SAM3 Segmentation node failing: Input type (torch.cuda.ByteTensor) and weight type (CUDABFloat16Type) should be the same

[ClayShoaf]

The SAM2 Segmentation node works fine.

!!! Exception during processing !!! Input type (torch.cuda.ByteTensor) and weight type (CUDABFloat16Type) should be the same
Traceback (most recent call last):
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/execution.py", line 516, in execute
    output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
                                                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/execution.py", line 330, in get_output_data
    return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/execution.py", line 304, in _async_map_node_over_list
    await process_inputs(input_dict, i)
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/execution.py", line 292, in process_inputs
    result = f(**inputs)
             ^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/AILab_SAM3Segment.py", line 208, in segment
    img_pil, mask_tensor, mask_rgb = self._run_single(
                                     ^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/AILab_SAM3Segment.py", line 171, in _run_single
    state = processor.set_image(img_pil)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 120, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/sam3/model/sam3_image_processor.py", line 59, in set_image
    state["backbone_out"] = self.model.backbone.forward_image(image)
                            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/sam3/model/vl_combiner.py", line 79, in forward_image
    return activation_ckpt_wrapper(self._forward_image_no_act_ckpt)(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/sam3/model/act_ckpt_utils.py", line 86, in act_ckpt_wrapper
    ret = module(*args, **kwargs)
          ^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/sam3/model/vl_combiner.py", line 86, in _forward_image_no_act_ckpt
    sam3_features, sam3_pos, sam2_features, sam2_pos = self.vision_backbone.forward(
                                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/sam3/model/necks.py", line 108, in forward
    xs = self.trunk(tensor_list)
         ^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/sam3/model/vitdet.py", line 814, in forward
    x = self.patch_embed(x)
        ^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/custom_nodes/comfyui-rmbg/sam3/model/vitdet.py", line 333, in forward
    x = self.proj(x)
        ^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/conv.py", line 548, in forward
    return self._conv_forward(input, self.weight, self.bias)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/ssd4TB/docker/comfy/storage/ComfyUI/env/lib/python3.12/site-packages/torch/nn/modules/conv.py", line 543, in _conv_forward
    return F.conv2d(
           ^^^^^^^^^
RuntimeError: Input type (torch.cuda.ByteTensor) and weight type (CUDABFloat16Type) should be the same

[1038lab]

This error is caused by a mismatch between the input tensor type and the model’s weight type, usually due to an incompatible PyTorch or CUDA setup. Please ensure ComfyUI-RMBG and all dependencies are correctly installed and updated for GPU support

[ClayShoaf]

Yea, I used the SAM3 install instructions in a separate environment and it looks like I ended up with an older version of torch. I don't really want to try messing with my ComfyUI torch install. That just seems like a day worth of troubleshooting will follow.