final fixes

app.py

@@ -256,8 +256,8 @@ def _background_bootstrap():
             import sys
             argv_bak = sys.argv
             try:
-                # Use official splits from nondisjoint/ and disjoint/ folders with default size limit (2000 samples for better early stopping)
-                sys.argv = ["prepare_polyvore.py", "--root", ds_root, "--max_samples", "2000"]
+                # Use official splits from nondisjoint/ and disjoint/ folders with default size limit (500 samples for faster training)
+                sys.argv = ["prepare_polyvore.py", "--root", ds_root, "--max_samples", "500"]
                 prepare_main()
             finally:
                 sys.argv = argv_bak
@@ -271,15 +271,15 @@ def _background_bootstrap():
         if not os.path.exists(resnet_ckpt):
             BOOT_STATUS = "training-resnet"
             subprocess.run([
-                "python", "train_resnet.py", "--data_root", ds_root, "--epochs", "50",
-                "--batch_size", "16", "--lr", "1e-3", "--early_stopping_patience", "10",
+                "python", "train_resnet.py", "--data_root", ds_root, "--epochs", "3",
+                "--batch_size", "8", "--lr", "1e-3", "--early_stopping_patience", "3",
                 "--out", os.path.join(export_dir, "resnet_item_embedder.pth")
             ], check=False)
         if not os.path.exists(vit_ckpt):
             BOOT_STATUS = "training-vit"
             subprocess.run([
-                "python", "train_vit_triplet.py", "--data_root", ds_root, "--epochs", "50",
-                "--batch_size", "16", "--lr", "5e-4", "--early_stopping_patience", "10",
+                "python", "train_vit_triplet.py", "--data_root", ds_root, "--epochs", "3",
+                "--batch_size", "8", "--lr", "5e-4", "--early_stopping_patience", "3",
                 "--export", os.path.join(export_dir, "vit_outfit_model.pth")
             ], check=False)
         service.reload_models()
@@ -681,8 +681,8 @@ def start_training_simple(dataset_size: str, res_epochs: int, vit_epochs: int):
             # Train ResNet first and wait for completion
             log_message += f"\n🚀 Starting ResNet training on {dataset_size} samples...\n"
             resnet_result = subprocess.run([
-                "python", "train_resnet.py", "--data_root", DATASET_ROOT, "--epochs", "50",
-                "--batch_size", "16", "--lr", "1e-3", "--early_stopping_patience", "10",
+                "python", "train_resnet.py", "--data_root", DATASET_ROOT, "--epochs", str(res_epochs),
+                "--batch_size", "8", "--lr", "1e-3", "--early_stopping_patience", "3",
                 "--out", os.path.join(export_dir, "resnet_item_embedder.pth")
             ] + dataset_args, capture_output=True, text=True, check=False)
             
@@ -708,8 +708,8 @@ def start_training_simple(dataset_size: str, res_epochs: int, vit_epochs: int):
             
             log_message += f"\n🚀 Starting ViT training on {dataset_size} samples...\n"
             vit_result = subprocess.run([
-                "python", "train_vit_triplet.py", "--data_root", DATASET_ROOT, "--epochs", "50",
-                "--batch_size", "16", "--lr", "5e-4", "--early_stopping_patience", "10",
+                "python", "train_vit_triplet.py", "--data_root", DATASET_ROOT, "--epochs", str(vit_epochs),
+                "--batch_size", "8", "--lr", "5e-4", "--early_stopping_patience", "3",
                 "--export", os.path.join(export_dir, "vit_outfit_model.pth")
             ] + dataset_args, capture_output=True, text=True, check=False)
             
@@ -785,12 +785,12 @@ with gr.Blocks(fill_height=True, title="Dressify - Advanced Outfit Recommendatio
         
         with gr.Row():
             gr.Markdown("#### 📊 **Current Behavior**")
-            gr.Markdown("• **Bootstrap**: Downloads full dataset (53K outfits) + generates splits with **2000 samples by default**\n• **Training**: Uses 2000 samples (good for early stopping demonstration!)\n• **Apply Button**: Regenerates splits with your selected size limit")
+            gr.Markdown("• **Bootstrap**: Downloads full dataset (53K outfits) + generates splits with **500 samples by default**\n• **Training**: Uses 500 samples (ultra-fast training!)\n• **Apply Button**: Regenerates splits with your selected size limit")
         
