[Bug][AMD] stabilityai/stable-audio-open-1.0 encounter RecursionError: maximum recursion depth exceeded #13274
Description
Describe the bug
Running the diffuser example from the model card https://huggingface.co/stabilityai/stable-audio-open-1.0
import torch
import soundfile as sf
from diffusers import StableAudioPipeline
pipe = StableAudioPipeline.from_pretrained("stabilityai/stable-audio-open-1.0", torch_dtype=torch.float16)
pipe = pipe.to("cuda")
# define the prompts
prompt = "The sound of a hammer hitting a wooden surface."
negative_prompt = "Low quality."
# set the seed for generator
generator = torch.Generator("cuda").manual_seed(0)
# run the generation
audio = pipe(
prompt,
negative_prompt=negative_prompt,
num_inference_steps=200,
audio_end_in_s=10.0,
num_waveforms_per_prompt=3,
generator=generator,
).audios
output = audio[0].T.float().cpu().numpy()
sf.write("hammer.wav", output, pipe.vae.sampling_rate)
on MI300X
throws the following error:
Loading pipeline components...: 33%|___________________________ | 2/6 [00:00<00:01, 2.05it/s]/usr/
local/lib/python3.12/dist-packages/torch/nn/utils/weight_norm.py:144: FutureWarning: `torch.nn.utils.weight_norm` is deprecated in favor of `torch.nn.u
tils.parametrizations.weight_norm`.
WeightNorm.apply(module, name, dim)
Loading pipeline components...: 100%|_______________________________________________________________________________| 6/6 [00:01<00:00, 5.02it/s]
0%| | 0/200 [00:00<?, ?it/s]/usr/
local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py:608: UserWarning: Should have tb<=t1 but got tb=500.00006103515625 and t1=50
0.0.
warnings.warn(f"Should have {tb_name}<=t1 but got {tb_name}={tb} and t1={self._end}.")
100%|___________________________________________________________________________________________________________| 199/200 [00:11<00:00, 18.59it/s]/usr/
local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py:599: UserWarning: Should have ta>=t0 but got ta=0.0 and t0=0.3.
warnings.warn(f"Should have ta>=t0 but got ta={ta} and t0={self._start}.")
100%|___________________________________________________________________________________________________________| 199/200 [00:11<00:00, 17.14it/s]
Traceback (most recent call last):
File "/usr/local/lib/python3.12/dist-packages/torch/utils/_contextlib.py", line 120, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/diffusers/pipelines/stable_audio/pipeline_stable_audio.py", line 736, in __call__
latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/diffusers/schedulers/scheduling_cosine_dpmsolver_multistep.py", line 662, in step
noise = self.noise_sampler(self.sigmas[self.step_index], self.sigmas[self.step_index + 1]).to(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/diffusers/schedulers/scheduling_dpmsolver_sde.py", line 139, in __call__
return self.tree(t0, t1) / (t1 - t0).abs().sqrt()
^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/diffusers/schedulers/scheduling_dpmsolver_sde.py", line 107, in __call__
w = torch.stack([tree(t0, t1) for tree in self.trees]) * (self.sign * sign)
^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 638, in __call__
intervals = self._last_interval._loc(ta, tb)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 275, in _loc
trampoline.trampoline(self._loc_inner(ta, tb, out))
File "/usr/local/lib/python3.12/dist-packages/trampoline/__init__.py", line 40, in trampoline
raise exception
File "/usr/local/lib/python3.12/dist-packages/trampoline/__init__.py", line 26, in trampoline
res = next(stack[-1])
^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 296, in _loc_inner
self._split(tb)
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 328, in _split
self._left_child._split(midway)
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 328, in _split
self._left_child._split(midway)
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 328, in _split
self._left_child._split(midway)
[Previous line repeated 984 more times]
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 323, in _split
self._split_exact(0.5 * (self._end + self._start))
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 341, in _split_exact
self._left_child = _Interval(start=self._start,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/dist-packages/torchsde/_brownian/brownian_interval.py", line 153, in __init__
self._start = top._round(start) # the left hand edge of the interval
^^^^^^^^^^^^^^^^^
RecursionError: maximum recursion depth exceeded
Reproduction
Running the diffuser example from the model card https://huggingface.co/stabilityai/stable-audio-open-1.0
import torch
import soundfile as sf
from diffusers import StableAudioPipeline
pipe = StableAudioPipeline.from_pretrained("stabilityai/stable-audio-open-1.0", torch_dtype=torch.float16)
pipe = pipe.to("cuda")
# define the prompts
prompt = "The sound of a hammer hitting a wooden surface."
negative_prompt = "Low quality."
# set the seed for generator
generator = torch.Generator("cuda").manual_seed(0)
# run the generation
audio = pipe(
prompt,
negative_prompt=negative_prompt,
num_inference_steps=200,
audio_end_in_s=10.0,
num_waveforms_per_prompt=3,
generator=generator,
).audios
output = audio[0].T.float().cpu().numpy()
sf.write("hammer.wav", output, pipe.vae.sampling_rate)
Logs
System Info
Copy-and-paste the text below in your GitHub issue and FILL OUT the two last points.
- 🤗 Diffusers version: 0.37.0
- Platform: Linux-5.15.0-116-generic-x86_64-with-glibc2.35
- Running on Google Colab?: No
- Python version: 3.12.12
- PyTorch version (GPU?): 2.9.1+git8907517 (True)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Huggingface_hub version: 0.36.2
- Transformers version: 4.57.6
- Accelerate version: 1.12.0
- PEFT version: 0.18.1
- Bitsandbytes version: not installed
- Safetensors version: 0.7.0
- xFormers version: not installed
- Accelerator: NA
- Using GPU in script?:
- Using distributed or parallel set-up in script?:
Who can help?
No response