\n"
+        "<|im_start|>user\n你好<|im_end|>\n"
+        "<|im_start|>assistant\n<think>\n\n</think>\n\n"
+    )
+    answer_str = "很高兴认识你！<|im_end|>\n"
+    _check_tokenization(tokenizer, (prompt_ids, answer_ids), (prompt_str, answer_str))
+
+
+@pytest.mark.parametrize("use_fast", [True, False])
+def test_reasoning_encode_multiturn(use_fast: bool):
+    tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-8B", use_fast=use_fast)
+    template = get_template_and_fix_tokenizer(tokenizer, DataArguments(template="qwen3"))
+    encoded_pairs = template.encode_multiturn(tokenizer, MESSAGES)
+    prompt_str_1 = "<|im_start|>user\nHow are you<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
+    answer_str_1 = "I am fine!<|im_end|>\n"
+    prompt_str_2 = "<|im_start|>user\n你好<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
+    answer_str_2 = "很高兴认识你！<|im_end|>\n"
+    _check_tokenization(
+        tokenizer,
+        (encoded_pairs[0][0], encoded_pairs[0][1], encoded_pairs[1][0], encoded_pairs[1][1]),
+        (prompt_str_1, answer_str_1, prompt_str_2, answer_str_2),
+    )
+
+
@pytest.mark.parametrize("use_fast", [True, False])
 def test_jinja_template(use_fast: bool):
    tokenizer = AutoTokenizer.from_pretrained(TINY_LLAMA3, use_fast=use_fast)
@ -227,6 +258,15 @@ def test_qwen2_5_template(use_fast: bool):

@pytest.mark.parametrize("use_fast", [True, False])
 def test_qwen3_template(use_fast: bool):
+    prompt_str = (
+        "<|im_start|>user\nHow are you<|im_end|>\n"
+        "<|im_start|>assistant\nI am fine!<|im_end|>\n"
+        "<|im_start|>user\n你好<|im_end|>\n"
+        "<|im_start|>assistant\n<think>\n\n</think>\n\n"
+    )
+    answer_str = "很高兴认识你！<|im_end|>\n"
+    _check_template("Qwen/Qwen3-8B", "qwen3", prompt_str, answer_str, use_fast)
+
    prompt_str = (
        "<|im_start|>user\nHow are you<|im_end|>\n"
        "<|im_start|>assistant\nI am fine!<|im_end|>\n"