[Issue]: Kandinsky 5: mat1 and mat2 must have the same dtype, but got Float and BFloat16 #4459
Description
Issue Description
Is there something should be configured before using Kandinsky 5?
Steps:
clean reinstall of latest dev in docker
download Kandinsky-5.0-T2I
Generate image -> Error
Kandinsky 5: mat1 and mat2 must have the same dtype, but got Float and BFloat16
using default settings.
Version Platform Description
Version: app=sd.next latest=2025-12-10T11:14:13Z hash=886f5770 branch=dev url=https://github.com/vladmandic/sdnext/tree/dev kanvas=main ui=main
Relevant log output
On branch dev
Your branch is up to date with 'origin/dev'.
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: extensions-builtin/sdnext-modernui
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: extensions-builtin/sdnext-modernui (new commits)
Activate python venv: /mnt/python/venv
Launch: /mnt/python/venv/bin/python3
11:48:19-288093 INFO Starting SD.Next
11:48:19-290575 INFO Logger: file="/app/sdnext.log" level=DEBUG
host="0b8869c383cf" size=262103 mode=append
11:48:19-291188 INFO Python: version=3.12.3 platform=Linux
bin="/mnt/python/venv/bin/python3"
venv="/mnt/python/venv"
11:48:19-307142 INFO Version: app=sd.next updated=2025-12-10
commit=886f57708 branch=dev
url=https://github.com/vladmandic/sdnext/tree/dev
kanvas=main ui=dev
11:48:19-677201 TRACE Repository branches: active=dev available=['dev',
'master', 'upstream']
11:48:20-021630 INFO Version: app=sd.next latest=2025-12-10T11:14:13Z
hash=886f5770 branch=dev
11:48:20-023960 INFO Platform: arch=x86_64 cpu=x86_64 system=Linux
release=6.17.0-7-generic python=3.12.3 locale=('C',
'UTF-8') docker=True
11:48:20-024670 DEBUG Packages: prefix=../mnt/python/venv
site=['../mnt/python/venv/lib/python3.12/site-packages'
]
11:48:20-025363 INFO Args: ['-f', '--use-ipex', '--uv', '--listen',
'--insecure', '--api-log', '--log', 'sdnext.log']
11:48:20-025830 DEBUG Setting environment tuning
11:48:20-026224 DEBUG Torch allocator:
"garbage_collection_threshold:0.80,max_split_size_mb:51
2"
11:48:20-027148 INFO Verifying torch installation
11:48:20-027510 DEBUG Torch overrides: cuda=False rocm=False ipex=True
directml=False openvino=False zluda=False
11:48:20-027978 INFO IPEX: Intel OneAPI toolkit detected
11:48:21-321322 INFO Torch detected: gpu="Intel(R) Graphics [0x7d51]"
vram=126026 units=128
11:48:21-353191 INFO Install: verifying requirements
11:48:21-359297 DEBUG Timestamp repository update time: Wed Dec 10 11:14:13
2025
11:48:21-360609 DEBUG Timestamp previous setup time: Wed Dec 10 11:37:08 2025
11:48:21-361526 INFO Extensions: disabled=[]
11:48:21-362411 INFO Extensions: path="extensions-builtin"
enabled=['stable-diffusion-webui-rembg',
'sd-extension-chainner', 'sdnext-kanvas',
'sd-extension-system-info', 'sdnext-modernui']
11:48:21-364136 INFO Extensions: path="/mnt/data/extensions" enabled=[]
11:48:21-365085 DEBUG Timestamp latest extensions time: Wed Dec 10 11:37:08
2025
11:48:21-366012 DEBUG Timestamp: version:1765365253 setup:1765366628
extension:1765366628
11:48:21-366961 INFO Startup: quick launch
11:48:21-367331 INFO Extensions: disabled=[]
11:48:21-367628 INFO Extensions: path="extensions-builtin"
enabled=['stable-diffusion-webui-rembg',
'sd-extension-chainner', 'sdnext-kanvas',
'sd-extension-system-info', 'sdnext-modernui']
11:48:21-368112 INFO Extensions: path="/mnt/data/extensions" enabled=[]
11:48:21-369560 DEBUG Extension preload: {'extensions-builtin': 0.0,
'/mnt/data/extensions': 0.0}
11:48:21-371021 INFO Installer time: total=2.27 torch=1.30 latest=0.73
base=0.15
11:48:21-372543 INFO Command line args: ['-f', '--use-ipex', '--uv',
'--listen', '--insecure', '--api-log', '--log',
'sdnext.log'] f=True uv=True use_ipex=True
insecure=True listen=True log=sdnext.log args=[]
11:48:21-374065 DEBUG Env flags: ['SD_VAE_DEBUG=true', 'SD_DOCKER=true',
'SD_DATADIR=/mnt/data', 'SD_MODELSDIR=/mnt/models']
11:48:21-375121 DEBUG Linker flags: preload="libjemalloc.so.2"
path=":/mnt/python/venv/lib/"
