Upload weights, notebooks, sample images

configs/config_inference.yaml

@@ -0,0 +1,29 @@
+# weights_path: run "unreflectanything download-weights" then use ~/.cache/unreflectanything/weights/ (or set path below)
+run: "gallant-bush-806"
+runs_dir: "/anvme/workspace/v120bb18-unreflectanything/runs"
+weights_path: "/anvme/workspace/v120bb18-unreflectanything/results/gallant-bush-806/models/full_model_weights.pt"
+
+model_config_path: "/anvme/workspace/v120bb18-unreflectanything/config_train.yaml"
+model_module: "models"
+input_dir: "/anvme/workspace/v120bb18-unreflectanything/benchmark/data/input"
+output_dir: "/anvme/workspace/v120bb18-unreflectanything/benchmark/data/OURS_L1"
+batch_size: 32
+num_workers: 4
+chunk_size: 8
+device: "cuda"
+image_extensions: [".png", ".jpg", ".jpeg"]
+resize_output: True
+brightness_threshold: 0.8
+inpaint_mask_dilation: 11
+monitor_usage: True
+
+# serene-terrain-817   :  SoftTHR ablation - Rebuttal
+# smooth-vortex-816    :  Dice ablation - Rebuttal
+# super-microwave-815  :  TV ablation - Rebuttal
+# magic-brook-814      :  DWConv ablation - Rebuttal
+# faithful-music-813   :  Learned mask token ablation - Rebuttal
+# fresh-fire-811       :  Positional encoding ablation- Rebuttal
+# clean-haze-809       :  Spec ablation - Rebuttal
+# dainty-paper-808     :  Seam ablation - Rebuttal
+# leafy-glade-807      :  RGB ablation - Rebuttal
+# gallant-bush-806     :  L1 ablation - Rebuttal

configs/decoder_pretrain_448.yaml

@@ -0,0 +1,307 @@
+### BASELINE: CONVERGES AFTER LONG
+
+parameters:
+
+  ### MODEL ARCHITECTURE                                                       
+  MODEL:
+    value:
+      MODEL_CLASS: "UnReflect_Model"  # Main model class name (must match class in models.py)
+      MODEL_MODULE: "models"  # Module name to import model classes from (default: "models")
+      RGB_ENCODER:
+        ENCODER: "facebook/dinov3-vitl16-pretrain-lvd1689m"  # DINOv3 encoder model name (HuggingFace format)
+        IMAGE_SIZE: 448  # Input image size (height and width in pixels)
+        RETURN_SELECTED_LAYERS: [3, 6, 9, 12]  # Transformer layer indices to extract features from (0-indexed)
+        RGB_ENCODER_LR: 0.0  # Learning rate for RGB encoder (0.0 = frozen, must be explicitly set)
+      DECODERS:
+        diffuse:
+          USE_FILM: False  # Enable FiLM (Feature-wise Linear Modulation) conditioning in decoder
+          FEATURE_DIM: 1024  # Feature dimension for decoder (should match encoder output)
+          REASSEMBLE_OUT_CHANNELS: [768,1024,1536,2048]  # Output channels for each decoder stage (DPT-style reassembly)
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder ("ignore", "project", etc.)
+          # FROM_PRETRAINED: "weights/rgb_decoder.pth"  # Path to pretrained decoder weights (optional)
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [448,448]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 3  # Number of output channels (3 for RGB diffuse image)
+          DECODER_LR: 1.0e-3 # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: null  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+          TRAIN_RGB_HEAD: True  # Whether to train RGB head (true/false, null = train if DECODER_LR != 0)
+        highlight:
+          USE_FILM: False  # Enable FiLM conditioning in highlight decoder
+          FEATURE_DIM: 1024  # Feature dimension for highlight decoder
+          REASSEMBLE_OUT_CHANNELS: [96,192,384,768]  # Output channels for each decoder stage
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [448,448]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 1  # Number of output channels (1 for highlight mask)
+          DECODER_LR: 5.0e-4  # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: null  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+      TOKEN_INPAINTER:
+        TOKEN_INPAINTER_CLASS: "TokenInpainter_Prior"  # Token inpainter class name
+        TOKEN_INPAINTER_MODULE: "token_inpainters"  # Module name to import token inpainter from
+        FROM_PRETRAINED: "weights/token_inpainter.pth"  # Path to pretrained token inpainter weights (optional)
+        TOKEN_INPAINTER_LR: 1.0e-5  # Learning rate for token inpainter (can differ from base LR)
+        DEPTH: 6  # Number of transformer blocks
+        HEADS: 16  # Number of attention heads
+        DROP: 0 # Dropout rate 
+        USE_POSITIONAL_ENCODING: True  # Enable 2D sinusoidal positional encodings
+        USE_FINAL_NORM: True  # Enable final LayerNorm before output projection
+        USE_LOCAL_PRIOR: True  # Blend local mean prior for masked seeds
+        LOCAL_PRIOR_WEIGHT: 0.5  # Weight for local prior blending (1.0 = only mask_token, 0.0 = only local mean)
+        LOCAL_PRIOR_KERNEL: 5  # Kernel size for local prior blending (> 1)
+        SEED_NOISE_STD: 0.02  # Standard deviation of noise added to masked seeds during training  
+  INPAINT_MASK_DILATION:
+    value: 11  # Dilation kernel size (pixels) for inpaint mask - Must be odd
+  USE_TORCH_COMPILE:  # Enable PyTorch 2.0 torch.compile for faster training (experimental)
+    value: True
+
+  ### DATA                                                                      
+  DATASETS:
+    value:
+      SCRREAM:
+        VAL_SCENES: ["scene10_full_00","scene11_full_00","scene044_full_00","scene04_reduced_00","scene04_reduced_01","scene04_reduced_02"]  # List of validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width] in pixels
+        RESIZE_MODE: "resize+crop"  # Image resizing mode: "resize", "crop", "resize+crop", or "pad"
+        FEW_IMAGES: False  # If True, load only first 10 images per scene (for quick debugging)
+        SAMPLE_EVERY_N: 4  # Load every Nth frame from each scene (1 = all frames, 4 = every 4th frame)
+        LOAD_RGB_ONLY: True  # If True, ignore polarization data and load only RGB images
+        
+      HOUSECAT6D:
+        VAL_SCENES: ["val_scene1","val_scene2"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 4  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      CROMO:
+        TRAIN_SCENES: ["kitchen"]  # Training scene names (list or string)
+        # VAL_SCENES: "station"  # Validation scene names (optional)
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 8  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+    
+      PSD:
+        TRAIN_SCENES: "PSD_Train"  # Training scene name (string or list)
+        VAL_SCENES: "PSD_Val"  # Validation scene name (string or list)
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 1  # Load every Nth frame (1 = all frames)
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      SCARED:
+        VAL_SCENES: ["v22","v23","v24","v25","v26","v27","v28","v29","v30","v31","v32","v33","v34"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 4  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing in dataset
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection (0-1)
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [1000, 1000]  # Size of highlight rectangle region [height, width]
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+      
+      STEREOMIS_TRACKING:
+        VAL_SCENES: ["P2_2"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 10  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      CHOLEC80:
+        VAL_SCENES: ["val"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 50  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      # POLARGB:
+      #   TRAIN_SCENES: "train"
+      #   VAL_SCENES: "test"
+      #   TARGET_SIZE: [448,448]
+      #   RESIZE_MODE: "resize+crop"
+      #   SAMPLE_EVERY_N: 1
+      #   LOAD_RGB_ONLY: True
+
+  BATCH_SIZE: # Max batch size with img size 448 is 32
+    value: 16  # Number of samples per batch (adjust based on GPU memory)
+  NUM_WORKERS:
+    value: 12  # Number of data loading worker processes (0 = main process only)
+  SHUFFLE:
+    value: True  # Shuffle training data each epoch (False for validation/test)
+  PIN_MEMORY:
+    value: True  # Pin memory in DataLoader for faster GPU transfer (recommended: True)
+  PREFETCH_FACTOR:
+    value: 2  # Number of batches to prefetch per worker (higher = more memory usage)
+  
+  ### HIGHLIGHTS                                                        
+  MOGE_MODEL:
+    value: "Ruicheng/moge-2-vits-normal"  # MoGe model name for normal estimation (HuggingFace format)
+  SURFACE_ROUGHNESS:
+    value: 8.0  # Blinn-Phong surface roughness exponent (higher = sharper highlights)
+  INTENSITY:
+    value: 0.0  # Specular highlight intensity multiplier
+  LIGHT_DISTANCE_RANGE:
+    value: [0.0, 1]  # Range for light source distance sampling [min, max] (normalized)
+  LIGHT_LEFT_RIGHT_ANGLE:
+    value: [0, 360]  # Range for light source horizontal angle [min, max] in degrees
+  LIGHT_ABOVE_BELOW_ANGLE:
+    value: [0, 360]  # Range for light source vertical angle [min, max] in degrees
+  DATASET_HIGHLIGHT_DILATION:
+    value: 25  # Dilation kernel size (pixels) for dataset highlight masks
+  DATASET_HIGHLIGHT_THRESHOLD:
+    value: 0.9  # Brightness/luminance threshold (0-1) for detecting highlights in dataset images
+  DATASET_HIGHLIGHT_USE_LUMINANCE:
+    value: True  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B) for dataset highlights; if False, use simple mean brightness
+  HIGHLIGHT_COLOR:
+    value: [1.0, 1.0, 1.0]  # RGB color for synthetic highlights (normalized 0-1)
+  CLAMP_RECONSTRUCTION:
+    value: True  # Clamp reconstructed images to [0, 1] range if True
+
+  ### OPTIMIZATION                                                                      
+  LEARNING_RATE:
+    value: 1.0e-3  # Base learning rate for optimizer
+  WEIGHT_DECAY:   
+    value: 0.0  # L2 regularization weight (0.0 = no weight decay)
+  EPOCHS:
+    value: 50  # Maximum number of training epochs
+  GRADIENT_ACCUMULATION_STEPS:
+    value: 1  # Number of steps to accumulate gradients before optimizer step (1 = no accumulation)
+  WARMUP:
+    value: 100  # Number of warmup steps for learning rate schedule (linear warmup from 0 to LR)
+  GRADIENT_CLIPPING_MAX_NORM:
+    value: 8  # Maximum gradient norm for clipping (set to -1 to disable clipping)
+  LR_SCHEDULER: 
+    value: 
+      ONPLATEAU:  # ReduceLROnPlateau scheduler (reduces LR when validation metric plateaus)
+        PATIENCE: 5  # Number of epochs to wait before reducing LR
+        FACTOR: 0.1  # Factor by which LR is reduced (new_lr = old_lr * factor)
+      # COSINE:  # CosineAnnealingLR scheduler (cosine annealing schedule)
+      #   N_PERIODS: 5  # Number of cosine periods over training
+      STEPWISE:  # StepLR scheduler (reduces LR at fixed step intervals)
+        N_STEPS: 4  # Number of times to reduce LR during training
+        GAMMA: 0.5  # Factor by which LR is reduced at each step (new_lr = old_lr * gamma)
+      # EXPONENTIAL:  # ExponentialLR scheduler (exponential decay)
+      #   GAMMA: 0.5  # Multiplicative factor for exponential decay
+
+  SWITCH_OPTIMIZER_EPOCH:
+    value: null  # Epoch number to switch from bootstrap to refining optimizer (null = no switch)
+  OPTIMIZER_BOOTSTRAP_NAME:
+    value: "AdamW"  # Optimizer name for initial training phase ("Adam", "SGD", etc.)
