mradermacher

/

model_requests

like 196

It's queued!

You can check for progress at http://hf.tst.eu/status.html or regularly check the model
summary page at https://hf.tst.eu/model#Goetia-26B-A4B-v1.4-LazyLora-heresy-GGUF for quants to appear.

I think the quant process went splat in the same (similar?) manner as other recent gemma 4 quant attempts.

Could you please try again without the vision component?

@simonko912

I queued it again so we can see what whent wrong. If just a vision is the issue it is unlikely to get nuked as we usualy try without vision if mmproj extraction fails.

This model unfortinately failed with the following error:

Goetia-26B-A4B-v1.4-LazyLora-heresy     INFO:hf-to-gguf:Exporting model...
Goetia-26B-A4B-v1.4-LazyLora-heresy     Traceback (most recent call last):
Goetia-26B-A4B-v1.4-LazyLora-heresy       File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 308, in <module>
Goetia-26B-A4B-v1.4-LazyLora-heresy         main()
Goetia-26B-A4B-v1.4-LazyLora-heresy       File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 302, in main
Goetia-26B-A4B-v1.4-LazyLora-heresy         model_instance.write()
Goetia-26B-A4B-v1.4-LazyLora-heresy       File "/llmjob/llama.cpp-cuda512/conversion/base.py", line 1035, in write
Goetia-26B-A4B-v1.4-LazyLora-heresy         self.prepare_tensors()
Goetia-26B-A4B-v1.4-LazyLora-heresy       File "/llmjob/llama.cpp-cuda512/conversion/base.py", line 865, in prepare_tensors
Goetia-26B-A4B-v1.4-LazyLora-heresy         for name, data_torch in chain(self.generate_extra_tensors(), self.get_tensors()):
Goetia-26B-A4B-v1.4-LazyLora-heresy       File "/llmjob/llama.cpp-cuda512/conversion/gemma.py", line 711, in generate_extra_tensors
Goetia-26B-A4B-v1.4-LazyLora-heresy         head_dim_full = (self.hparams["global_head_dim"])
Goetia-26B-A4B-v1.4-LazyLora-heresy     KeyError: 'global_head_dim'

It looks like everything came together famously in the end.

Thanks again.