11:48:21-376138 DEBUG Starting module: <module 'webui' from '/app/webui.py'>
11:48:26-586371 DEBUG System: cores=16 affinity=16 threads=16
11:48:26-587954 INFO Torch: torch==2.9.1+xpu torchvision==0.24.1+xpu
11:48:26-588343 INFO Packages: diffusers==0.36.0.dev0 transformers==4.57.3
accelerate==1.12.0 gradio==3.43.2 pydantic==1.10.21
numpy==2.1.2 cv2==4.12.0
11:48:26-898238 DEBUG ONNX: version=1.23.2,
available=['AzureExecutionProvider',
'CPUExecutionProvider']
11:48:26-929977 DEBUG State initialized: id=127234988517584
11:48:26-942010 INFO Device detect: memory=123.0 default=balanced
optimization=highvram
11:48:26-942541 DEBUG Triton: pass=False fn=<module>:has_triton time=0.00
11:48:26-944007 DEBUG Read: file="/mnt/data/config.json" json=14 bytes=749
time=0.000 fn=<module>:load
11:48:26-944819 INFO Engine: backend=Backend.DIFFUSERS compute=ipex
device=xpu attention="Scaled-Dot-Product" mode=no_grad
11:48:26-945656 DEBUG Read: file="html/reference.json" json=145 bytes=68779
time=0.000 fn=_call_with_frames_removed:<module>
11:48:26-946200 DEBUG Torch attention: type="sdpa" kernels=['Flash',
'Memory', 'Math'] overrides=[]
11:48:26-946707 DEBUG Torch attention installed: flashattn=False
sageattention=False
11:48:26-947067 DEBUG Torch attention status: flash=False flash3=False
aiter=False sage=False flex=True npu=False xla=False
xformers=False
11:48:27-570095 DEBUG Triton: pass=False fn=<module>:set_cuda_params
time=0.00
11:48:27-570936 INFO Torch parameters: backend=ipex device=xpu config=Auto
dtype=torch.bfloat16 context=no_grad nohalf=False
nohalfvae=False upcast=False deterministic=False
tunable=[False, False] fp16=pass bf16=pass triton=fail
optimization="Scaled-Dot-Product"
11:48:27-582484 DEBUG Quantization: registered=SDNQ
11:48:27-582922 INFO Device: device=Intel(R) Graphics [0x7d51] n=1 ipex=
driver=1.6.31294+20
11:48:27-605714 TRACE Trace: VAE
11:48:27-652052 DEBUG Entering start sequence
11:48:27-652486 INFO Base path: data="/mnt/data"
11:48:27-652811 INFO Base path: models="/mnt/models"
11:48:27-653416 DEBUG Initializing
11:48:27-672773 INFO Available VAEs: path="/mnt/models/VAE" items=0
11:48:27-673328 INFO Available UNets: path="/mnt/models/UNET" items=0
11:48:27-673764 INFO Available TEs: path="/mnt/models/Text-encoder" items=0
11:48:27-674819 INFO Available Models:
safetensors="/mnt/models/Stable-diffusion":0
diffusers="/mnt/models/Diffusers":1 reference=145
items=1 time=0.00
11:48:27-677658 INFO Available LoRAs: path="/mnt/models/Lora" items=0
folders=2 time=0.00
11:48:27-682070 INFO Available Styles: path="/mnt/models/styles" items=288
time=0.00
11:48:27-706793 INFO Available Detailer: path="/mnt/models/yolo" items=11
downloaded=0
11:48:27-707407 DEBUG Extensions: disabled=[]
11:48:27-707735 INFO Load extensions
11:48:27-853310 DEBUG Extensions init time: total=0.15
11:48:28-204052 DEBUG Read: file="html/upscalers.json" json=4 bytes=2640
time=0.000 fn=__init__:__init__
11:48:28-204964 DEBUG Read:
file="extensions-builtin/sd-extension-chainner/models.j
son" json=25 bytes=2803 time=0.000
fn=__init__:find_scalers
11:48:28-205902 DEBUG Available chaiNNer: path="/mnt/models/chaiNNer"
defined=25 discovered=0 downloaded=0
11:48:28-207685 INFO Available Upscalers: items=76 downloaded=0 user=0
time=0.35 types=['None', 'Resize', 'Latent',
'AsymmetricVAE', 'WanUpscale', 'DCC', 'VIPS',
'ChaiNNer', 'SeedVR', 'AuraSR', 'Diffusion', 'SwinIR',
'RealESRGAN', 'ESRGAN', 'SCUNet']
11:48:28-237869 INFO Networks: type="video" engines=13 models=67 errors=0
time=0.03
11:48:28-239815 INFO Huggingface: transfer=rust parallel=True direct=False
token="None" cache="/mnt/models/huggingface" init
11:48:28-243446 DEBUG Huggingface: cache="/mnt/models/huggingface" size=27302
MB
11:48:28-243948 DEBUG UI start sequence
11:48:28-244293 DEBUG UI image support: kanvas=main
11:48:28-245896 INFO UI locale: name="Auto"
11:48:28-246356 INFO UI theme: type=Modern name="Default" available=35
11:48:28-246906 DEBUG UI theme:
css="extensions-builtin/sdnext-modernui/themes/Default.
