Help please: RuntimeError: Boolean value of Tensor with more than one value is ambiguous

[xXSuperSnakerXx]

Hi, i have an error

RuntimeError: Boolean value of Tensor with more than one value is ambiguous

I searched information but i can fix it, anyone know how fix this error?

Before this error i execute "Git pull" in cmd for update automatic 1111, after that stable diffusion gives this error, i reinstalled python and automatic 1111 but this error persist.

Anyone can help me?... Thanks...

[thiagocrepaldi]

same here. my repro is

import torch
from diffusers import DiffusionPipeline

device = "cpu"

model_id = "stabilityai/stable-diffusion-xl-base-1.0"
model = DiffusionPipeline.from_pretrained(model_id).to(device)
random_input = torch.randn(1, 4, 256, 256).to(device)
timestep = torch.tensor([1.0]).to(device)
encoder_hidden_states = torch.randn(1, 1, 2048).to(device)
added_cond_kwargs = {
    "text_embeds": torch.randn(1, 2560).to(device),  # Example tensor, adjust shape as needed
    "time_ids": torch.tensor([1]).to(device)  # Replace 'some_value' with the appropriate value or tensor shape for time_ids
}

args = (random_input, timestep, encoder_hidden_states, None, None, None, None, added_cond_kwargs)
image = model(*args)

Error:

(ptca) root@e9b784a5c219:/opt/pytorch# python test_sdxl.py 
/opt/conda/envs/ptca/lib/python3.11/site-packages/transformers/utils/generic.py:441: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  _torch_pytree._register_pytree_node(
[2024-02-20 20:43:25,716] [INFO] [real_accelerator.py:158:get_accelerator] Setting ds_accelerator to cuda (auto detect)
/opt/conda/envs/ptca/lib/python3.11/site-packages/transformers/utils/generic.py:309: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  _torch_pytree._register_pytree_node(
/opt/conda/envs/ptca/lib/python3.11/site-packages/diffusers/utils/outputs.py:63: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  torch.utils._pytree._register_pytree_node(
/opt/conda/envs/ptca/lib/python3.11/site-packages/transformers/utils/generic.py:309: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  _torch_pytree._register_pytree_node(
/opt/conda/envs/ptca/lib/python3.11/site-packages/diffusers/utils/outputs.py:63: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  torch.utils._pytree._register_pytree_node(
Loading pipeline components...: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:01<00:00,  4.48it/s]
Traceback (most recent call last):
  File "/opt/pytorch/test_sdxl.py", line 18, in <module>
    image = fake_model(*fake_args)
            ^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/pytorch/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/envs/ptca/lib/python3.11/site-packages/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py", line 1076, in __call__
    height = height or self.default_sample_size * self.vae_scale_factor
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Boolean value of Tensor with more than one value is ambiguous

[RETXAM]

I have the same problem, when I increase batch size it fails and gives me the same error of "RuntimeError: Boolean value of Tensor with more than one value is ambiguous" not a coder or anything like that so I'd appreciate any simple version of a solution.