+  OPTIMIZER_REFINING_NAME:
+    value: "AdamW"  # Optimizer name for refining phase (used after SWITCH_OPTIMIZER_EPOCH)
+  EARLY_STOPPING_PATIENCE:
+    value: 20  # Number of epochs without improvement before stopping training
+  SAVE_INTERVAL:
+    value: 1000  # Number of training steps between model checkpoints
+  
+  DATASET_HIGHLIGHT_SUPERVISION_THRESHOLD:
+    value: 0.1  # Pixel highlights above this threshold (should be low) are excluded from supervision
+
+  ### LOSS WEIGHTS (relative to the total loss, NOT NORMALIZED LATER)                                                                              
+  SPECULAR_LOSS_WEIGHT:
+    value: 1.0  # Weight for specular component reconstruction loss
+  DIFFUSE_LOSS_WEIGHT:
+    value: 1.0  # Weight for diffuse component reconstruction loss
+  HIGHLIGHT_LOSS_WEIGHT:
+    value: 1.0  # Weight for highlight mask regression loss
+  IMAGE_RECONSTRUCTION_LOSS_WEIGHT:
+    value: 0.0  # Weight for full image reconstruction loss
+  SATURATION_RING_LOSS_WEIGHT:
+    value: 0.5  # Weight for saturation ring consistency loss (around highlight regions)
+  RING_KERNEL_SIZE:
+    value: 11  # Kernel size (odd number) for saturation ring dilation around highlights
+  RING_VAR_WEIGHT:
+    value: 0.5  # Weight for variance matching in saturation ring loss (vs mean matching)
+  RING_TEXTURE_WEIGHT:
+    value: 0.0  # Weight for texture consistency term in saturation ring loss
+  HLREG_W_L1:
+    value: 1.0  # Weight for L1 loss in highlight regression
+  HLREG_USE_CHARB:
+    value: True  # Use Charbonnier loss (smooth L1) instead of standard L1 if True
+  HLREG_W_DICE:
+    value: 0.2  # Weight for Dice loss in highlight regression (for mask overlap)
+  HLREG_W_SSIM:
+    value: 0.0  # Weight for SSIM loss in highlight regression
+  HLREG_W_GRAD:
+    value: 0.0  # Weight for gradient loss in highlight regression
+  HLREG_W_TV:
+    value: 0.0  # Weight for total variation loss in highlight regression
+  HLREG_BALANCE_MODE:
+    value: "auto"   # Class balancing mode for highlight regression: 'none' | 'auto' | 'pos_weight'
+  HLREG_POS_WEIGHT:
+    value: 1.0      # Positive class weight (used only if BALANCE_MODE == 'pos_weight')
+  HLREG_FOCAL_GAMMA:
+    value: 2.0      # Focal loss gamma parameter (0.0 = standard BCE, 1.0-2.0 helps with gradient vanishing)
+  
+  WEIGHT_TOKEN_INPAINT:
+    value: 0.0  # Weight for token-space inpainting loss (L1 + cosine similarity in feature space)
+  WEIGHT_CONTEXT_IDENTITY: 
+    value: 0.0  # LEAVE TO 0.0: Weight for L1 loss on context (non-masked) regions (identity preservation) 
+  WEIGHT_TV_IN_HOLE:
+    value: 0.0  # LEAVE TO 0.0: Weight for total variation loss inside masked/hole regions
+  RING_DILATE_KERNEL:
+    value: 17  # Dilation kernel size (odd number) for creating ring mask around highlights
+  WEIGHT_SEAM:
+    value: 0.5  # Weight for gradient matching loss on saturation ring
+  SEAM_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 in seam loss (smooth L1 for boundary consistency)
+  SEAM_WEIGHT_GRAD:
+    value: 0.0  # Weight for gradient matching term inside seam loss (0.0 = disable gradient term)
+  TOKEN_FEAT_ALPHA:
+    value: 0.5  # Mixing factor for token feature loss: alpha * L1 + (1-alpha) * (1-cosine_sim)
+
+  ### DIFFUSE HIGHLIGHT PENALTY
+  WEIGHT_DIFFUSE_HIGHLIGHT_PENALTY:
+    value: 0.0  # Weight for penalty loss on highlights in diffuse decoder output (0.0 = disabled)
+  DIFFUSE_HL_THRESHOLD:
+    value: 0.8  # Brightness/luminance threshold for detecting highlights in diffuse (0.0-1.0)
+  DIFFUSE_HL_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 for diffuse highlight penalty
+  DIFFUSE_HL_PENALTY_MODE:
+    value: "brightness"  # Penalty mode: "brightness" (penalize brightness/luminance above threshold) or "pixel" (penalize RGB values directly)
+  DIFFUSE_HL_TARGET_BRIGHTNESS:
+    value: null  # Target brightness/luminance for penalized pixels (null = use threshold value)
+  DIFFUSE_HL_USE_LUMINANCE:
+    value: True  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B); if False, use simple mean brightness
+
+  ### LOGGING, RESULTS AND WANDB                                                                 
+  LOG_INTERVAL:
+    value: 1  # Number of training steps between console log outputs
+  WANDB_LOG_INTERVAL:
+    value: 1  # Number of training steps between WandB metric logs
+  IMAGE_LOG_INTERVAL:
+    value: 5  # Number of training steps between image logging to WandB
+  NO_WANDB:
+    value: False  # Disable WandB logging if True (useful for local debugging)
+  MODEL_WATCHER_FREQ_WANDB:
+    value: 50  # Frequency (in steps) for logging model parameter histograms to WandB
+  WANDB_ENTITY:
+    value: "unreflect-anything"  # WandB organization/entity name
+  WANDB_PROJECT:
+    value: "UnReflectAnything"  # WandB project name
+  NOTES:
+    value: "896 Decoder Pretraining"  # Notes/description for this training run
+  

configs/decoder_pretrain_896.yaml

@@ -0,0 +1,295 @@
+### BASELINE: CONVERGES AFTER LONG
+
+parameters:
+
+  ### MODEL ARCHITECTURE                                                       
+  MODEL:
+    value:
+      MODEL_CLASS: "UnReflect_Model"  # Main model class name (must match class in models.py)  # <<<<<<<<< DECODER PRETRAINING: NOT USING TOKEN INPAINTER (DIRECT FROM DINO)
+      MODEL_MODULE: "models"  # Module name to import model classes from (default: "models")
+      RGB_ENCODER:
+        ENCODER: "facebook/dinov3-vitl16-pretrain-lvd1689m"  # DINOv3 encoder model name (HuggingFace format)
+        IMAGE_SIZE: 896  # Input image size (height and width in pixels)
+        RETURN_SELECTED_LAYERS: [3, 6, 9, 12]  # Transformer layer indices to extract features from (0-indexed)
+        RGB_ENCODER_LR: 0.0  # Learning rate for RGB encoder (0.0 = frozen, must be explicitly set)
+      DECODERS:
+        diffuse:
+          USE_FILM: False  # Enable FiLM (Feature-wise Linear Modulation) conditioning in decoder
+          FEATURE_DIM: 1024  # Feature dimension for decoder (should match encoder output)
+          REASSEMBLE_OUT_CHANNELS: [768,1024,1536,2048]  # Output channels for each decoder stage (DPT-style reassembly)
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder ("ignore", "project", etc.)
