What happened?
whatever blizaine added with the Yue2 update, loks like he added new H3 checkpoints, and the wgp.py code is bringing up errors related to tagging of architecture

Steps to reproduce

1. update Maestro AMD, start.

Your system (OS / GPU / RAM / VRAM / etc.)
Ubuntu / Radeon AI Pro 9700 / 32gb vram / 64gb ram / Ryzen 9 5900x

Logs / full error output

Paste the full log output here

████
██ Starting Shell 678f1554-39d4-418e-81bf-62cd15e8f41c
<>eval "$(conda shell.bash hook)" ; conda deactivate ; conda deactivate ; conda deactivate ; conda activate base ; source /home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/env-amd/bin/activate /home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/env-amd && python "/home/mj84/pinokio/api/Maestro-AMD.git/verify_rocm_torch.py" --preflight
(env-amd) (base) <>

██ Terminated Shell 678f1554-39d4-418e-81bf-62cd15e8f41c
████

████
██ Starting Shell 943632e9-97d1-40dc-8e31-e1c968d8490d
<>eval "$(conda shell.bash hook)" ; conda deactivate ; conda deactivate ; conda deactivate ; conda activate base ; source /home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/env-amd/bin/activate /home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/env-amd && python "/home/mj84/pinokio/api/Maestro-AMD.git/configure_amd_defaults.py"
[amd_defaults] AMD defaults already in place
(env-amd) (base) <>

██ Terminated Shell 943632e9-97d1-40dc-8e31-e1c968d8490d
████

████
██ Starting Shell 6e448b35-f646-45b8-84b6-08e6f075abe3
<>eval "$(conda shell.bash hook)" ; conda deactivate ; conda deactivate ; conda deactivate ; conda activate base ; source /home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/env-amd/bin/activate /home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/env-amd && python launch.py
[Maestro] Installing download stall protection...
[safe_download] tqdm download hook installed
[Maestro] Importing WanGP engine...
(null): No such file or directory
[Runtime] FlashAttention is unavailable (No module named 'flash_attn'); continuing with SageAttention/SDPA.
[GGUF][llama.cpp CUDA] kernels unavailable, using fallback
[Runtime] Python 3.11.15 | PyTorch 2.11.0+rocm7.2 | CUDA none | AMD Radeon Graphics (sm_120)
[Runtime] Kernels: Triton=ready (3.8.0), SageAttention=missing, Lightx2v NVFP4=missing, FlashAttention=missing, H3 Sol Engine=ready
[Runtime][ACTION REQUIRED] a CUDA 13 PyTorch build is required for native RTX 50 kernels; the Lightx2v NVFP4 kernel is unavailable; NVFP4 will use a very slow fallback.
[Runtime][ACTION REQUIRED] Run Update in Pinokio. Maestro will create the dedicated RTX 50 Python 3.11 / CUDA 13 environment automatically.
[Runtime][ACTION REQUIRED] If Update already completed, use Advanced > Repair RTX 50 Runtime in Maestro's Pinokio menu.
[GuideLoader] Guide not found: prompt_enhancer/dramabox_speech_rules.md
[GuideLoader] Guide not found: prompt_enhancer/dramabox_dialogue_rules.md
Traceback (most recent call last):
File "/home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/launch.py", line 152, in
import wgp
File "/home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/wgp.py", line 3483, in
load_model_definitions()
File "/home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/wgp.py", line 3452, in load_model_definitions
model_def = init_model_def(model_type, model_def)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/wgp.py", line 3414, in init_model_def

██ Detached from Shell 6e448b35-f646-45b8-84b6-08e6f075abe3

base_model_type = get_base_model_type(model_type)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "/home/mj84/pinokio/api/Maestro-AMD.git/Maestro/app/wgp.py", line 2877, in get_base_model_type
return model_def["architecture"]
~~~~~~~~~^^^^^^^^^^^^^^^^
KeyError: 'architecture'

===================================================

input.event

[
"Error:"
]

██ Terminated Shell 6e448b35-f646-45b8-84b6-08e6f075abe3
████

Disconnected...