Add 4x upscale button for any History image via PiD-Image-Upscaler API

Revert FP8: torchao Float8Tensor is incompatible with ZeroGPU tensor packing (aten.empty_like unimplemented). Keep dynamic duration + xlarge

FP8 weight-only transformer via torchao: fits default large worker (48GB) -> 1x quota instead of 2x

Dynamic GPU duration based on steps x images x resolution (better queue priority, caps hang damage)

Fix ZeroGPU init abort: request xlarge worker + 120s duration (matches official Qwen/Qwen-Image-2512 space config for this model)

Revert: remove RealESRGAN/utils.py

Revert: remove RealESRGAN/rrdbnet_arch.py

Revert: remove RealESRGAN/model.py

Revert: remove RealESRGAN/arch_utils.py

Revert: remove RealESRGAN/__init__.py

Revert: remove mrflow_utils.py

Revert: remove MrFlow/RealESRGAN deps from requirements.txt

Revert: remove MrFlow/RealESRGAN entirely, restore last known-good app.py (hf_transfer fix only)

Make MrFlow opt-in via a checkbox instead of always-on in normal mode

Add vendored RealESRGAN module (fixed for modern huggingface_hub)

Vendor RealESRGAN locally instead of broken pip package

Add mrflow_utils (direct_sigma_schedule) ported from Xingyu-Zheng/MrFlow

Add Real-ESRGAN dependency for MrFlow

Add MrFlow staged sampling for normal mode (12+1 low-res/refine + Real-ESRGAN x2)

Add hf_transfer dependency

Fix runtime error: enable hf_transfer and raise download timeout for large model download

sfgsfgsdfg

adfasdfad

asdfadsf

vae tiling

Update loras.json

Update loras.json

asdfasd

No changes were provided to summarize.

No changes were provided to summarize in a commit message.

ahora se permiten versiones de loras

Update app.py

Update app.py

Update app.py

Update app.py

Cambios varios

feat: Rename "Users" to "Usuarios" in the admin panel's user management section and page title.

feat: Configure EasyAdmin dashboard with menu items for User, Category, and Product entities, and set the default redirect to the User list.

Update app.py

fix: Add GET and POST methods to the login route definition.

Refactor run_lora_multi for multi-LoRA support, MOCK mode, and UI improvements

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py