+          # FROM_PRETRAINED: "weights/rgb_decoder.pth"  # Path to pretrained decoder weights (optional)    # <<<<<<<<< DECODER PRETRAINING: NO WEIGHTS HERE
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [896,896]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 3  # Number of output channels (3 for RGB diffuse image)
+          DECODER_LR: 5.0e-5 # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: null  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+          TRAIN_RGB_HEAD: True  # Whether to train RGB head (true/false, null = train if DECODER_LR != 0)
+      TOKEN_INPAINTER:
+        TOKEN_INPAINTER_CLASS: "TokenInpainter_Prior"  # Token inpainter class name
+        TOKEN_INPAINTER_MODULE: "token_inpainters"  # Module name to import token inpainter from
+        FROM_PRETRAINED: "weights/token_inpainter.pth"  # Path to pretrained token inpainter weights (optional)
+        TOKEN_INPAINTER_LR: 1.0e-5  # Learning rate for token inpainter (can differ from base LR)
+        DEPTH: 6  # Number of transformer blocks
+        HEADS: 16  # Number of attention heads
+        DROP: 0 # Dropout rate 
+        USE_POSITIONAL_ENCODING: True  # Enable 2D sinusoidal positional encodings
+        USE_FINAL_NORM: True  # Enable final LayerNorm before output projection
+        USE_LOCAL_PRIOR: True  # Blend local mean prior for masked seeds
+        LOCAL_PRIOR_WEIGHT: 0.5  # Weight for local prior blending (1.0 = only mask_token, 0.0 = only local mean)
+        LOCAL_PRIOR_KERNEL: 5  # Kernel size for local prior blending (> 1)
+        SEED_NOISE_STD: 0.02  # Standard deviation of noise added to masked seeds during training  
+  INPAINT_MASK_DILATION:
+    value: 11  # Dilation kernel size (pixels) for inpaint mask - Must be odd
+  USE_TORCH_COMPILE:  # Enable PyTorch 2.0 torch.compile for faster training (experimental)
+    value: False
+
+  ### DATA                                                                      
+  DATASETS:
+    value:
+      SCRREAM:
+        VAL_SCENES: ["scene10_full_00","scene11_full_00","scene044_full_00","scene04_reduced_00","scene04_reduced_01","scene04_reduced_02"]  # List of validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width] in pixels
+        RESIZE_MODE: "resize+crop"  # Image resizing mode: "resize", "crop", "resize+crop", or "pad"
+        FEW_IMAGES: False  # If True, load only first 10 images per scene (for quick debugging)
+        SAMPLE_EVERY_N: 4  # Load every Nth frame from each scene (1 = all frames, 4 = every 4th frame)
+        LOAD_RGB_ONLY: True  # If True, ignore polarization data and load only RGB images
+        
+      HOUSECAT6D:
+        VAL_SCENES: ["val_scene1","val_scene2"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 4  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      CROMO:
+        TRAIN_SCENES: ["kitchen","station","facades"]  # Training scene names (list or string)
+        # VAL_SCENES: "station"  # Validation scene names (optional)
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 4  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+    
+      PSD:
+        TRAIN_SCENES: "PSD_Train"  # Training scene name (string or list)
+        VAL_SCENES: "PSD_Val"  # Validation scene name (string or list)
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 1  # Load every Nth frame (1 = all frames)
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      SCARED:
+        VAL_SCENES: ["v22","v23","v24","v25","v26","v27","v28","v29","v30","v31","v32","v33","v34"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 4  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing in dataset
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection (0-1)
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [1000, 1000]  # Size of highlight rectangle region [height, width]
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+      
+      STEREOMIS_TRACKING:
+        VAL_SCENES: ["P2_2"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 5  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      CHOLEC80:
+        VAL_SCENES: ["val"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 50  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      # POLARGB:
+      #   TRAIN_SCENES: "train"
+      #   VAL_SCENES: "test"
+      #   TARGET_SIZE: [896,896]
+      #   RESIZE_MODE: "resize+crop"
+      #   SAMPLE_EVERY_N: 1
+      #   LOAD_RGB_ONLY: True
+
+  BATCH_SIZE: # Max batch size with img size 896x896 is 20
+    value: 20  # Number of samples per batch (adjust based on GPU memory)
+  NUM_WORKERS:
+    value: 12  # Number of data loading worker processes (0 = main process only)
+  SHUFFLE:  
+    value: True  # Shuffle training data each epoch (False for validation/test)
+  PIN_MEMORY:
+    value: True  # Pin memory in DataLoader for faster GPU transfer (recommended: True)
+  PREFETCH_FACTOR:
+    value: 2  # Number of batches to prefetch per worker (higher = more memory usage)
+  
+  ### HIGHLIGHTS                                                        
+  MOGE_MODEL:
+    value: "Ruicheng/moge-2-vits-normal"  # MoGe model name for normal estimation (HuggingFace format)
+  SURFACE_ROUGHNESS:
+    value: 8.0  # Blinn-Phong surface roughness exponent (higher = sharper highlights)
+  INTENSITY:
+    value: 0.0  # Specular highlight intensity multiplier
+  LIGHT_DISTANCE_RANGE:
+    value: [0.0, 1]  # Range for light source distance sampling [min, max] (normalized)
+  LIGHT_LEFT_RIGHT_ANGLE:
+    value: [0, 360]  # Range for light source horizontal angle [min, max] in degrees
+  LIGHT_ABOVE_BELOW_ANGLE:
+    value: [0, 360]  # Range for light source vertical angle [min, max] in degrees
+  DATASET_HIGHLIGHT_DILATION:
+    value: 25  # Dilation kernel size (pixels) for dataset highlight masks
+  DATASET_HIGHLIGHT_THRESHOLD:
+    value: 0.9  # Brightness/luminance threshold (0-1) for detecting highlights in dataset images
+  DATASET_HIGHLIGHT_USE_LUMINANCE:
+    value: True  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B) for dataset highlights; if False, use simple mean brightness
+  HIGHLIGHT_COLOR:
+    value: [1.0, 1.0, 1.0]  # RGB color for synthetic highlights (normalized 0-1)
+  CLAMP_RECONSTRUCTION:
+    value: True  # Clamp reconstructed images to [0, 1] range if True
+
+  ### OPTIMIZATION                                                                      
+  EPOCHS:
+    value: 20  # Maximum number of training epochs<
+  LEARNING_RATE:
+    value: 1.0e-4  # Base learning rate for optimizer
+  WEIGHT_DECAY:   
+    value: 0.0  # L2 regularization weight (0.0 = no weight decay)
+  GRADIENT_ACCUMULATION_STEPS:
+    value: 1  # Number of steps to accumulate gradients before optimizer step (1 = no accumulation)
+  WARMUP:
+    value: 100  # Number of warmup steps for learning rate schedule (linear warmup from 0 to LR)
+  GRADIENT_CLIPPING_MAX_NORM:
+    value: 8  # Maximum gradient norm for clipping (set to -1 to disable clipping)
+  LR_SCHEDULER: 
+    value: 
+      ONPLATEAU:  # ReduceLROnPlateau scheduler (reduces LR when validation metric plateaus)
+        PATIENCE: 5  # Number of epochs to wait before reducing LR
+        FACTOR: 0.1  # Factor by which LR is reduced (new_lr = old_lr * factor)
+      COSINE:  # CosineAnnealingLR scheduler (cosine annealing schedule)
+        N_PERIODS: 1  # Number of cosine periods over training
+      # STEPWISE:  # StepLR scheduler (reduces LR at fixed step intervals)
+      #   N_STEPS: 5  # Number of times to reduce LR during training
+      #   GAMMA: 0.25  # Factor by which LR is reduced at each step (new_lr = old_lr * gamma)
+      # EXPONENTIAL:  # ExponentialLR scheduler (exponential decay)
+      #   GAMMA: 0.5  # Multiplicative factor for exponential decay
+
+  SWITCH_OPTIMIZER_EPOCH:
+    value: null  # Epoch number to switch from bootstrap to refining optimizer (null = no switch)
+  OPTIMIZER_BOOTSTRAP_NAME:
+    value: "AdamW"  # Optimizer name for initial training phase ("Adam", "SGD", etc.)
