Hi — Qwen Image 2.1 generates normally from text alone, but every generation fails as soon as I add an image reference.
Error:
RuntimeError: expected scalar type Float but found BFloat16
The traceback consistently goes through the Qwen3-VL vision encoder:
transformers/models/qwen3_vl/modeling_qwen3_vl.py
get_image_features → self.visual(pixel_values, grid_thw=...)
...
torch.nn.functional.layer_norm
RuntimeError: expected scalar type Float but found BFloat16
Environment:
- Windows
- Pinokio / Maestro
- NVIDIA GeForce RTX 5090 Laptop GPU, 24 GB VRAM
- Qwen Image 2.1 7B
- Loaded text encoder:
Qwen3-VL-8B-Instruct_int8_convrot.safetensors
Steps to reproduce:
- Select Qwen Image 2.1 7B.
- Generate from a text prompt without references — this succeeds.
- Attach any image as a reference and use Reference Images mode.
- Submit the same generation — it fails during reference-image encoding.
This previously worked, then began failing without any intentional change to the prompt or reference images. Restarting does not help, and it happens with every reference image I try.
It looks like the reference-image path is sending BF16 activations into a Float32 layer in the Qwen3-VL vision encoder, possibly involving the INT8 ConvRot encoder or offloading logic.
AboutMaestroBlizaine/Maestro
Hi — Qwen Image 2.1 generates normally from text alone, but every generation fails as soon as I add an image reference.
Error:
The traceback consistently goes through the Qwen3-VL vision encoder:
Environment:
Qwen3-VL-8B-Instruct_int8_convrot.safetensorsSteps to reproduce:
This previously worked, then began failing without any intentional change to the prompt or reference images. Restarting does not help, and it happens with every reference image I try.
It looks like the reference-image path is sending BF16 activations into a Float32 layer in the Qwen3-VL vision encoder, possibly involving the INT8 ConvRot encoder or offloading logic.