Commit ·
fe18223
1
Parent(s): 114b4d9
new update fix
Browse files
README.md
CHANGED
|
@@ -29,7 +29,7 @@ Minecraftify is a Hugging Face Gradio Space that turns uploaded photos into a fa
|
|
| 29 |
|
| 30 |
* **Space:** Minecraftify
|
| 31 |
* **Demo video:** [YouTube walkthrough](#)
|
| 32 |
-
* **Blog post:** [Project blog](
|
| 33 |
|
| 34 |
## Current Status
|
| 35 |
|
|
@@ -62,7 +62,7 @@ The model is tuned to:
|
|
| 62 |
|
| 63 |
* **Base model:** `black-forest-labs/FLUX.2-klein-4B`
|
| 64 |
* **LoRA adapter:** `AnimeOverlord/flux2-klein-4b-mc-v2`
|
| 65 |
-
* **Dataset:**
|
| 66 |
* **Training script:** `train_dreambooth_lora_flux2_klein_img2img.py`
|
| 67 |
|
| 68 |
## Fine-Tuning Setup
|
|
@@ -219,7 +219,7 @@ cd diffusers/examples/dreambooth && accelerate launch train_dreambooth_lora_flux
|
|
| 219 |
|
| 220 |
### Notes
|
| 221 |
|
| 222 |
-
* The dataset contains **
|
| 223 |
* The training run uses a paired img2img setup with:
|
| 224 |
|
| 225 |
* `source_image` as the conditioning image
|
|
@@ -238,7 +238,7 @@ YOURACCOUNT/flux2-klein-4b-mc
|
|
| 238 |
|
| 239 |
## Blog and Video Links
|
| 240 |
|
| 241 |
-
* **Blog:** [Read the build notes](
|
| 242 |
* **YouTube:** [Watch the walkthrough](#)
|
| 243 |
|
| 244 |
## Link to Notebooks Used
|
|
|
|
| 29 |
|
| 30 |
* **Space:** Minecraftify
|
| 31 |
* **Demo video:** [YouTube walkthrough](#)
|
| 32 |
+
* **Blog post:** [Project blog](https://huggingface.co/blog/build-small-hackathon/minecraftify)
|
| 33 |
|
| 34 |
## Current Status
|
| 35 |
|
|
|
|
| 62 |
|
| 63 |
* **Base model:** `black-forest-labs/FLUX.2-klein-4B`
|
| 64 |
* **LoRA adapter:** `AnimeOverlord/flux2-klein-4b-mc-v2`
|
| 65 |
+
* **Dataset:** 376 image pairs created with Qwen-Edit-25-12
|
| 66 |
* **Training script:** `train_dreambooth_lora_flux2_klein_img2img.py`
|
| 67 |
|
| 68 |
## Fine-Tuning Setup
|
|
|
|
| 219 |
|
| 220 |
### Notes
|
| 221 |
|
| 222 |
+
* The dataset contains **376 images** created with **Qwen-Edit-25-12**.
|
| 223 |
* The training run uses a paired img2img setup with:
|
| 224 |
|
| 225 |
* `source_image` as the conditioning image
|
|
|
|
| 238 |
|
| 239 |
## Blog and Video Links
|
| 240 |
|
| 241 |
+
* **Blog:** [Read the build notes](https://huggingface.co/blog/build-small-hackathon/minecraftify)
|
| 242 |
* **YouTube:** [Watch the walkthrough](#)
|
| 243 |
|
| 244 |
## Link to Notebooks Used
|
app.py
CHANGED
|
@@ -762,9 +762,6 @@ def _load_pipe():
|
|
| 762 |
|
| 763 |
pipe.set_progress_bar_config(disable=False)
|
| 764 |
|
| 765 |
-
shutil.rmtree(LORA_DIR, ignore_errors=True)
|
| 766 |
-
shutil.rmtree(LORA_CACHE_DIR, ignore_errors=True)
|
| 767 |
-
|
| 768 |
_pipe = pipe
|
| 769 |
_pipe_ready = True
|
| 770 |
log_event("Pipeline ready")
|
|
@@ -1007,12 +1004,6 @@ with gr.Blocks(**BLOCKS_KWARGS) as demo:
|
|
| 1007 |
)
|
| 1008 |
|
| 1009 |
with gr.Accordion("⚙️ Advanced Settings", open=False):
|
| 1010 |
-
steps = gr.Slider(
|
| 1011 |
-
minimum=1, maximum=15, value=2, step=1,
|
| 1012 |
-
label="⚡ Inference Steps",
|
| 1013 |
-
info="Higher = better quality, slower generation",
|
| 1014 |
-
)
|
| 1015 |
-
|
| 1016 |
guidance_scale = gr.Slider(
|
| 1017 |
minimum=1.0, maximum=10.0, value=3.0, step=0.5,
|
| 1018 |
label="🧭 Guidance Scale",
|
|
|
|
| 762 |
|
| 763 |
pipe.set_progress_bar_config(disable=False)
|
| 764 |
|
|
|
|
|
|
|
|
|
|
| 765 |
_pipe = pipe
|
| 766 |
_pipe_ready = True
|
| 767 |
log_event("Pipeline ready")
|
|
|
|
| 1004 |
)
|
| 1005 |
|
| 1006 |
with gr.Accordion("⚙️ Advanced Settings", open=False):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1007 |
guidance_scale = gr.Slider(
|
| 1008 |
minimum=1.0, maximum=10.0, value=3.0, step=0.5,
|
| 1009 |
label="🧭 Guidance Scale",
|