+  OPTIMIZER_REFINING_NAME:
+    value: "AdamW"  # Optimizer name for refining phase (used after SWITCH_OPTIMIZER_EPOCH)
+  EARLY_STOPPING_PATIENCE:
+    value: 20  # Number of epochs without improvement before stopping training
+  SAVE_INTERVAL:
+    value: 1000  # Number of training steps between model checkpoints
+  
+  DATASET_HIGHLIGHT_SUPERVISION_THRESHOLD:
+    value: 0.1  # Pixel highlights above this threshold (should be low) are excluded from supervision
+
+  ### LOSS WEIGHTS (relative to the total loss, NOT NORMALIZED LATER)                                                                              
+  SPECULAR_LOSS_WEIGHT:
+    value: 0.0  # Weight for specular component reconstruction loss
+  DIFFUSE_LOSS_WEIGHT:
+    value: 1.0  # Weight for diffuse component reconstruction loss
+  HIGHLIGHT_LOSS_WEIGHT:
+    value: 0.0  # Weight for highlight mask regression loss
+  IMAGE_RECONSTRUCTION_LOSS_WEIGHT:
+    value: 0.0  # Weight for full image reconstruction loss
+  SATURATION_RING_LOSS_WEIGHT:
+    value: 0.0  # Weight for saturation ring consistency loss (around highlight regions)
+  RING_KERNEL_SIZE:
+    value: 11  # Kernel size (odd number) for saturation ring dilation around highlights
+  RING_VAR_WEIGHT:
+    value: 0.5  # Weight for variance matching in saturation ring loss (vs mean matching)
+  RING_TEXTURE_WEIGHT:
+    value: 0.0  # Weight for texture consistency term in saturation ring loss
+  HLREG_W_L1:
+    value: 1.0  # Weight for L1 loss in highlight regression
+  HLREG_USE_CHARB:
+    value: True  # Use Charbonnier loss (smooth L1) instead of standard L1 if True
+  HLREG_W_DICE:
+    value: 0.2  # Weight for Dice loss in highlight regression (for mask overlap)
+  HLREG_W_SSIM:
+    value: 0.0  # Weight for SSIM loss in highlight regression
+  HLREG_W_GRAD:
+    value: 0.0  # Weight for gradient loss in highlight regression
+  HLREG_W_TV:
+    value: 0.0  # Weight for total variation loss in highlight regression
+  HLREG_BALANCE_MODE:
+    value: "auto"   # Class balancing mode for highlight regression: 'none' | 'auto' | 'pos_weight'
+  HLREG_POS_WEIGHT:
+    value: 1.0      # Positive class weight (used only if BALANCE_MODE == 'pos_weight')
+  HLREG_FOCAL_GAMMA:
+    value: 2.0      # Focal loss gamma parameter (0.0 = standard BCE, 1.0-2.0 helps with gradient vanishing)
+  
+  WEIGHT_TOKEN_INPAINT:
+    value: 0.0  # Weight for token-space inpainting loss (L1 + cosine similarity in feature space)
+  WEIGHT_CONTEXT_IDENTITY: 
+    value: 0.0  # LEAVE TO 0.0: Weight for L1 loss on context (non-masked) regions (identity preservation) 
+  WEIGHT_TV_IN_HOLE:
+    value: 0.0  # LEAVE TO 0.0: Weight for total variation loss inside masked/hole regions
+  RING_DILATE_KERNEL:
+    value: 17  # Dilation kernel size (odd number) for creating ring mask around highlights
+  WEIGHT_SEAM:
+    value: 0.0  # Weight for gradient matching loss on saturation ring
+  SEAM_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 in seam loss (smooth L1 for boundary consistency)
+  SEAM_WEIGHT_GRAD:
+    value: 0.0  # Weight for gradient matching term inside seam loss (0.0 = disable gradient term)
+  TOKEN_FEAT_ALPHA:
+    value: 0.5  # Mixing factor for token feature loss: alpha * L1 + (1-alpha) * (1-cosine_sim)
+
+  ### DIFFUSE HIGHLIGHT PENALTY
+  WEIGHT_DIFFUSE_HIGHLIGHT_PENALTY:
+    value: 0.0  # Weight for penalty loss on highlights in diffuse decoder output (0.0 = disabled)
+  DIFFUSE_HL_THRESHOLD:
+    value: 0.8  # Brightness/luminance threshold for detecting highlights in diffuse (0.0-1.0)
+  DIFFUSE_HL_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 for diffuse highlight penalty
+  DIFFUSE_HL_PENALTY_MODE:
+    value: "brightness"  # Penalty mode: "brightness" (penalize brightness/luminance above threshold) or "pixel" (penalize RGB values directly)
+  DIFFUSE_HL_TARGET_BRIGHTNESS:
+    value: null  # Target brightness/luminance for penalized pixels (null = use threshold value)
+  DIFFUSE_HL_USE_LUMINANCE:
+    value: True  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B); if False, use simple mean brightness
+
+  ### LOGGING, RESULTS AND WANDB                                                                 
+  LOG_INTERVAL:
+    value: 1  # Number of training steps between console log outputs
+  WANDB_LOG_INTERVAL:
+    value: 1  # Number of training steps between WandB metric logs
+  IMAGE_LOG_INTERVAL:
+    value: 5  # Number of training steps between image logging to WandB
+  NO_WANDB:
+    value: False  # Disable WandB logging if True (useful for local debugging)
+  MODEL_WATCHER_FREQ_WANDB:
+    value: 50  # Frequency (in steps) for logging model parameter histograms to WandB
+  WANDB_ENTITY:
+    value: "unreflect-anything"  # WandB organization/entity name
+  WANDB_PROJECT:
+    value: "UnReflectAnything"  # WandB project name
+  NOTES:
+    value: "896 Decoder Pretraining"  # Notes/description for this training run
+  

configs/figuremeout.yaml

@@ -0,0 +1,306 @@
+### BASELINE: CONVERGES AFTER LONG
+
+parameters:
+
+  ### MODEL ARCHITECTURE                                                       
+  MODEL:
+    value:
+      MODEL_CLASS: "UnReflect_Model "  # Main model class name (must match class in models.py)
+      MODEL_MODULE: "models"  # Module name to import model classes from (default: "models")
+      RGB_ENCODER:
+        ENCODER: "facebook/dinov3-vitl16-pretrain-lvd1689m"  # DINOv3 encoder model name (HuggingFace format)
+        IMAGE_SIZE: 448  # Input image size (height and width in pixels)
+        RETURN_SELECTED_LAYERS: [3, 6, 9  , 12]  # Transformer layer indices to extract features from (0-indexed)
+        RGB_ENCODER_LR: 0.0  # Learning rate for RGB encoder (0.0 = frozen, must be explicitly set)
+      DECODERS:
+        diffuse:
+          USE_FILM: False  # Enable FiLM (Feature-wise Linear Modulation) conditioning in decoder
+          FEATURE_DIM: 1024  # Feature dimension for decoder (should match encoder output)
+          REASSEMBLE_OUT_CHANNELS: [768,1024,1536,2048]  # Output channels for each decoder stage (DPT-style reassembly)
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder ("ignore", "project", etc.)
+          # FROM_PRETRAINED: "weights/rgb_decoder.pth"  # Path to pretrained decoder weights (optional)
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [448,448]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 3  # Number of output channels (3 for RGB diffuse image)
+          DECODER_LR: 1.0e-4 # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: 1  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+          TRAIN_RGB_HEAD: True  # Whether to train RGB head (true/false, null = train if DECODER_LR != 0)
+        highlight:
+          USE_FILM: False  # Enable FiLM conditioning in highlight decoder
+          FEATURE_DIM: 1024  # Feature dimension for highlight decoder
+          REASSEMBLE_OUT_CHANNELS: [96,192,384,768]  # Output channels for each decoder stage
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [448,448]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 1  # Number of output channels (1 for highlight mask)
+          DECODER_LR: 5.0e-4  # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: null  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+      TOKEN_INPAINTER:
+        TOKEN_INPAINTER_CLASS: "TokenInpainter_Naive"  # Token inpainter class name
+        # TOKEN_INPAINTER_MODULE: "token_inpainters"  # Module name to import token inpainter from
+        # FROM_PRETRAINED: "weights/token_inpainter.pth"  # Path to pretrained token inpainter weights (optional)
+        TOKEN_INPAINTER_LR: 0.0  # Learning rate for token inpainter (can differ from base LR)
+        DEPTH: 6  # Number of transformer blocks
+        HEADS: 16  # Number of attention heads
+        DROP: 0 # Dropout rate 
+        USE_POSITIONAL_ENCODING: True  # Enable 2D sinusoidal positional encodings
+        USE_FINAL_NORM: True  # Enable final LayerNorm before output projection
+        USE_LOCAL_PRIOR: True  # Blend local mean prior for masked seeds
+        LOCAL_PRIOR_WEIGHT: 0.5  # Weight for local prior blending (1.0 = only mask_token, 0.0 = only local mean)
+        LOCAL_PRIOR_KERNEL: 5  # Kernel size for local prior blending (> 1)
+        SEED_NOISE_STD: 0.02  # Standard deviation of noise added to masked seeds during training  
+  INPAINT_MASK_DILATION:
+    value: 1  # Dilation kernel size (pixels) for inpaint mask - Must be odd
+  USE_TORCH_COMPILE:  # Enable PyTorch 2.0 torch.compile for faster training (experimental)
+    value: False
+
+  ### DATA                                                                      
+  DATASETS:
+    value:
+      SCRREAM:
+        VAL_SCENES: ["scene10_full_00","scene11_full_00","scene044_full_00","scene04_reduced_00","scene04_reduced_01","scene04_reduced_02"]  # List of validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width] in pixels
+        RESIZE_MODE: "resize+crop"  # Image resizing mode: "resize", "crop", "resize+crop", or "pad"
+        FEW_IMAGES: False  # If True, load only first 10 images per scene (for quick debugging)
+        SAMPLE_EVERY_N: 2  # Load every Nth frame from each scene (1 = all frames, 4 = every 4th frame)
+        LOAD_RGB_ONLY: True  # If True, ignore polarization data and load only RGB images
+        
+      HOUSECAT6D:
+        VAL_SCENES: ["val_scene1","val_scene2"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 2  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      CROMO:
+        TRAIN_SCENES: ["kitchen"]  # Training scene names (list or string)
+        # VAL_SCENES: "station"  # Validation scene names (optional)
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 2  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+    
+      PSD:
+        TRAIN_SCENES: "PSD_Train"  # Training scene name (string or list)
+        VAL_SCENES: "PSD_Val"  # Validation scene name (string or list)
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 1  # Load every Nth frame (1 = all frames)
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      SCARED:
+        VAL_SCENES: ["v22","v23","v24","v25","v26","v27","v28","v29","v30","v31","v32","v33","v34"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 8  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing in dataset
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection (0-1)
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [1000, 1000]  # Size of highlight rectangle region [height, width]
