Conversion to GGUF format

[Allgot]

Hello there!
I was really exited to see the introduction of a newer, larger model that natively supports Korean.

I'm currently fine-tuning the model for my own task, but I've run into some trouble while trying to convert it into GGUF format.
I noticed that you successfully converted it to GGUF format, so I was wondering if you could share the script or method you used for the conversion. I keep encountering a ValueError(f"Can not map tensor {name!r}") during the process.

Once again, I really appreciate your hard work on advancing Korean LLMs. I look forward to seeing more of your future work (and hopefully a response to this issue as well😅)

[Allgot]

Update: although I managed to successfully converted it by skipping all the problematic layers, I'm still curious how you handle this :)

(FYR, these are the skipped layers: attention.k_proj.weight, attention.out_proj.weight, attention.q_proj.weight, attention.v_proj.weight, ln_1.weight, ln_2.weight, mlp.c_fc_0.weight, mlp.c_fc_1.weight, mlp.c_proj.weight.)

Update2: things are getting weird.. I noticed that the map_tensor_name() function in convert_hf_to_gguf.py suddenly fails after successfully mapping a few layers with similar names; e.g., it successfully maps transformer.h.2.attn.attention.k_proj.weight, but then fails to map transformer.h.33.attn.attention.k_proj.weight. It seems like a subtle bug or something—myb it's just specific to me?