css" base="['base.css', 'timesheet.css']" user="None"
11:48:28-248633 DEBUG UI initialize: tab=txt2img
11:48:28-266965 DEBUG Read: file="html/reference.json" json=145 bytes=68779
time=0.000 fn=list_items:list_reference
11:48:28-270154 DEBUG Networks: type="reference" items={'total': 145,
'ready': 1, 'hidden': 0, 'experimental': 0, 'base': 91,
'distilled': 18, 'quantized': 17, 'community': 15,
'cloud': 2}
11:48:28-273095 DEBUG Networks: type="model" items=144 subfolders=8
tab=txt2img folders=['/mnt/models/Stable-diffusion',
'models/Reference', '/mnt/models/Stable-diffusion']
list=0.01 thumb=0.00 desc=0.00 info=0.00 workers=12
11:48:28-274299 DEBUG Networks: type="lora" items=0 subfolders=1 tab=txt2img
folders=['/mnt/models/Lora'] list=0.00 thumb=0.00
desc=0.00 info=0.00 workers=12
11:48:28-279364 DEBUG Networks: type="style" items=288 subfolders=3
tab=txt2img folders=['/mnt/models/styles', 'html']
list=0.00 thumb=0.00 desc=0.00 info=0.00 workers=12
11:48:28-281195 DEBUG Networks: type="wildcards" items=0 subfolders=1
tab=txt2img folders=['/mnt/models/wildcards'] list=0.00
thumb=0.00 desc=0.00 info=0.00 workers=12
11:48:28-282066 DEBUG Networks: type="embedding" items=0 subfolders=1
tab=txt2img folders=['/mnt/models/embeddings']
list=0.00 thumb=0.00 desc=0.00 info=0.00 workers=12
11:48:28-282910 DEBUG Networks: type="vae" items=0 subfolders=1 tab=txt2img
folders=['/mnt/models/VAE'] list=0.00 thumb=0.00
desc=0.00 info=0.00 workers=12
11:48:28-283753 DEBUG Networks: type="history" items=0 subfolders=1
tab=txt2img folders=[] list=0.00 thumb=0.00 desc=0.00
info=0.00 workers=12
11:48:28-398098 DEBUG UI initialize: tab=img2img
11:48:28-506635 DEBUG UI initialize: tab=control models="/mnt/models/control"
11:48:28-901720 DEBUG UI initialize: tab=video
11:48:29-019155 DEBUG UI initialize: tab=process
11:48:29-068342 DEBUG UI initialize: tab=caption
11:48:29-151657 DEBUG UI initialize: tab=models
11:48:29-224390 DEBUG UI initialize: tab=gallery
11:48:29-267627 DEBUG Read: file="/mnt/data/ui-config.json" json=0 bytes=2
time=0.000 fn=__init__:read_from_file
11:48:29-269381 DEBUG UI initialize: tab=settings
11:48:29-726919 DEBUG Settings: sections=23 settings=383/608 quicksettings=1
11:48:29-784302 DEBUG UI initialize: tab=info
11:48:29-807155 DEBUG UI initialize: tab=extensions
11:48:29-811677 INFO Extension list is empty: refresh required
11:48:29-846537 DEBUG Extension list: processed=3 installed=3 enabled=3
disabled=0 visible=3 hidden=0
11:48:30-357351 DEBUG Root paths: ['/app', '/mnt/data', '/mnt/models']
11:48:30-442399 INFO Local URL: http://localhost:7860/
11:48:30-443409 WARNING Public URL: enabled without authentication
11:48:30-444075 WARNING Public URL: enabled with insecure flag
11:48:30-444754 INFO External URL: http://172.18.0.2:7860
11:48:30-605208 INFO Public URL: http://xx.xx.xx.xx:7860
11:48:30-605965 INFO API docs: http://localhost:7860/docs
11:48:30-606329 INFO API redocs: http://localhost:7860/redocs
11:48:30-607421 DEBUG API middleware: [<class
'starlette.middleware.base.BaseHTTPMiddleware'>, <class
'starlette.middleware.gzip.GZipMiddleware'>]
11:48:30-607963 DEBUG API initialize
11:48:30-744379 DEBUG Scripts setup: time=0.272 ['XYZ Grid:0.033', 'IP
Adapters:0.032']
11:48:30-745048 DEBUG Model metadata: file="/mnt/data/metadata.json" no
changes
11:48:30-745421 INFO Model: autoload=True
selected="Diffusers/kandinskylab/Kandinsky-5.0-T2I-Lite
-sft-Diffusers [25da1e82e6]"
11:48:30-746392 DEBUG Model requested: fn=threading.py:run:<lambda>
11:48:30-747140 DEBUG Search model:
name="Diffusers/kandinskylab/Kandinsky-5.0-T2I-Lite-sft
-Diffusers [25da1e82e6]"
matched="/mnt/models/Diffusers/models--kandinskylab--Ka