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+      
+      STEREOMIS_TRACKING:
+        VAL_SCENES: ["P2_2"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 4  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      CHOLEC80:
+        VAL_SCENES: ["val"]  # Validation scene names
+        TARGET_SIZE: [448,448]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 10  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      # POLARGB:
+      #   TRAIN_SCENES: "train"
+      #   VAL_SCENES: "test"
+      #   TARGET_SIZE: [448,448]
+      #   RESIZE_MODE: "resize+crop"
+      #   SAMPLE_EVERY_N: 1
+      #   LOAD_RGB_ONLY: True
+
+  BATCH_SIZE: # Max batch size with img size 448 is 32
+    value: 32 # Number of samples per batch (adjust based on GPU memory)
+  NUM_WORKERS:
+    value: 8  # Number of data loading worker processes (0 = main process only)
+  SHUFFLE:
+    value: True  # Shuffle training data each epoch (False for validation/test)
+  PIN_MEMORY:
+    value: True  # Pin memory in DataLoader for faster GPU transfer (recommended: True)
+  PREFETCH_FACTOR:
+    value: 2  # Number of batches to prefetch per worker (higher = more memory usage)
+  
+  ### HIGHLIGHTS                                                        
+  MOGE_MODEL:
+    value: "Ruicheng/moge-2-vits-normal"  # MoGe model name for normal estimation (HuggingFace format)
+  SURFACE_ROUGHNESS:
+    value: 8.0  # Blinn-Phong surface roughness exponent (higher = sharper highlights)
+  INTENSITY:
+    value: 0.0  # Specular highlight intensity multiplier
+  LIGHT_DISTANCE_RANGE:
+    value: [0.0, 1]  # Range for light source distance sampling [min, max] (normalized)
+  LIGHT_LEFT_RIGHT_ANGLE:
+    value: [0, 360]  # Range for light source horizontal angle [min, max] in degrees
+  LIGHT_ABOVE_BELOW_ANGLE:
+    value: [0, 360]  # Range for light source vertical angle [min, max] in degrees
+  DATASET_HIGHLIGHT_DILATION:
+    value: 25  # Dilation kernel size (pixels) for dataset highlight masks
+  DATASET_HIGHLIGHT_THRESHOLD:
+    value: 0.9  # Brightness/luminance threshold (0-1) for detecting highlights in dataset images
+  DATASET_HIGHLIGHT_USE_LUMINANCE:
+    value: True  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B) for dataset highlights; if False, use simple mean brightness
+  HIGHLIGHT_COLOR:
+    value: [1.0, 1.0, 1.0]  # RGB color for synthetic highlights (normalized 0-1)
+  CLAMP_RECONSTRUCTION:
+    value: True  # Clamp reconstructed images to [0, 1] range if True
+
+  ### OPTIMIZATION                                                                      
+  LEARNING_RATE:
+    value: 1.0e-3  # Base learning rate for optimizer
+  WEIGHT_DECAY:   
+    value: 0.0  # L2 regularization weight (0.0 = no weight decay)
+  EPOCHS:
+    value: 25  # Maximum number of training epochs
+  GRADIENT_ACCUMULATION_STEPS:
+    value: 1  # Number of steps to accumulate gradients before optimizer step (1 = no accumulation)
+  WARMUP:
+    value: 200  # Number of warmup steps for learning rate schedule (linear warmup from 0 to LR)
+  GRADIENT_CLIPPING_MAX_NORM:
+    value: 8  # Maximum gradient norm for clipping (set to -1 to disable clipping)
+  LR_SCHEDULER: 
+    value: 
+      ONPLATEAU:  # ReduceLROnPlateau scheduler (reduces LR when validation metric plateaus)
+        PATIENCE: 5  # Number of epochs to wait before reducing LR
+        FACTOR: 0.1  # Factor by which LR is reduced (new_lr = old_lr * factor)
+      # COSINE:  # CosineAnnealingLR scheduler (cosine annealing schedule)
+      #   N_PERIODS: 5  # Number of cosine periods over training
+      STEPWISE:  # StepLR scheduler (reduces LR at fixed step intervals)
+        N_STEPS: 4  # Number of times to reduce LR during training
+        GAMMA: 0.5  # Factor by which LR is reduced at each step (new_lr = old_lr * gamma)
+      # EXPONENTIAL:  # ExponentialLR scheduler (exponential decay)
+      #   GAMMA: 0.5  # Multiplicative factor for exponential decay
+
+  SWITCH_OPTIMIZER_EPOCH:
+    value: null  # Epoch number to switch from bootstrap to refining optimizer (null = no switch)
+  OPTIMIZER_BOOTSTRAP_NAME:
+    value: "AdamW"  # Optimizer name for initial training phase ("Adam", "SGD", etc.)
+  OPTIMIZER_REFINING_NAME:
+    value: "AdamW"  # Optimizer name for refining phase (used after SWITCH_OPTIMIZER_EPOCH)
+  EARLY_STOPPING_PATIENCE:
+    value: 10  # Number of epochs without improvement before stopping training
+  SAVE_INTERVAL:
+    value: 1000  # Number of training steps between model checkpoints
+  
+  DATASET_HIGHLIGHT_SUPERVISION_THRESHOLD:
+    value: 0.1  # Pixel highlights above this threshold (should be low) are excluded from supervision
+
+  ### LOSS WEIGHTS (relative to the total loss, NOT NORMALIZED LATER)                                                                              
+  SPECULAR_LOSS_WEIGHT:
+    value: 0.0  # Weight for specular component reconstruction loss
+  DIFFUSE_LOSS_WEIGHT:
+    value: 1.0  # Weight for diffuse component reconstruction loss
+  HIGHLIGHT_LOSS_WEIGHT:
+    value: 0.0  # Weight for highlight mask regression loss
+  IMAGE_RECONSTRUCTION_LOSS_WEIGHT:
+    value: 0.0  # Weight for full image reconstruction loss
+  SATURATION_RING_LOSS_WEIGHT:
+    value: 0.0  # Weight for saturation ring consistency loss (around highlight regions)
+  RING_KERNEL_SIZE:
+    value: 11  # Kernel size (odd number) for saturation ring dilation around highlights
+  RING_VAR_WEIGHT:
+    value: 0.5  # Weight for variance matching in saturation ring loss (vs mean matching)
+  RING_TEXTURE_WEIGHT:
+    value: 1.0  # Weight for texture consistency term in saturation ring loss
+  HLREG_W_L1:
+    value: 1.0  # Weight for L1 loss in highlight regression
+  HLREG_USE_CHARB:
+    value: True  # Use Charbonnier loss (smooth L1) instead of standard L1 if True
+  HLREG_W_DICE:
+    value: 0.2  # Weight for Dice loss in highlight regression (for mask overlap)
+  HLREG_W_SSIM:
+    value: 0.0  # Weight for SSIM loss in highlight regression
+  HLREG_W_GRAD:
+    value: 0.0  # Weight for gradient loss in highlight regression
+  HLREG_W_TV:
+    value: 0.0  # Weight for total variation loss in highlight regression
+  HLREG_BALANCE_MODE:
+    value: "auto"   # Class balancing mode for highlight regression: 'none' | 'auto' | 'pos_weight'
+  HLREG_POS_WEIGHT:
+    value: 1.0      # Positive class weight (used only if BALANCE_MODE == 'pos_weight')
+  HLREG_FOCAL_GAMMA:
+    value: 2.0      # Focal loss gamma parameter (0.0 = standard BCE, 1.0-2.0 helps with gradient vanishing)
+  
+  WEIGHT_TOKEN_INPAINT:
+    value: 0.0  # Weight for token-space inpainting loss (L1 + cosine similarity in feature space)
+  WEIGHT_CONTEXT_IDENTITY: 
+    value: 0.0  # LEAVE TO 0.0: Weight for L1 loss on context (non-masked) regions (identity preservation) 
+  WEIGHT_TV_IN_HOLE:
+    value: 0.0  # LEAVE TO 0.0: Weight for total variation loss inside masked/hole regions
+  RING_DILATE_KERNEL:
+    value: 17  # Dilation kernel size (odd number) for creating ring mask around highlights
+  WEIGHT_SEAM:
+    value: 0.5  # Weight for gradient matching loss on saturation ring
+  SEAM_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 in seam loss (smooth L1 for boundary consistency)
+  SEAM_WEIGHT_GRAD:
+    value: 0.0  # Weight for gradient matching term inside seam loss (0.0 = disable gradient term)
+  TOKEN_FEAT_ALPHA:
+    value: 0.5  # Mixing factor for token feature loss: alpha * L1 + (1-alpha) * (1-cosine_sim)
+
+  ### DIFFUSE HIGHLIGHT PENALTY
+  WEIGHT_DIFFUSE_HIGHLIGHT_PENALTY:
+    value: 0.0  # Weight for penalty loss on highlights in diffuse decoder output (0.0 = disabled)
+  DIFFUSE_HL_THRESHOLD:
+    value: 0.8  # Brightness/luminance threshold for detecting highlights in diffuse (0.0-1.0)
+  DIFFUSE_HL_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 for diffuse highlight penalty
+  DIFFUSE_HL_PENALTY_MODE:
+    value: "brightness"  # Penalty mode: "brightness" (penalize brightness/luminance above threshold) or "pixel" (penalize RGB values directly)
+  DIFFUSE_HL_TARGET_BRIGHTNESS:
+    value: null  # Target brightness/luminance for penalized pixels (null = use threshold value)
+  DIFFUSE_HL_USE_LUMINANCE:
+    value: False  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B); if False, use simple mean brightness
+
+  ### LOGGING, RESULTS AND WANDB                                                                 
+  LOG_INTERVAL:
+    value: 1  # Number of training steps between console log outputs
+  WANDB_LOG_INTERVAL:
+    value: 1  # Number of training steps between WandB metric logs
+  IMAGE_LOG_INTERVAL:
+    value: 5  # Number of training steps between image logging to WandB
+  NO_WANDB:
+    value: False  # Disable WandB logging if True (useful for local debugging)
+  MODEL_WATCHER_FREQ_WANDB:
+    value: 50  # Frequency (in steps) for logging model parameter histograms to WandB
+  WANDB_ENTITY:
+    value: "unreflect-anything"  # WandB organization/entity name
+  WANDB_PROJECT:
+    value: "UnReflectAnything"  # WandB project name
+  NOTES:
+    value: "Final - TKI Learns, Decoder Learns faster"  # Notes/description for this training run

configs/pretrain_tki.yaml

@@ -0,0 +1,306 @@
+### BASELINE: CONVERGES AFTER LONG
+
+parameters:
+
+  ### MODEL ARCHITECTURE                                                       
+  MODEL:
+    value:
+      MODEL_CLASS: "UnReflect_Model_TokenInpainter"  # Main model class name (must match class in models.py)
+      MODEL_MODULE: "models"  # Module name to import model classes from (default: "models")
+      RGB_ENCODER:
+        ENCODER: "facebook/dinov3-vitl16-pretrain-lvd1689m"  # DINOv3 encoder model name (HuggingFace format)
+        IMAGE_SIZE: 896  # Input image size (height and width in pixels)
+        RETURN_SELECTED_LAYERS: [3, 6, 9  , 12]  # Transformer layer indices to extract features from (0-indexed)
+        RGB_ENCODER_LR: 0.0  # Learning rate for RGB encoder (0.0 = frozen, must be explicitly set)
+      DECODERS:
+        diffuse:
+          USE_FILM: False  # Enable FiLM (Feature-wise Linear Modulation) conditioning in decoder
+          FEATURE_DIM: 1024  # Feature dimension for decoder (should match encoder output)
+          REASSEMBLE_OUT_CHANNELS: [768,1024,1536,2048]  # Output channels for each decoder stage (DPT-style reassembly)
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder ("ignore", "project", etc.)
