Add GLM-4-MoE tool calling support

docs/source/grpo_trainer.md

@@ -741,6 +741,7 @@ The returned images are automatically injected into the conversation and passed
 Tested with:
 
 - [**Gemma4**](https://huggingface.co/collections/google/gemma-4) — e.g., `google/gemma-4-E2B-it`
+- **GLM-4-MoE** ([4.5](https://huggingface.co/collections/zai-org/glm-45), [4.6](https://huggingface.co/collections/zai-org/glm-46) or [4.7](https://huggingface.co/collections/zai-org/glm-47)) — e.g., `zai-org/GLM-4.7`
 - [**Qwen3**](https://huggingface.co/collections/Qwen/qwen3) — e.g., `Qwen/Qwen3-0.6B`
 - [**Qwen3.5**](https://huggingface.co/collections/Qwen/qwen35) — e.g., `Qwen/Qwen3.5-2B`
 

tests/test_chat_template_utils.py

@@ -114,6 +114,7 @@ def test_clone_with_sequence_classification_model(self):
 @pytest.mark.parametrize(
     "tokenizer_name",
     [
+        pytest.param("trl-internal-testing/tiny-Glm4MoeForCausalLM", id="glm4moe"),
         pytest.param("trl-internal-testing/tiny-Qwen3MoeForSequenceClassification", id="qwen3"),
         pytest.param("trl-internal-testing/tiny-Qwen3_5ForConditionalGeneration", id="qwen35"),
     ],
@@ -464,6 +465,7 @@ def test_behavior_unchanged_generation_prompt_with_enable_thinking_false(self, t
 @pytest.mark.parametrize(
     "tokenizer_name",
     [
+        pytest.param("trl-internal-testing/tiny-Glm4MoeForCausalLM", id="glm4moe"),
         pytest.param("trl-internal-testing/tiny-Qwen3MoeForSequenceClassification", id="qwen3"),
         pytest.param("trl-internal-testing/tiny-Qwen3_5ForConditionalGeneration", id="qwen35"),
         pytest.param(

trl/chat_template_utils.py

@@ -186,11 +186,47 @@ def clone_chat_template(
     },
 }
 
+glm4moe_schema = {
+    "x-regex": r"^(?:\n?<think>\n?(?:(?P<reasoning_content>.*?\S.*?)\n?|[\s]*)</think>\s*)?(?P<content>.*?)(?:\n(?=<tool_call>))?(?=(?:<tool_call>|$))(?P<tool_calls>(?:<tool_call>.+?</tool_call>\s*)+)?$",
+    "type": "object",
+    "properties": {
+        "role": {"const": "assistant"},
+        "content": {"type": "string"},
+        "reasoning_content": {"type": "string"},
+        "tool_calls": {
+            "type": "array",
+            "x-regex-iterator": r"<tool_call>\s*(.+?)\s*</tool_call>",
+            "items": {
+                "type": "object",
+                "properties": {
+                    "type": {"const": "function"},
+                    "function": {
+                        "type": "object",
+                        "properties": {
+                            "name": {"type": "string", "x-regex": r"^(\S+)"},
+                            "arguments": {
+                                "type": "object",
+                                "x-regex-key-value": r"<arg_key>(?P<key>[^<]+)</arg_key>\s*\n<arg_value>(?P<value>.*?)</arg_value>",
+                                "default": {},
+                                "additionalProperties": {
+                                    "x-parser": "json",
+                                    "x-parser-args": {"allow_non_json": True},
+                                },
+                            },
+                        },
+                    },
+                },
+            },
+        },
+    },
+}
+
+glm4moe_chat_template = (_CHAT_TEMPLATES_DIR / "glm4moe.jinja").read_text()
+
 qwen3_chat_template = (_CHAT_TEMPLATES_DIR / "qwen3.jinja").read_text()
 
 qwen3_5_chat_template_2b_and_below = (_CHAT_TEMPLATES_DIR / "qwen3_5_2b_and_below.jinja").read_text()
 
-
 qwen3_5_chat_template_4b_and_above = (_CHAT_TEMPLATES_DIR / "qwen3_5_4b_and_above.jinja").read_text()
 