ndinsky-5.0-T2I-Lite-sft-Diffusers/snapshots/25da1e82e6
adefdbc83c02a71f762287c5e471a4" type=alias
11:48:30-747719 INFO Load model:
select="Diffusers/kandinskylab/Kandinsky-5.0-T2I-Lite-s
ft-Diffusers [25da1e82e6]"
11:48:30-750510 INFO Autodetect model: detect="Kandinsky 5.0" class=None
file="/mnt/models/Diffusers/models--kandinskylab--Kandi
nsky-5.0-T2I-Lite-sft-Diffusers/snapshots/25da1e82e6ade
fdbc83c02a71f762287c5e471a4"
11:48:30-751194 DEBUG Cache clear
11:48:30-752471 DEBUG Load model: type=Kandinsky50
repo="kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers
" config={'low_cpu_mem_usage': True, 'torch_dtype':
torch.bfloat16, 'load_connected_pipeline': True,
'safety_checker': None, 'requires_safety_checker':
False} offload=balanced dtype=torch.bfloat16
args={'torch_dtype': torch.bfloat16}
11:48:30-753448 DEBUG Load model:
transformer="kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Di
ffusers" cls=Kandinsky5Transformer3DModel
subfolder=transformer quant="None" loader=default
args={'torch_dtype': torch.bfloat16}
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; diffusers/0.36.0.dev0; session_id/94ced552e9644c9788717e9c44118d5e; telemetry/off' https://huggingface.co/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/resolve/main/transformer/config.json
Request 99c7dbd3-216c-4331-99aa-7399abc5b323: HEAD https://huggingface.co/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/resolve/main/transformer/config.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; diffusers/0.36.0.dev0; session_id/94ced552e9644c9788717e9c44118d5e; telemetry/off' https://huggingface.co/api/resolve-cache/models/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/25da1e82e6adefdbc83c02a71f762287c5e471a4/transformer%2Fconfig.json
Request cb801469-c25e-40ec-adf9-f89f5b7f12c5: HEAD https://huggingface.co/api/resolve-cache/models/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/25da1e82e6adefdbc83c02a71f762287c5e471a4/transformer%2Fconfig.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; diffusers/0.36.0.dev0; file_type/model; framework/pytorch' https://huggingface.co/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/resolve/25da1e82e6adefdbc83c02a71f762287c5e471a4/transformer/diffusion_pytorch_model.safetensors.index.json
Request 7f8a55be-b7a1-4c62-8687-ac4e073867a2: HEAD https://huggingface.co/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/resolve/25da1e82e6adefdbc83c02a71f762287c5e471a4/transformer/diffusion_pytorch_model.safetensors.index.json (authenticated: False)
11:48:31-547668 DEBUG Load model:
text_encoder="hunyuanvideo-community/HunyuanImage-2.1-D
iffusers" cls=Qwen2_5_VLForConditionalGeneration
quant="None" loader=default shared=True
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; transformers/4.57.3; session_id/9a638f87398540629504663bba1e8e3d; torch/2.9.1+xpu; telemetry/off' https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/text_encoder/config.json
Request 85649972-1ba0-4b8b-842b-3d141d965511: HEAD https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/text_encoder/config.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; transformers/4.57.3; session_id/9a638f87398540629504663bba1e8e3d; torch/2.9.1+xpu; telemetry/off' https://huggingface.co/api/resolve-cache/models/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/7e7b7a177de58591aeaffca0929f4765003d7ced/text_encoder%2Fconfig.json
Request eba7b2be-837d-4527-97dd-aa0091d30bef: HEAD https://huggingface.co/api/resolve-cache/models/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/7e7b7a177de58591aeaffca0929f4765003d7ced/text_encoder%2Fconfig.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; transformers/4.57.3; session_id/9a638f87398540629504663bba1e8e3d; torch/2.9.1+xpu; telemetry/off' https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/text_encoder/config.json