+          FROM_PRETRAINED: "weights/decoder_896.pth"  # Path to pretrained decoder weights (optional)
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [896,896]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 3  # Number of output channels (3 for RGB diffuse image)
+          DECODER_LR: 0.0 # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: 1  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+          TRAIN_RGB_HEAD: True  # Whether to train RGB head (true/false, null = train if DECODER_LR != 0)
+        highlight:
+          USE_FILM: False  # Enable FiLM conditioning in highlight decoder
+          FEATURE_DIM: 1024  # Feature dimension for highlight decoder
+          REASSEMBLE_OUT_CHANNELS: [96,192,384,768]  # Output channels for each decoder stage
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [896,896]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 1  # Number of output channels (1 for highlight mask)
+          DECODER_LR: 5.0e-4  # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: null  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+      TOKEN_INPAINTER:
+        TOKEN_INPAINTER_CLASS: "TokenInpainter_Prior"  # Token inpainter class name
+        TOKEN_INPAINTER_MODULE: "token_inpainters"  # Module name to import token inpainter from
+        # FROM_PRETRAINED: "weights/token_inpainter.pth"  # Path to pretrained token inpainter weights (optional)
+        TOKEN_INPAINTER_LR: 1.0e-4  # Learning rate for token inpainter (can differ from base LR)
+        DEPTH: 6  # Number of transformer blocks
+        HEADS: 16  # Number of attention heads
+        DROP: 0 # Dropout rate 
+        USE_POSITIONAL_ENCODING: True  # Enable 2D sinusoidal positional encodings
+        USE_FINAL_NORM: True  # Enable final LayerNorm before output projection
+        USE_LOCAL_PRIOR: True  # Blend local mean prior for masked seeds
+        LOCAL_PRIOR_WEIGHT: 0.5  # Weight for local prior blending (1.0 = only mask_token, 0.0 = only local mean)
+        LOCAL_PRIOR_KERNEL: 5  # Kernel size for local prior blending (> 1)
+        SEED_NOISE_STD: 0.02  # Standard deviation of noise added to masked seeds during training  
+  INPAINT_MASK_DILATION:
+    value: 1  # Dilation kernel size (pixels) for inpaint mask - Must be odd
+  USE_TORCH_COMPILE:  # Enable PyTorch 2.0 torch.compile for faster training (experimental)
+    value: False
+
+  ### DATA                                                                      
+  DATASETS:
+    value:
+      SCRREAM:
+        VAL_SCENES: ["scene10_full_00","scene11_full_00","scene044_full_00","scene04_reduced_00","scene04_reduced_01","scene04_reduced_02"]  # List of validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width] in pixels
+        RESIZE_MODE: "resize+crop"  # Image resizing mode: "resize", "crop", "resize+crop", or "pad"
+        FEW_IMAGES: False  # If True, load only first 10 images per scene (for quick debugging)
+        SAMPLE_EVERY_N: 2  # Load every Nth frame from each scene (1 = all frames, 4 = every 4th frame)
+        LOAD_RGB_ONLY: True  # If True, ignore polarization data and load only RGB images
+        
+      HOUSECAT6D:
+        VAL_SCENES: ["val_scene1","val_scene2"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 2  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      CROMO:
+        TRAIN_SCENES: ["kitchen"]  # Training scene names (list or string)
+        # VAL_SCENES: "station"  # Validation scene names (optional)
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 2  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+    
+      PSD:
+        TRAIN_SCENES: "PSD_Train"  # Training scene name (string or list)
+        VAL_SCENES: "PSD_Val"  # Validation scene name (string or list)
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        FEW_IMAGES: False  # Load only first 10 images if True
+        SAMPLE_EVERY_N: 1  # Load every Nth frame (1 = all frames)
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+      SCARED:
+        VAL_SCENES: ["v22","v23","v24","v25","v26","v27","v28","v29","v30","v31","v32","v33","v34"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 8  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing in dataset
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection (0-1)
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [1000, 1000]  # Size of highlight rectangle region [height, width]
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+      
+      STEREOMIS_TRACKING:
+        VAL_SCENES: ["P2_2"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 4  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      CHOLEC80:
+        VAL_SCENES: ["val"]  # Validation scene names
+        TARGET_SIZE: [896,896]  # Target image size [height, width]
+        RESIZE_MODE: "resize+crop"  # Image resizing mode
+        SAMPLE_EVERY_N: 10  # Load every Nth frame
+        LOAD_RGB_ONLY: True  # Ignore polarization data if True
+        FEW_IMAGES: False  # Load only first 10 images if True
+        HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+        HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+        HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+        HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+        HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+        HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+      # POLARGB:
+      #   TRAIN_SCENES: "train"
+      #   VAL_SCENES: "test"
+      #   TARGET_SIZE: [896,896]
+      #   RESIZE_MODE: "resize+crop"
+      #   SAMPLE_EVERY_N: 1
+      #   LOAD_RGB_ONLY: True
+
+  BATCH_SIZE: # Max batch size with img size 896 is 32
+    value: 16 # Number of samples per batch (adjust based on GPU memory)
+  NUM_WORKERS:
+    value: 8  # Number of data loading worker processes (0 = main process only)
+  SHUFFLE:
+    value: True  # Shuffle training data each epoch (False for validation/test)
+  PIN_MEMORY:
+    value: True  # Pin memory in DataLoader for faster GPU transfer (recommended: True)
+  PREFETCH_FACTOR:
+    value: 2  # Number of batches to prefetch per worker (higher = more memory usage)
+  
+  ### HIGHLIGHTS                                                        
+  MOGE_MODEL:
+    value: "Ruicheng/moge-2-vits-normal"  # MoGe model name for normal estimation (HuggingFace format)
+  SURFACE_ROUGHNESS:
+    value: 8.0  # Blinn-Phong surface roughness exponent (higher = sharper highlights)
+  INTENSITY:
+    value: 2.0  # Specular highlight intensity multiplier
+  LIGHT_DISTANCE_RANGE:
+    value: [0.0, 1]  # Range for light source distance sampling [min, max] (normalized)
+  LIGHT_LEFT_RIGHT_ANGLE:
+    value: [0, 360]  # Range for light source horizontal angle [min, max] in degrees
+  LIGHT_ABOVE_BELOW_ANGLE:
+    value: [0, 360]  # Range for light source vertical angle [min, max] in degrees
+  DATASET_HIGHLIGHT_DILATION:
+    value: 25  # Dilation kernel size (pixels) for dataset highlight masks
+  DATASET_HIGHLIGHT_THRESHOLD:
+    value: 0.9  # Brightness/luminance threshold (0-1) for detecting highlights in dataset images
+  DATASET_HIGHLIGHT_USE_LUMINANCE:
+    value: True  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B) for dataset highlights; if False, use simple mean brightness
+  HIGHLIGHT_COLOR:
+    value: [1.0, 1.0, 1.0]  # RGB color for synthetic highlights (normalized 0-1)
+  CLAMP_RECONSTRUCTION:
+    value: True  # Clamp reconstructed images to [0, 1] range if True
+
+  ### OPTIMIZATION                                                                      
+  LEARNING_RATE:
+    value: 1.0e-3  # Base learning rate for optimizer
+  WEIGHT_DECAY:   
+    value: 0.0  # L2 regularization weight (0.0 = no weight decay)
+  EPOCHS:
+    value: 25  # Maximum number of training epochs
+  GRADIENT_ACCUMULATION_STEPS:
+    value: 1  # Number of steps to accumulate gradients before optimizer step (1 = no accumulation)
+  WARMUP:
+    value: 200  # Number of warmup steps for learning rate schedule (linear warmup from 0 to LR)
+  GRADIENT_CLIPPING_MAX_NORM:
+    value: 8  # Maximum gradient norm for clipping (set to -1 to disable clipping)
+  LR_SCHEDULER: 
+    value: 
+      ONPLATEAU:  # ReduceLROnPlateau scheduler (reduces LR when validation metric plateaus)
+        PATIENCE: 5  # Number of epochs to wait before reducing LR
+        FACTOR: 0.1  # Factor by which LR is reduced (new_lr = old_lr * factor)
+      # COSINE:  # CosineAnnealingLR scheduler (cosine annealing schedule)
+      #   N_PERIODS: 5  # Number of cosine periods over training
+      STEPWISE:  # StepLR scheduler (reduces LR at fixed step intervals)
+        N_STEPS: 4  # Number of times to reduce LR during training
+        GAMMA: 0.5  # Factor by which LR is reduced at each step (new_lr = old_lr * gamma)
+      # EXPONENTIAL:  # ExponentialLR scheduler (exponential decay)
+      #   GAMMA: 0.5  # Multiplicative factor for exponential decay
+
+  SWITCH_OPTIMIZER_EPOCH:
+    value: null  # Epoch number to switch from bootstrap to refining optimizer (null = no switch)
+  OPTIMIZER_BOOTSTRAP_NAME:
+    value: "AdamW"  # Optimizer name for initial training phase ("Adam", "SGD", etc.)
