convert : correct gemma 3n conversion

[ngxson]

Address problem reported by @danielhanchen : https://huggingface.co/unsloth/gemma-3n-E4B-it-GGUF/discussions/6#6860763b91da860136961b2b

[ggerganov]

Btw, the conversion of pretrained Gemma 3n models fails like this:

python convert_hf_to_gguf.py google/gemma-3n-E4B/ --outfile models/gemma-3n-e4b/ggml-model-f16.gguf --outtype f16

Traceback (most recent call last):
  File "llama.cpp/convert_hf_to_gguf.py", line 6718, in <module>
    main()
  File "llama.cpp/convert_hf_to_gguf.py", line 6712, in main
    model_instance.write()
  File "llama.cpp/convert_hf_to_gguf.py", line 410, in write
    self.prepare_metadata(vocab_only=False)
  File "llama.cpp/convert_hf_to_gguf.py", line 523, in prepare_metadata
    self.set_vocab()
  File "llama.cpp/convert_hf_to_gguf.py", line 4411, in set_vocab
    with open(self.dir_model / "chat_template.jinja") as f:
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: 'google/gemma-3n-E4B/chat_template.jinja'

[danielhanchen]

Oh hi - this looks fine!
Re the jinja template missing - this is for a pretrained model right - I guess it should be an if statement?

[CISC]

Btw, the conversion of pretrained Gemma 3n models fails like this:

FileNotFoundError: [Errno 2] No such file or directory: 'google/gemma-3n-E4B/chat_template.jinja'

Fixed in #14508