Request ee47dc2b-9055-43bc-bcc8-f44724da9902: HEAD https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/text_encoder/config.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; transformers/4.57.3; session_id/9a638f87398540629504663bba1e8e3d; torch/2.9.1+xpu; telemetry/off' https://huggingface.co/api/resolve-cache/models/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/7e7b7a177de58591aeaffca0929f4765003d7ced/text_encoder%2Fconfig.json
Request de325a42-4bd2-416d-8b10-479f104d5bd2: HEAD https://huggingface.co/api/resolve-cache/models/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/7e7b7a177de58591aeaffca0929f4765003d7ced/text_encoder%2Fconfig.json (authenticated: False)
Progress 13.60it/s ██████████████ 100% 4/4 00:00 00:00 Loading checkpoint shards
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; transformers/4.57.3; session_id/9a638f87398540629504663bba1e8e3d; torch/2.9.1+xpu; telemetry/off' https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/text_encoder/generation_config.json
Request c14548cf-66f7-47fd-b325-19d12d0b7c0b: HEAD https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/text_encoder/generation_config.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; transformers/4.57.3; session_id/9a638f87398540629504663bba1e8e3d; torch/2.9.1+xpu; telemetry/off' https://huggingface.co/api/resolve-cache/models/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/7e7b7a177de58591aeaffca0929f4765003d7ced/text_encoder%2Fgeneration_config.json
Request c4f2800e-378b-48d2-a743-636ef2d9b051: HEAD https://huggingface.co/api/resolve-cache/models/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/7e7b7a177de58591aeaffca0929f4765003d7ced/text_encoder%2Fgeneration_config.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3; transformers/4.57.3; session_id/9a638f87398540629504663bba1e8e3d; torch/2.9.1+xpu; telemetry/off' https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/custom_generate/generate.py
Request 74fb33f0-eec4-4f03-8eba-1d9b3dde10c2: HEAD https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers/resolve/main/custom_generate/generate.py (authenticated: False)
Send: curl -X GET -H 'Accept: */*' -H 'Accept-Encoding: gzip, deflate' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3' https://huggingface.co/api/models/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers
Request 5ee30564-54f6-46be-83b2-b8bf73a03aef: GET https://huggingface.co/api/models/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3' https://huggingface.co/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/resolve/main/model_index.json
Request 9b9b30c6-88b9-445f-9a1e-80c6679f90d7: HEAD https://huggingface.co/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/resolve/main/model_index.json (authenticated: False)
Send: curl -X HEAD -H 'Accept: */*' -H 'Accept-Encoding: identity' -H 'Connection: keep-alive' -H 'user-agent: unknown/None; hf_hub/0.36.0; python/3.12.3' https://huggingface.co/api/resolve-cache/models/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/25da1e82e6adefdbc83c02a71f762287c5e471a4/model_index.json
Request 8af52673-80a4-4c03-b277-a57fcce95b6b: HEAD https://huggingface.co/api/resolve-cache/models/kandinskylab/Kandinsky-5.0-T2I-Lite-sft-Diffusers/25da1e82e6adefdbc83c02a71f762287c5e471a4/model_index.json (authenticated: False)
Progress 6.44it/s ██████████████ 100% 1/1 00:00 00:00 Loading checkpoint shards
Progress 6.26it/s █████████ 100% 7/7 00:01 00:00 Loading pipeline components...