+  OPTIMIZER_REFINING_NAME:
+    value: "AdamW"  # Optimizer name for refining phase (used after SWITCH_OPTIMIZER_EPOCH)
+  EARLY_STOPPING_PATIENCE:
+    value: 10  # Number of epochs without improvement before stopping training
+  SAVE_INTERVAL:
+    value: 1000  # Number of training steps between model checkpoints
+  
+  DATASET_HIGHLIGHT_SUPERVISION_THRESHOLD:
+    value: 0.1  # Pixel highlights above this threshold (should be low) are excluded from supervision
+
+  ### LOSS WEIGHTS (relative to the total loss, NOT NORMALIZED LATER)                                                                              
+  SPECULAR_LOSS_WEIGHT:
+    value: 0.0  # Weight for specular component reconstruction loss
+  DIFFUSE_LOSS_WEIGHT:
+    value: 0.0  # Weight for diffuse component reconstruction loss
+  HIGHLIGHT_LOSS_WEIGHT:
+    value: 0.0  # Weight for highlight mask regression loss
+  IMAGE_RECONSTRUCTION_LOSS_WEIGHT:
+    value: 0.0  # Weight for full image reconstruction loss
+  SATURATION_RING_LOSS_WEIGHT:
+    value: 0.0  # Weight for saturation ring consistency loss (around highlight regions)
+  RING_KERNEL_SIZE:
+    value: 11  # Kernel size (odd number) for saturation ring dilation around highlights
+  RING_VAR_WEIGHT:
+    value: 0.5  # Weight for variance matching in saturation ring loss (vs mean matching)
+  RING_TEXTURE_WEIGHT:
+    value: 1.0  # Weight for texture consistency term in saturation ring loss
+  HLREG_W_L1:
+    value: 1.0  # Weight for L1 loss in highlight regression
+  HLREG_USE_CHARB:
+    value: True  # Use Charbonnier loss (smooth L1) instead of standard L1 if True
+  HLREG_W_DICE:
+    value: 0.2  # Weight for Dice loss in highlight regression (for mask overlap)
+  HLREG_W_SSIM:
+    value: 0.0  # Weight for SSIM loss in highlight regression
+  HLREG_W_GRAD:
+    value: 0.0  # Weight for gradient loss in highlight regression
+  HLREG_W_TV:
+    value: 0.0  # Weight for total variation loss in highlight regression
+  HLREG_BALANCE_MODE:
+    value: "auto"   # Class balancing mode for highlight regression: 'none' | 'auto' | 'pos_weight'
+  HLREG_POS_WEIGHT:
+    value: 1.0      # Positive class weight (used only if BALANCE_MODE == 'pos_weight')
+  HLREG_FOCAL_GAMMA:
+    value: 2.0      # Focal loss gamma parameter (0.0 = standard BCE, 1.0-2.0 helps with gradient vanishing)
+  
+  WEIGHT_TOKEN_INPAINT:
+    value: 1.0  # Weight for token-space inpainting loss (L1 + cosine similarity in feature space)
+  WEIGHT_CONTEXT_IDENTITY: 
+    value: 0.0  # LEAVE TO 0.0: Weight for L1 loss on context (non-masked) regions (identity preservation) 
+  WEIGHT_TV_IN_HOLE:
+    value: 0.0  # LEAVE TO 0.0: Weight for total variation loss inside masked/hole regions
+  RING_DILATE_KERNEL:
+    value: 17  # Dilation kernel size (odd number) for creating ring mask around highlights
+  WEIGHT_SEAM:
+    value: 0.5  # Weight for gradient matching loss on saturation ring
+  SEAM_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 in seam loss (smooth L1 for boundary consistency)
+  SEAM_WEIGHT_GRAD:
+    value: 0.0  # Weight for gradient matching term inside seam loss (0.0 = disable gradient term)
+  TOKEN_FEAT_ALPHA:
+    value: 0.5  # Mixing factor for token feature loss: alpha * L1 + (1-alpha) * (1-cosine_sim)
+
+  ### DIFFUSE HIGHLIGHT PENALTY
+  WEIGHT_DIFFUSE_HIGHLIGHT_PENALTY:
+    value: 0.1  # Weight for penalty loss on highlights in diffuse decoder output (0.0 = disabled)
+  DIFFUSE_HL_THRESHOLD:
+    value: 0.8  # Brightness/luminance threshold for detecting highlights in diffuse (0.0-1.0)
+  DIFFUSE_HL_USE_CHARB:
+    value: True  # Use Charbonnier loss instead of L1 for diffuse highlight penalty
+  DIFFUSE_HL_PENALTY_MODE:
+    value: "brightness"  # Penalty mode: "brightness" (penalize brightness/luminance above threshold) or "pixel" (penalize RGB values directly)
+  DIFFUSE_HL_TARGET_BRIGHTNESS:
+    value: null  # Target brightness/luminance for penalized pixels (null = use threshold value)
+  DIFFUSE_HL_USE_LUMINANCE:
+    value: False  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B); if False, use simple mean brightness
+
+  ### LOGGING, RESULTS AND WANDB                                                                 
+  LOG_INTERVAL:
+    value: 1  # Number of training steps between console log outputs
+  WANDB_LOG_INTERVAL:
+    value: 1  # Number of training steps between WandB metric logs
+  IMAGE_LOG_INTERVAL:
+    value: 5  # Number of training steps between image logging to WandB
+  NO_WANDB:
+    value: False  # Disable WandB logging if True (useful for local debugging)
+  MODEL_WATCHER_FREQ_WANDB:
+    value: 50  # Frequency (in steps) for logging model parameter histograms to WandB
+  WANDB_ENTITY:
+    value: "unreflect-anything"  # WandB organization/entity name
+  WANDB_PROJECT:
+    value: "UnReflectAnything"  # WandB project name
+  NOTES:
+    value: "Final - TKI Learns, Decoder Learns faster"  # Notes/description for this training run

configs/pretrained_config.yaml

@@ -0,0 +1,306 @@
+### BASELINE: CONVERGES AFTER LONG
+
+parameters:
+
+  ### MODEL ARCHITECTURE                                                       
+  MODEL:
+    value:
+      MODEL_CLASS: "UnReflect_Model_TokenInpainter"  # Main model class name (must match class in models.py)
+      MODEL_MODULE: "models"  # Module name to import model classes from (default: "models")
+      RGB_ENCODER:
+        ENCODER: "facebook/dinov3-vitl16-pretrain-lvd1689m"  # DINOv3 encoder model name (HuggingFace format)
+        IMAGE_SIZE: 448  # Input image size (height and width in pixels)
+        RETURN_SELECTED_LAYERS: [3, 6, 9, 12]  # Transformer layer indices to extract features from (0-indexed)
+        RGB_ENCODER_LR: 0.0  # Learning rate for RGB encoder (0.0 = frozen, must be explicitly set)
+      DECODERS:
+        diffuse:
+          USE_FILM: False  # Enable FiLM (Feature-wise Linear Modulation) conditioning in decoder
+          FEATURE_DIM: 1024  # Feature dimension for decoder (should match encoder output)
+          REASSEMBLE_OUT_CHANNELS: [768,1024,1536,2048]  # Output channels for each decoder stage (DPT-style reassembly)
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder ("ignore", "project", etc.)