         with gr.Row():
             global_dataset_size = gr.Dropdown(
-                choices=["160", "2000", "5000", "10000", "25000", "50000", "full"],
-                value="2000",
+                choices=["160", "500", "2000", "5000", "10000", "25000", "50000", "full"],
+                value="500",
                 label="Global Dataset Size (Affects Prep + Training)"
             )
             gr.Markdown("**160**: Ultra-fast testing (~30 sec prep, ~1-2 min training)\n**2000**: Fast testing (~1-2 min prep, ~2-5 min training)\n**5000**: Fast testing (~2-3 min prep, ~5-10 min training)\n**10000**: Good testing (~3-5 min prep, ~10-20 min training)\n**full**: Production (~5-10 min prep, ~1-4 hours training)")
@@ -798,11 +798,11 @@ with gr.Blocks(fill_height=True, title="Dressify - Advanced Outfit Recommendatio
         with gr.Row():
             # Apply dataset size button
             apply_size_btn = gr.Button("🔄 Apply Dataset Size & Regenerate Splits", variant="primary")
-            size_status = gr.Textbox(label="Dataset Size Status", value="Dataset size: 2000 samples (click Apply to regenerate splits)", interactive=False)
+            size_status = gr.Textbox(label="Dataset Size Status", value="Dataset size: 500 samples (click Apply to regenerate splits)", interactive=False)
             
             # Current dataset info
             gr.Markdown("#### 📊 **Current Dataset Status**")
-            gr.Markdown("• **Full dataset downloaded**: 53,306 outfits (required for system)\n• **Splits generated**: **2000 samples by default** (good for early stopping!)\n• **Training will use**: 2000 samples (good for early stopping demonstration!)\n• **Scale up**: Use Apply button to increase to larger sizes")
+            gr.Markdown("• **Full dataset downloaded**: 53,306 outfits (required for system)\n• **Splits generated**: **500 samples by default** (ultra-fast training!)\n• **Training will use**: 500 samples (ultra-fast training!)\n• **Scale up**: Use Apply button to increase to larger sizes")
         
         def apply_dataset_size(size: str):
             """Apply global dataset size and regenerate splits."""
@@ -855,8 +855,8 @@ with gr.Blocks(fill_height=True, title="Dressify - Advanced Outfit Recommendatio
                 gr.Markdown("#### 📊 Dataset Size Control")
                 gr.Markdown("Start small for testing, increase for production training")
                 dataset_size = gr.Dropdown(
-                    choices=["160", "2000", "5000", "10000", "25000", "50000", "full"],
-                    value="2000",
+                    choices=["160", "500", "2000", "5000", "10000", "25000", "50000", "full"],
+                    value="500",
                     label="Training Dataset Size"
                 )
                 gr.Markdown("**2000**: Quick testing (~2-5 min)\n**5000**: Fast validation (~5-10 min)\n**10000**: Good validation (~10-20 min)\n**25000+**: Production training")
@@ -944,8 +944,8 @@ with gr.Blocks(fill_height=True, title="Dressify - Advanced Outfit Recommendatio
                 
                 # Quick training
                 gr.Markdown("**Quick Training (Basic Parameters)**")
-                epochs_res = gr.Slider(1, 50, value=10, step=1, label="ResNet epochs")
-                epochs_vit = gr.Slider(1, 100, value=20, step=1, label="ViT epochs")
+                epochs_res = gr.Slider(1, 50, value=3, step=1, label="ResNet epochs")
+                epochs_vit = gr.Slider(1, 100, value=3, step=1, label="ViT epochs")
                 start_btn = gr.Button("🚀 Start Quick Training", variant="secondary")
                 
                 # Advanced training
@@ -1048,16 +1048,16 @@ with gr.Blocks(fill_height=True, title="Dressify - Advanced Outfit Recommendatio
                 gr.Markdown("#### 📊 Dataset Size Control")
                 gr.Markdown("Start small for testing, increase for production training")
                 dataset_size = gr.Dropdown(
-                    choices=["160", "2000", "5000", "10000", "25000", "50000", "full"],
-                    value="2000",
+                    choices=["160", "500", "2000", "5000", "10000", "25000", "50000", "full"],
+                    value="500",
                     label="Training Dataset Size"
                 )
                 gr.Markdown("**2000**: Quick testing (~2-5 min)\n**5000**: Fast validation (~5-10 min)\n**10000**: Good validation (~10-20 min)\n**25000+**: Production training")
             
             with gr.Column(scale=1):
                 gr.Markdown("#### ⚙️ Training Parameters")
-                epochs_res = gr.Slider(1, 50, value=10, step=1, label="ResNet epochs")
-                epochs_vit = gr.Slider(1, 100, value=20, step=1, label="ViT epochs")
+                epochs_res = gr.Slider(1, 50, value=3, step=1, label="ResNet epochs")
+                epochs_vit = gr.Slider(1, 100, value=3, step=1, label="ViT epochs")
         
         train_log = gr.Textbox(label="Training Log", lines=10)
         start_btn = gr.Button("Start Training")