11:48:34-312733 DEBUG GC: current={'gpu': 0.0, 'ram': 3.34, 'oom': 0}
prev={'gpu': 0.0, 'ram': 3.35} load={'gpu': 0, 'ram':
3} gc={'gpu': 0.0, 'py': 282}
fn=load_diffuser_force:load_kandinsky5 why=load
time=0.36
11:48:34-317306 DEBUG Setting model: component=vae {'slicing': True,
'tiling': False}
11:48:34-319528 DEBUG Setting model: attention="Scaled-Dot-Product"
11:48:34-320944 INFO Offload: type=balanced op=init watermark=0.2-0.8
gpu=24.60-98.40:123.00 cpu=123.000 limit=0.00
always=['T5EncoderModel', 'UMT5EncoderModel']
never=['CLIPTextModel', 'CLIPTextModelWithProjection',
'AutoencoderKL'] pre=True streams=False
11:48:34-336683 DEBUG Module: name=text_encoder
cls=Qwen2_5_VLForConditionalGeneration size=15.445
params=8292166656 quant=None
11:48:34-343565 DEBUG Module: name=transformer
cls=Kandinsky5Transformer3DModel size=11.217
params=6022080576 quant=None
11:48:34-345834 DEBUG Module: name=text_encoder_2 cls=CLIPTextModel
size=0.229 params=123060480 quant=None
11:48:34-348512 DEBUG Module: name=vae cls=AutoencoderKL size=0.156
params=83819683 quant=None
11:48:34-349483 INFO Model class=Kandinsky5T2IPipeline modules=4 size=27.048
11:48:34-353804 INFO Load model: family=kandinsky5 time={'total': 3.61,
'load': 3.57} native=1024 memory={'ram': {'total':
123.07, 'rss': 2.14, 'used': 3.35, 'free': 116.59,
'avail': 119.73, 'buffers': 0.29, 'cached': 3.99},
'gpu': {'used': 0.0, 'total': 123.07, 'active': 0.0,
'peak': 0.0, 'retries': 0, 'oom': 0, 'swap': 0}, 'job':
''}
11:48:34-357326 INFO Startup time: total=19.84 checkpoint=3.61 launch=2.60
gradio=2.53 loader=2.29 installer=2.29 torch=1.85
libraries=1.28 ui-extensions=0.56 ui-defaults=0.44
upscalers=0.36 bnb=0.28 ui-control=0.28 diffusers=0.25
ui-networks=0.21 extensions=0.15 ui-models=0.12
ui-txt2img=0.10
11:48:34-362483 DEBUG Save: file="/mnt/data/config.json" json=14 bytes=749
time=0.006
11:48:35-219706 DEBUG UI: connected
11:48:35-221362 INFO API user=None code=200 http/1.1 GET /sdapi/v1/version
10.9.8.241 0.0005
Traceback (most recent call last):
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/image_utils.py", line 480, in load_image
b64 = base64.decodebytes(image.encode())
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.12/base64.py", line 554, in decodebytes
return binascii.a2b_base64(s)
^^^^^^^^^^^^^^^^^^^^^^
binascii.Error: Invalid base64-encoded string: number of data characters (25) cannot be 1 more than a multiple of 4
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/mnt/python/venv/lib/python3.12/site-packages/gradio/queueing.py", line 388, in call_prediction
output = await route_utils.call_process_api(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/gradio/route_utils.py", line 219, in call_process_api
output = await app.get_blocks().process_api(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/gradio/blocks.py", line 1437, in process_api
result = await self.call_function(
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/gradio/blocks.py", line 1109, in call_function
prediction = await anyio.to_thread.run_sync(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/anyio/to_thread.py", line 61, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/anyio/_backends/_asyncio.py", line 2525, in run_sync_in_worker_thread
return await future
^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/anyio/_backends/_asyncio.py", line 986, in run
result = context.run(func, *args)
^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/gradio/utils.py", line 641, in wrapper
response = f(*args, **kwargs)
^^^^^^^^^^^^^^^^^^
File "/app/modules/call_queue.py", line 13, in f
res = func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/app/modules/ui_common.py", line 424, in update_token_counter
ids = shared.sd_model.tokenizer(prompt)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/models/qwen2_vl/processing_qwen2_vl.py", line 144, in __call__
image_inputs = self.image_processor(images=images, **output_kwargs["images_kwargs"])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/image_processing_utils_fast.py", line 732, in __call__
