Headers too large

[FlavianDiethelmEPS]

Hi everyone,

I am trying to convert the whisper-large-v3-turbo-swiss-german model to a ggml.bin file for use in Vibe. They provide a sample script for what I'm trying to do, however I get errors during the conversion phase.

My project structure:

project
|- .venv
|- whisper
|- whisper-large-v3-turbo-swiss-german
|- whisper.cpp
|- script.sh

The command that causes the error:

uv run ./whisper.cpp/models/convert-h5-to-ggml.py ./whisper-large-v3-turbo-swiss-german/ ./whisper .

The output I receive:

Traceback (most recent call last):
  File "/workspaces/python/./whisper.cpp/models/convert-h5-to-ggml.py", line 97, in <module>
    model = WhisperForConditionalGeneration.from_pretrained(dir_model)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/python/.venv/lib/python3.11/site-packages/transformers/modeling_utils.py", line 288, in _wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/python/.venv/lib/python3.11/site-packages/transformers/modeling_utils.py", line 5060, in from_pretrained
    with safe_open(checkpoint_files[0], framework="pt") as f:
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
safetensors_rust.SafetensorError: Error while deserializing header: header too large

For context I have experience with code in general, but I'm a total noob when it comes to AI wizardry.

Kind Regards
Flavian