[data] fix ollama template (#6902)

* fix ollama template

* add meta info

* use half precision

tests/data/test_template.py

@@ -123,6 +123,7 @@ def test_ollama_modelfile():
     tokenizer = AutoTokenizer.from_pretrained(TINY_LLAMA)
     template = get_template_and_fix_tokenizer(tokenizer, DataArguments(template="llama3"))
     assert template.get_ollama_modelfile(tokenizer) == (
+        "# ollama modelfile auto-generated by llamafactory\n\n"
         "FROM .\n\n"
         'TEMPLATE """<|begin_of_text|>'
         "{{ if .System }}<|start_header_id|>system<|end_header_id|>\n\n{{ .System }}<|eot_id|>{{ end }}"