return self.preprocess(images, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/models/qwen2_vl/image_processing_qwen2_vl_fast.py", line 141, in preprocess
return super().preprocess(images, videos, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/image_processing_utils_fast.py", line 757, in preprocess
return self._preprocess_image_like_inputs(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/models/qwen2_vl/image_processing_qwen2_vl_fast.py", line 160, in _preprocess_image_like_inputs
images = self._prepare_image_like_inputs(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/image_processing_utils_fast.py", line 633, in _prepare_image_like_inputs
images = self._prepare_images_structure(images, expected_ndims=expected_ndims)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/image_processing_utils_fast.py", line 564, in _prepare_images_structure
images = self.fetch_images(images)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/image_processing_base.py", line 532, in fetch_images
return load_image(image_url_or_urls)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/python/venv/lib/python3.12/site-packages/transformers/image_utils.py", line 483, in load_image
raise ValueError(
ValueError: Incorrect image source. Must be a valid URL starting with `http://` or `https://`, a valid path to an image file, or a base64 encoded string. Got car drifting on the Red square. Failed with Invalid base64-encoded string: number of data characters (25) cannot be 1 more than a multiple of 4
11:49:31-904197 DEBUG Select input: type=<class 'str'> source=None init=None
mask=None mode=Image time=0.00
11:49:32-961175 DEBUG Sampler: "Default" cls=FlowMatchEulerDiscreteScheduler
config={'num_train_timesteps': 1000, 'shift': 5.0,
'use_dynamic_shifting': False, 'base_shift': 0.5,
'max_shift': 1.15, 'base_image_seq_len': 256,
'max_image_seq_len': 4096, 'invert_sigmas': False,
'shift_terminal': None, 'use_karras_sigmas': False,
'use_exponential_sigmas': False, 'use_beta_sigmas':
False, 'time_shift_type': 'exponential',
'stochastic_sampling': False}
11:49:32-988940 INFO Processing modifiers: apply
11:49:33-013696 INFO Base: pipeline=Kandinsky5T2IPipeline task=TEXT_2_IMAGE
batch=1/1x1 set={'prompt': 1, 'negative_prompt': 1,
'guidance_scale': 6, 'generator': 'xpu:[1849669577]',
'num_inference_steps': 20, 'output_type': 'np',
'width': 1024, 'height': 1024, 'parser': 'fixed'}
11:49:33-041537 DEBUG Encode: prompt="['car drifting on the Red square']"
hijack=True
11:49:46-001339 DEBUG Encode: prompt="['']" hijack=True
Progress ?it/s 0% 0/20 00:08 ? Base
11:49:54-773356 ERROR Processing: step=base args={'prompt': ['car drifting on
the Red square'], 'negative_prompt': [''],
'guidance_scale': 6, 'generator':
[<modules.intel.ipex.hijacks.torch_Generator object at
0x73b7fab70a10>], 'callback_on_step_end': <function
diffusers_callback at 0x73b7fd0fd300>,
'callback_on_step_end_tensor_inputs': ['latents'],
'num_inference_steps': 20, 'output_type': 'np',
'width': 1024, 'height': 1024} mat1 and mat2 must have
the same dtype, but got Float and BFloat16
11:49:54-775736 ERROR Processing: RuntimeError
╭───────────────────── Traceback (most recent call last) ──────────────────────╮
│/app/modules/processing_diffusers.py:180 in process_base │
│ │
│ 179 │ │ │ taskid = shared.state.begin('Inference') │
│❱ 180 │ │ │ output = shared.sd_model(**base_args) │
│ 181 │ │ │ shared.state.end(taskid) │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/utils/_contextlib.py:120 │
│ │
│ 119 │ │ with ctx_factory(): │
│❱ 120 │ │ │ return func(*args, **kwargs) │
│ 121 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/diffusers/pipelines/kandinsky5/ │
│ │
│ 730 │ │ │ │ # Predict noise residual │
│❱ 731 │ │ │ │ pred_velocity = self.transformer( │
│ 732 │ │ │ │ │ hidden_states=latents.to(dtype), │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1775 │
│ │
│ 1774 │ │ else: │
│❱ 1775 │ │ │ return self._call_impl(*args, **kwargs) │
