store data

app.py

@@ -34,6 +34,38 @@ nag_pipe = NAGStableDiffusion3Pipeline.from_pretrained(
     torch_dtype=torch.bfloat16,
     token="hf_token",
 )
+
+
+from huggingface_hub import CommitScheduler, InferenceClient
+
+
+IMAGE_DATASET_DIR = Path(f"{uuid4()}")
+IMAGE_DATASET_DIR.mkdir(parents=True, exist_ok=True)
+IMAGE_JSONL_PATH = IMAGE_DATASET_DIR / "metadata.jsonl"
+
+scheduler = CommitScheduler(
+    repo_id="weathon/vsf-log",
+    repo_type="dataset",
+    folder_path=IMAGE_DATASET_DIR,
+    path_in_repo=IMAGE_DATASET_DIR.name,
+)
+from PIL import Image
+import json
+from datetime import datetime
+from pathlib import Path
+from uuid import uuid4
+def save_image(prompt: str, negative_prompt: str, img_vsf: Image, img_nag: Image, parameters: dict) -> None:
+    vsf_image_path = IMAGE_DATASET_DIR / f"{uuid4()}_vsf.png"
+    nag_image_path = IMAGE_DATASET_DIR / f"{uuid4()}_nag.png"
+
+    with scheduler.lock:
+        img_vsf.save(vsf_image_path)
+        img_nag.save(nag_image_path)
+        with IMAGE_JSONL_PATH.open("a") as f:
+            json.dump({"prompt": prompt, "negative_prompt": negative_prompt, "vsf_image_path": str(vsf_image_path), "nag_image_path": str(nag_image_path), "parameters": parameters, "timestamp": datetime.utcnow().isoformat()}, f)
+            f.write("\n")
+
+
 # pipe = pipe.to("cuda")
 # nag_pipe = nag_pipe.to("cuda")
 import os
@@ -69,7 +101,16 @@ def generate_video(positive_prompt, negative_prompt, guidance_scale, bias, step,
     nag_path = f"images/{uuid.uuid4().hex}_nag.png"
     output_nag.save(nag_path)
     print(f"NAG Image saved to {nag_path}")
-
+    save_image(positive_prompt, negative_prompt, output, output_nag, {
+        "guidance_scale": guidance_scale,
+        "bias": bias,
+        "step": step,
+        "seed": seed,
+        "nag_guidance": nag_guidance,
+        "nag_alpha": nag_alpha,
+        "nag_tau": nag_tau,
+        "nag_step": nag_step,
+    })
     return output_path, nag_path
 
 import json
@@ -111,8 +152,21 @@ def load_abstract_prompt():
 #     ]
 #     )
 
+
+import json
+from datetime import datetime
+from pathlib import Path
+from uuid import uuid4
+
+import gradio as gr
+import numpy as np
+from PIL import Image
+
+
 with gr.Blocks(title="Value Sign Flip SD3.5 Demo") as demo:
-    gr.Markdown("# Value Sign Flip SD3.5 Demo \n\n This demo is based on SD3.5-L-Turbo model and uses Value Sign Flip technique to generate videos with different guidance scales and biases. More on [GitHub](https://github.com/weathon/VSF/blob/main/wan.md)\n\nPositive prompt should be at least 1 sentence long or the results will be weird.")
+    gr.Markdown("# Value Sign Flip SD3.5 Demo \n\n This demo is based on SD3.5-L-Turbo model and uses Value Sign Flip technique to generate videos with different guidance scales and biases. More on [GitHub](https://github.com/weathon/VSF/blob/main/wan.md)\n\nPositive prompt should be at least 1 sentence long or the results will be weird. ")
+    gr.Markdown("To help with further research, all generations will be logged anonymously. If you do not wish to participate, please do not use the demo. Please keep prompts safe for work and non-offensive. ") 
+    
     # gr.Markdown("# Value Sign Flip Wan 2.1 Demo \n\n This demo is based on Wan 2.1 T2V model and uses Value Sign Flip technique to generate videos with different guidance scales and biases. More on [GitHub](https://github.com/weathon/VSF/blob/main/wan.md)\n\nPositive prompt should be at least 2 sentence long or the results will be weird.")
 
     with gr.Row():