search

runew0lf / RuinedFooocus

Focus on prompting and generating
GNU General Public License v3.0
378 stars 22 forks source link

ver 1.5.1 Error with sd3_medium_incl_clips_t5xxlfp16.safetensors #148

Open ptits opened 5 months ago

ptits commented 5 months ago 
it runs well with sd3_medium_incl_clips.safetensors

but with sd3_medium_incl_clips_t5xxlfp16.safetensors it gives errors:
python entry_with_update.py --share
You have the latest version
/data/ptits/focus/RuinedDS3/RuinedFooocus/venv/lib/python3.10/site-packages/huggingface_hub/file_download.py:1132: FutureWarning: `resume_download` is deprecated and will be removed in version 1.0.0. Downloads always resume when possible. If you want to force a new download, use `force_download=True`.
  warnings.warn(
Python 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0]
RuinedFooocus version: 1.51.0
Comfy Backend update check complete.
playsound is relying on another python subprocess. Please use `pip install pygobject` if you want playsound to run more efficiently.
Total VRAM 81051 MB, total RAM 1290100 MB
pytorch version: 2.3.1+cu121
WARNING[XFORMERS]: xFormers can't load C++/CUDA extensions. xFormers was built for:
    PyTorch 2.1.1+cu121 with CUDA 1201 (you have 2.3.1+cu121)
    Python  3.10.13 (you have 3.10.12)
  Please reinstall xformers (see https://github.com/facebookresearch/xformers#installing-xformers)
  Memory-efficient attention, SwiGLU, sparse and more won't be available.
  Set XFORMERS_MORE_DETAILS=1 for more details
xformers version: 0.0.23
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA A100-SXM4-80GB : native
VAE dtype: torch.bfloat16
Using pytorch cross attention
Running on local URL:  http://127.0.0.1:7860
Loading base model: sd_xl_base_1.0_0.9vae.safetensors
IMPORTANT: You are using gradio version 3.50.2, however version 4.29.0 is available, please upgrade.
--------
Running on public URL: https://9184c63067f4becfda.gradio.live

This share link expires in 72 hours. For free permanent hosting and GPU upgrades, run `gradio deploy` from Terminal to deploy to Spaces (https://huggingface.co/spaces)
model_type EPS
Using pytorch attention in VAE
Using pytorch attention in VAE
Base model loaded: sd_xl_base_1.0_0.9vae.safetensors
App started successful. Use the app with http://127.0.0.1:7860/ or 127.0.0.1:7860 or https://9184c63067f4becfda.gradio.live
Loading base model: sd3_medium_incl_clips_t5xxlfp16.safetensors
model_type FLOW
Using pytorch attention in VAE
Using pytorch attention in VAE
Base model loaded: sd3_medium_incl_clips_t5xxlfp16.safetensors
LoRAs loaded: []
Requested to load SD3ClipModel_
Loading 1 new model
Time taken: 1.97 seconds Pipeline process
Exception in thread Thread-5 (worker):
Traceback (most recent call last):
  File "/usr/lib/python3.10/threading.py", line 1016, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.10/threading.py", line 953, in run
    self._target(*self._args, **self._kwargs)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/modules/async_worker.py", line 341, in worker
    handler(task)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/modules/async_worker.py", line 57, in handler
    process(gen_data)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/modules/async_worker.py", line 240, in process
    imgs = pipeline.process(
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/venv/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/modules/sdxl_pipeline.py", line 280, in process
    if self.textencode("+", positive_prompt, clip_skip):
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/modules/sdxl_pipeline.py", line 202, in textencode
    self.conditions[id]["cache"] = CLIPTextEncode().encode(
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/repositories/ComfyUI-from-StabilityAI-Official/nodes.py", line 58, in encode
    cond, pooled = clip.encode_from_tokens(tokens, return_pooled=True)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/repositories/ComfyUI-from-StabilityAI-Official/comfy/sd.py", line 142, in encode_from_tokens
    cond, pooled = self.cond_stage_model.encode_token_weights(tokens)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/repositories/ComfyUI-from-StabilityAI-Official/comfy/sd3_clip.py", line 124, in encode_token_weights
    t5_out, t5_pooled = self.t5xxl.encode_token_weights(token_weight_pars_t5)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/repositories/ComfyUI-from-StabilityAI-Official/comfy/sd1_clip.py", line 40, in encode_token_weights
    out, pooled = self.encode(to_encode)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/repositories/ComfyUI-from-StabilityAI-Official/comfy/sd1_clip.py", line 201, in encode
    return self(tokens)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
    return forward_call(*args, **kwargs)
  File "/data/ptits/focus/RuinedDS3/RuinedFooocus/repositories/ComfyUI-from-StabilityAI-Official/comfy/sd1_clip.py", line 186, in forward
    z = outputs[1].float()
AttributeError: 'NoneType' object has no attribute 'float'
ptits commented 5 months ago

I just reproduced it on Windows 10, not only on Ubuntu as above

version 1.5.1, launched with update

========================

Exception in thread Thread-8 (worker): Traceback (most recent call last): File "threading.py", line 1016, in _bootstrap_inner File "threading.py", line 953, in run File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\modules\async_worker.py", line 341, in worker handler(task) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\modules\async_worker.py", line 57, in handler process(gen_data) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\modules\async_worker.py", line 240, in process imgs = pipeline.process( File "f:\RuinedFooocus_win64_1-25-2\python_embeded\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, *kwargs) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\modules\sdxl_pipeline.py", line 280, in process if self.textencode("+", positive_prompt, clip_skip): File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\modules\sdxl_pipeline.py", line 202, in textencode self.conditions[id]["cache"] = CLIPTextEncode().encode( File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\repositories\ComfyUI-from-StabilityAI-Official\nodes.py", line 58, in encode cond, pooled = clip.encode_from_tokens(tokens, return_pooled=True) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\repositories\ComfyUI-from-StabilityAI-Official\comfy\sd.py", line 142, in encode_from_tokens cond, pooled = self.cond_stage_model.encode_token_weights(tokens) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\repositories\ComfyUI-from-StabilityAI-Official\comfy\sd3_clip.py", line 124, in encode_token_weights t5_out, t5_pooled = self.t5xxl.encode_token_weights(token_weight_pars_t5) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\repositories\ComfyUI-from-StabilityAI-Official\comfy\sd1_clip.py", line 40, in encode_token_weights out, pooled = self.encode(to_encode) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\repositories\ComfyUI-from-StabilityAI-Official\comfy\sd1_clip.py", line 201, in encode return self(tokens) File "f:\RuinedFooocus_win64_1-25-2\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl return forward_call(args, **kwargs) File "F:\RuinedFooocus_win64_1-25-2\RuinedFooocus\repositories\ComfyUI-from-StabilityAI-Official\comfy\sd1_clip.py", line 186, in forward z = outputs[1].float() AttributeError: 'NoneType' object has no attribute 'float'

ptits commented 5 months ago

Any news?

runew0lf commented 5 months ago

yeah.. doesnt work :D