[model] support GLM-OCR SFT (#10183)

src/llamafactory/data/collator.py

@@ -213,6 +213,7 @@ class MultiModalDataCollatorForSeq2Seq(DataCollatorForSeq2Seq):
             and getattr(self.model.config, "model_type", None)
             in [
                 "glm4v",
+                "glm_ocr",
                 "Keye",
                 "qwen2_vl",
                 "qwen2_5_vl",

src/llamafactory/data/template.py

@@ -1061,6 +1061,22 @@ register_template(
 )
 
 
+# copied from glm4 template
+register_template(
+    name="glm_ocr",
+    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,
+    mm_plugin=get_mm_plugin(name="glm4v", image_token="<|image|>", video_token="<|video|>"),
+)
+
+
 # copied from glm4_moe template
 register_template(
     name="glm4_7",