 
@@ -223,6 +259,9 @@ def add_response_schema(tokenizer: PreTrainedTokenizer) -> PreTrainedTokenizer:
     {'role': 'assistant', 'content': '', 'tool_calls': [{'type': 'function', 'function': {'name': 'multiply', 'arguments': {'a': 3, 'b': 4}}}]}
     ```
     """
+    if tokenizer.chat_template == glm4moe_chat_template:
+        tokenizer.response_schema = glm4moe_schema
+        return tokenizer
     if tokenizer.chat_template == qwen3_chat_template:
         tokenizer.response_schema = qwen3_schema
         return tokenizer

trl/chat_templates/README.md

@@ -11,6 +11,10 @@ Jinja2 chat templates stored here serve two purposes:
 
 Used for identity comparison only.
 
+### `glm4moe.jinja`
+
+Original GLM-4-MoE chat template.
+
 ### `qwen3.jinja`
 
 Original Qwen3 chat template.

trl/chat_templates/glm4moe.jinja

@@ -0,0 +1,103 @@
+[gMASK]<sop>
+{%- if tools -%}
+<|system|>
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{% for tool in tools %}
+{{ tool | tojson(ensure_ascii=False) }}
+{% endfor %}
+</tools>
+
+For each function call, output the function name and arguments within the following XML format:
+<tool_call>{function-name}
+<arg_key>{arg-key-1}</arg_key>
+<arg_value>{arg-value-1}</arg_value>
+<arg_key>{arg-key-2}</arg_key>
+<arg_value>{arg-value-2}</arg_value>
+...
+</tool_call>{%- endif -%}
+{%- macro visible_text(content) -%}
+    {%- if content is string -%}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if item is mapping and item.type == 'text' -%}
+                {{- item.text }}
+            {%- elif item is string -%}
+                {{- item }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{- content }}
+    {%- endif -%}
+{%- endmacro -%}
+{%- set ns = namespace(last_user_index=-1) %}
+{%- for m in messages %}
+    {%- if m.role == 'user' %}
+        {% set ns.last_user_index = loop.index0 -%}
+    {%- endif %}
+{%- endfor %}
+{% for m in messages %}
+{%- if m.role == 'user' -%}<|user|>
+{{ visible_text(m.content) }}
+{{- '/nothink' if (enable_thinking is defined and not enable_thinking and not visible_text(m.content).endswith("/nothink")) else '' -}}
+{%- elif m.role == 'assistant' -%}
+<|assistant|>
+{%- set reasoning_content = '' %}
+{%- set content = visible_text(m.content) %}
+{%- if m.reasoning_content is string %}
+    {%- set reasoning_content = m.reasoning_content %}
+{%- else %}
+    {%- if '</think>' in content %}
+        {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+        {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+    {%- endif %}
+{%- endif %}
+{%- if loop.index0 > ns.last_user_index and reasoning_content -%}
+{{ '\n<think>' + reasoning_content.strip() +  '</think>'}}
+{%- else -%}
+{{ '\n<think></think>' }}
+{%- endif -%}
+{%- if content.strip() -%}
+{{ '\n' + content.strip() }}
+{%- endif -%}
+{% if m.tool_calls %}
+{% for tc in m.tool_calls %}
+{%- if tc.function %}
+    {%- set tc = tc.function %}
+{%- endif %}
+{{ '\n<tool_call>' + tc.name }}
+{% set _args = tc.arguments %}
+{% for k, v in _args.items() %}
+<arg_key>{{ k }}</arg_key>
+<arg_value>{{ v | tojson(ensure_ascii=False) if v is not string else v }}</arg_value>
+{% endfor %}
+</tool_call>{% endfor %}
+{% endif %}
+{%- elif m.role == 'tool' -%}
+{%- if m.content is string -%}
+{%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+    {{- '<|observation|>' }}
+{%- endif %}
+{{- '\n<tool_response>\n' }}
+{{- m.content }}
+{{- '\n</tool_response>' }}
+{%- else -%}
+<|observation|>{% for tr in m.content %}
+
+<tool_response>
+{{ tr.output if tr.output is defined else tr }}
+</tool_response>{% endfor -%}
+{% endif -%}
+{%- elif m.role == 'system' -%}
+<|system|>
+{{ visible_text(m.content) }}
+{%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    <|assistant|>{{- '\n<think></think>' if (enable_thinking is defined and not enable_thinking) else '' -}}
+{%- endif -%}