│ 1776 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1786 │
│ │
│ 1785 │ │ │ │ or _global_forward_hooks or _global_forward_pre_hooks) │
│❱ 1786 │ │ │ return forward_call(*args, **kwargs) │
│ 1787 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/accelerate/hooks.py:175 in new_ │
│ │
│ 174 │ │ else: │
│❱ 175 │ │ │ output = module._old_forward(*args, **kwargs) │
│ 176 │ │ return module._hf_hook.post_forward(module, output) │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/diffusers/models/transformers/t │
│ │
│ 628 │ │ text_embed = self.text_embeddings(text_embed) │
│❱ 629 │ │ time_embed = self.time_embeddings(time) │
│ 630 │ │ time_embed = time_embed + self.pooled_text_embeddings(pooled_te │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1775 │
│ │
│ 1774 │ │ else: │
│❱ 1775 │ │ │ return self._call_impl(*args, **kwargs) │
│ 1776 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1786 │
│ │
│ 1785 │ │ │ │ or _global_forward_hooks or _global_forward_pre_hooks) │
│❱ 1786 │ │ │ return forward_call(*args, **kwargs) │
│ 1787 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/amp/autocast_mode.py:44 i │
│ │
│ 43 │ │ with autocast_instance: │
│❱ 44 │ │ │ return func(*args, **kwargs) │
│ 45 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/diffusers/models/transformers/t │
│ │
│ 171 │ │ time_embed = torch.cat([torch.cos(args), torch.sin(args)], dim= │
│❱ 172 │ │ time_embed = self.out_layer(self.activation(self.in_layer(time_ │
│ 173 │ │ return time_embed │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1775 │
│ │
│ 1774 │ │ else: │
│❱ 1775 │ │ │ return self._call_impl(*args, **kwargs) │
│ 1776 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1786 │
│ │
│ 1785 │ │ │ │ or _global_forward_hooks or _global_forward_pre_hooks) │
│❱ 1786 │ │ │ return forward_call(*args, **kwargs) │
│ 1787 │
│ │
│/mnt/python/venv/lib/python3.12/site-packages/torch/nn/modules/linear.py:134 │
│ │
│ 133 │ │ """ │
│❱ 134 │ │ return F.linear(input, self.weight, self.bias) │
│ 135 │
╰──────────────────────────────────────────────────────────────────────────────╯
RuntimeError: mat1 and mat2 must have the same dtype, but got Float and BFloat16
11:49:55-186148 DEBUG Search model:
name="Diffusers/kandinskylab/Kandinsky-5.0-T2I-Lite-sft
-Diffusers [25da1e82e6]"
matched="/mnt/models/Diffusers/models--kandinskylab--Ka
ndinsky-5.0-T2I-Lite-sft-Diffusers/snapshots/25da1e82e6
adefdbc83c02a71f762287c5e471a4" type=alias
11:49:55-200050 DEBUG Analyzed:
model="Diffusers/kandinskylab/Kandinsky-5.0-T2I-Lite-sf
t-Diffusers" type=kandinsky5
class=Kandinsky5T2IPipeline size=0 mtime="2025-12-10
11:00:51" modules=[name="transformer"
cls=Kandinsky5Transformer3DModel config=True
device=xpu:0 dtype=torch.bfloat16 params=6022080576
modules=1307, name="vae" cls=AutoencoderKL config=True
device=cpu dtype=torch.bfloat16 params=83819683
modules=241, name="text_encoder"
cls=Qwen2_5_VLForConditionalGeneration config=True
device=xpu:0 dtype=torch.bfloat16 params=8292166656
modules=763, name="tokenizer" cls=Qwen2VLProcessor
config=False, name="text_encoder_2" cls=CLIPTextModel
config=True device=xpu:0 dtype=torch.bfloat16
params=123060480 modules=152, name="tokenizer_2"
cls=CLIPTokenizer config=False, name="scheduler"
cls=FlowMatchEulerDiscreteScheduler config=True]
11:49:55-202693 INFO Processing modifiers: unapply
11:49:55-204009 DEBUG Process: batch=1/1 interrupted
11:49:55-205473 INFO Processed: images=0 its=0.00 ops=['txt2img']
11:49:55-206321 DEBUG Processed: timers={'total': 53.79, 'post': 22.2,
'onload': 18.28, 'te': 13.24}
11:49:55-207397 DEBUG Processed: memory={'ram': {'total': 123.07, 'rss':
3.14, 'used': 33.51, 'free': 59.89, 'avail': 89.56,
'buffers': 0.29, 'cached': 58.3}, 'gpu': {'used':
27.76, 'total': 123.07, 'active': 26.96, 'peak': 26.96,
'retries': 0, 'oom': 0, 'swap': 0}, 'job': ''}Backend
Diffusers
Compute
Intel IPEX
Interface
ModernUI
Branch
Dev
Model
Other
Acknowledgements
- I have read the above and searched for existing issues
- I confirm that this is classified correctly and its not an extension issue