image gen issues!

What happened?

When ever i try to generate a image i get a driver crash. Video generation still works! this is at setting 4 for low ram and vram.
Steps to reproduce

1.
ive tried qwen and krea
Your system (OS / GPU / RAM / VRAM / etc.)

Im running a 9070xt with 32 gb of ram.
Logs / full error output
[Memory at failure, before cleanup] system RAM available 0.11/31.11 GiB; process RSS 25.70 GiB; process private commit 74.84 GiB; PyTorch VRAM allocated 2.35 GiB; reserved 2.37 GiB; CUDA free 13.21/15.92 GiB
Traceback (most recent call last):
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\wgp.py", line 9373, in generate_video
samples = call_with_sticky_interrupt(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\services\job_lifecycle.py", line 176, in call_with_sticky_interrupt
return callable_(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\models\ltx2\ltx2.py", line 2594, in generate
pipeline_output = active_two_stage(
^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch\utils_contextlib.py", line 124, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\models\ltx2\ltx_pipelines\ti2vid_two_stages_ref.py", line 415, in call
contexts = self.text_encoder_cache.encode(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\shared\utils\text_encoder_cache.py", line 80, in encode
encoded_batch = encode_fn(missing_prompts)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\models\ltx2\ltx_pipelines\ti2vid_two_stages_ref.py", line 410, in
encode_text(text_encoder, prompts=prompts),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\models\ltx2\ltx_core\text_encoders\gemma\encoders\base_encoder.py", line 468, in encode_text
result.append(text_encoder.encode_raw(prompt))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\models\ltx2\ltx_core\text_encoders\gemma\encoders\base_encoder.py", line 55, in encode_raw
outputs = language_model(
^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch\nn\modules\module.py", line 1778, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch\nn\modules\module.py", line 1789, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\mmgp\offload.py", line 3359, in check_change_module
return previous_method(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\transformers\utils\generic.py", line 1064, in wrapper
outputs = func(self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\transformers\models\gemma3\modeling_gemma3.py", line 570, in forward
layer_outputs = decoder_layer(
^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\transformers\modeling_layers.py", line 94, in call
return super().call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch\nn\modules\module.py", line 1778, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch\nn\modules\module.py", line 1789, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\transformers\utils\generic.py", line 1023, in wrapped_forward
output = orig_forward(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\mmgp\offload.py", line 3336, in check_load_into_GPU_needed_other
check_load_into_GPU_needed()
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\mmgp\offload.py", line 3325, in check_load_into_GPU_needed
self.gpu_load_blocks(model_id, blocks_name)
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch_dynamo\eval_frame.py", line 1446, in _fn
return fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\mmgp\offload.py", line 2764, in gpu_load_blocks
cpu_to_gpu(self.transfer_stream, self.blocks_of_modules[next_blocks_entry]) #, self.default_stream
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\mmgp\offload.py", line 2715, in cpu_to_gpu
q = p.to("cuda", non_blocking=True)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch\utils_device.py", line 122, in torch_function
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\shared\qtypes\scaled_fp8.py", line 413, in torch_function
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\shared\qtypes\scaled_fp8.py", line 442, in torch_dispatch
out_data = op(t._data, device=device, **(kwargs or {}))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\pinokio\api\Maestro-AMD.git\Maestro\app\env-amd\Lib\site-packages\torch_ops.py", line 1279, in call
return self._op(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^
torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 58.00 MiB. GPU 0 has a total capacity of 15.92 GiB of which 13.21 GiB is free. Of the allocated memory 2.35 GiB is allocated by PyTorch, and 16.68 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://docs.pytorch.org/docs/stable/notes/cuda.html#optimizing-memory-usage-with-pytorch-cuda-alloc-conf)

[ERROR] The generation of the video has encountered an error: it is likely that you have insufficient VRAM and you should therefore reduce the video resolution or its number of frames.

Paste the full log output here