+          FROM_PRETRAINED: "weights/rgb_decoder.pth"  # Path to pretrained decoder weights (optional)
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [448,448]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 3  # Number of output channels (3 for RGB diffuse image)
+          DECODER_LR: 1.0e-5 # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: 1  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+          TRAIN_RGB_HEAD: True  # Whether to train RGB head (true/false, null = train if DECODER_LR != 0)
+        highlight:
+          USE_FILM: False  # Enable FiLM conditioning in highlight decoder
+          FEATURE_DIM: 1024  # Feature dimension for highlight decoder
+          REASSEMBLE_OUT_CHANNELS: [96,192,384,768]  # Output channels for each decoder stage
+          REASSEMBLE_FACTORS: [4.0, 2.0, 1.0, 0.5]  # Spatial upsampling factors for each stage
+          READOUT_TYPE: "ignore"  # Readout type for DPT decoder
+          USE_BN: False  # Use batch normalization in decoder
+          DROPOUT: 0.1  # Dropout rate in decoder layers
+          OUTPUT_IMAGE_SIZE: [448,448]  # Output image resolution [height, width]
+          OUTPUT_CHANNELS: 1  # Number of output channels (1 for highlight mask)
+          DECODER_LR: 5.0e-4  # Custom learning rate for decoder (0.0 = frozen, 1.0 = same as base LR)
+          NUM_FUSION_BLOCKS_TRAINABLE: null  # Number of fusion blocks to train (0-4, null = train all if DECODER_LR != 0)
+      TOKEN_INPAINTER:
+        TOKEN_INPAINTER_CLASS: "TokenInpainter_Prior"  # Token inpainter class name
+        TOKEN_INPAINTER_MODULE: "token_inpainters"  # Module name to import token inpainter from
+        FROM_PRETRAINED: "weights/token_inpainter.pth"  # Path to pretrained token inpainter weights 
+        TOKEN_INPAINTER_LR: 1.0e-5  # Learning rate for token inpainter (can differ from base LR)
+        DEPTH: 6  # Number of transformer blocks
+        HEADS: 16  # Number of attention heads
+        DROP: 0 # Dropout rate 
+        USE_POSITIONAL_ENCODING: True  # Enable 2D sinusoidal positional encodings
+        USE_FINAL_NORM: True  # Enable final LayerNorm before output projection
+        USE_LOCAL_PRIOR: True  # Blend local mean prior for masked seeds
+        LOCAL_PRIOR_WEIGHT: 0.5  # Weight for local prior blending (1.0 = only mask_token, 0.0 = only local mean)
+        LOCAL_PRIOR_KERNEL: 5  # Kernel size for local prior blending (> 1)
+        SEED_NOISE_STD: 0.02  # Standard deviation of noise added to masked seeds during training  
+  INPAINT_MASK_DILATION:
+    value: 1  # Dilation kernel size (pixels) for inpaint mask - Must be odd
+  USE_TORCH_COMPILE:  # Enable PyTorch 2.0 torch.compile for faster training (experimental)
+    value: False
+
+  # ### DATA                                                                      
+  # DATASETS:
+  #   value:
+  #     SCRREAM:
+  #       VAL_SCENES: ["scene10_full_00","scene11_full_00","scene044_full_00","scene04_reduced_00","scene04_reduced_01","scene04_reduced_02"]  # List of validation scene names
+  #       TARGET_SIZE: [448,448]  # Target image size [height, width] in pixels
+  #       RESIZE_MODE: "resize+crop"  # Image resizing mode: "resize", "crop", "resize+crop", or "pad"
+  #       FEW_IMAGES: False  # If True, load only first 10 images per scene (for quick debugging)
+  #       SAMPLE_EVERY_N: 2  # Load every Nth frame from each scene (1 = all frames, 4 = every 4th frame)
+  #       LOAD_RGB_ONLY: True  # If True, ignore polarization data and load only RGB images
+        
+  #     HOUSECAT6D:
+  #       VAL_SCENES: ["val_scene1","val_scene2"]  # Validation scene names
+  #       TARGET_SIZE: [448,448]  # Target image size [height, width]
+  #       RESIZE_MODE: "resize+crop"  # Image resizing mode
+  #       FEW_IMAGES: False  # Load only first 10 images if True
+  #       SAMPLE_EVERY_N: 2  # Load every Nth frame
+  #       LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+  #     CROMO:
+  #       TRAIN_SCENES: ["kitchen"]  # Training scene names (list or string)
+  #       # VAL_SCENES: "station"  # Validation scene names (optional)
+  #       TARGET_SIZE: [448,448]  # Target image size [height, width]
+  #       RESIZE_MODE: "resize"  # Image resizing mode
+  #       FEW_IMAGES: False  # Load only first 10 images if True
+  #       SAMPLE_EVERY_N: 2  # Load every Nth frame
+  #       LOAD_RGB_ONLY: True  # Ignore polarization data if True
+    
+  #     PSD:
+  #       TRAIN_SCENES: "PSD_Train"  # Training scene name (string or list)
+  #       VAL_SCENES: "PSD_Val"  # Validation scene name (string or list)
+  #       TARGET_SIZE: [448,448]  # Target image size [height, width]
+  #       RESIZE_MODE: "resize+crop"  # Image resizing mode
+  #       FEW_IMAGES: False  # Load only first 10 images if True
+  #       SAMPLE_EVERY_N: 1  # Load every Nth frame (1 = all frames)
+  #       LOAD_RGB_ONLY: True  # Ignore polarization data if True
+
+  #     SCARED:
+  #       VAL_SCENES: ["v22","v23","v24","v25","v26","v27","v28","v29","v30","v31","v32","v33","v34"]  # Validation scene names
+  #       TARGET_SIZE: [448,448]  # Target image size [height, width]
+  #       RESIZE_MODE: "resize" #"resize+crop"  # Image resizing mode
+  #       SAMPLE_EVERY_N: 4  # Load every Nth frame
+  #       LOAD_RGB_ONLY: True  # Ignore polarization data if True
+  #       FEW_IMAGES: False  # Load only first 10 images if True
+  #       HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing in dataset
+  #       HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection (0-1)
+  #       HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+  #       HIGHLIGHT_RECT_SIZE: [1000, 1000]  # Size of highlight rectangle region [height, width]
+  #       HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+  #       HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+      
+  #     STEREOMIS_TRACKING:
+  #       VAL_SCENES: ["P2_2"]  # Validation scene names
+  #       TARGET_SIZE: [448,448]  # Target image size [height, width]
+  #       RESIZE_MODE: "resize+crop"  # Image resizing mode
+  #       SAMPLE_EVERY_N: 4  # Load every Nth frame
+  #       LOAD_RGB_ONLY: True  # Ignore polarization data if True
+  #       FEW_IMAGES: False  # Load only first 10 images if True
+  #       HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+  #       HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+  #       HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+  #       HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+  #       HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+  #       HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+  #     CHOLEC80:
+  #       VAL_SCENES: ["val"]  # Validation scene names
+  #       TARGET_SIZE: [448,448]  # Target image size [height, width]
+  #       RESIZE_MODE: "resize+crop"  # Image resizing mode
+  #       SAMPLE_EVERY_N: 10  # Load every Nth frame
+  #       LOAD_RGB_ONLY: True  # Ignore polarization data if True
+  #       FEW_IMAGES: False  # Load only first 10 images if True
+  #       HIGHLIGHT_ENABLE: False  # Enable highlight detection/processing
+  #       HIGHLIGHT_BRIGHTNESS_THRESHOLD: 0.9  # Brightness threshold for highlight detection
+  #       HIGHLIGHT_RETURN_MASK: True  # Return highlight mask in dataset output
+  #       HIGHLIGHT_RECT_SIZE: [800, 800]  # Size of highlight rectangle region
+  #       HIGHLIGHT_RETURN_RECT_AS_RGB: False  # Return highlight rectangle as RGB if True
+  #       HIGHLIGHT_RETURN_RECT: True  # Return highlight rectangle region if True
+
+  #     # POLARGB:
+  #     #   TRAIN_SCENES: "train"
+  #     #   VAL_SCENES: "test"
+  #     #   TARGET_SIZE: [448,448]
+  #     #   RESIZE_MODE: "resize+crop"
+  #     #   SAMPLE_EVERY_N: 1
+  #     #   LOAD_RGB_ONLY: True
+
+  # BATCH_SIZE: # Max batch size with img size 448 is 32
+  #   value: 16 # Number of samples per batch (adjust based on GPU memory)
+  # NUM_WORKERS:
+  #   value: 8  # Number of data loading worker processes (0 = main process only)
+  # SHUFFLE:
+  #   value: True  # Shuffle training data each epoch (False for validation/test)
+  # PIN_MEMORY:
+  #   value: True  # Pin memory in DataLoader for faster GPU transfer (recommended: True)
+  # PREFETCH_FACTOR:
+  #   value: 2  # Number of batches to prefetch per worker (higher = more memory usage)
+  
+  # ### HIGHLIGHTS                                                        
+  # MOGE_MODEL:
+  #   value: "Ruicheng/moge-2-vits-normal"  # MoGe model name for normal estimation (HuggingFace format)
+  # SURFACE_ROUGHNESS:
+  #   value: 8.0  # Blinn-Phong surface roughness exponent (higher = sharper highlights)
+  # INTENSITY:
+  #   value: 2.0  # Specular highlight intensity multiplier
+  # LIGHT_DISTANCE_RANGE:
+  #   value: [0.0, 1]  # Range for light source distance sampling [min, max] (normalized)
+  # LIGHT_LEFT_RIGHT_ANGLE:
+  #   value: [0, 360]  # Range for light source horizontal angle [min, max] in degrees
+  # LIGHT_ABOVE_BELOW_ANGLE:
+  #   value: [0, 360]  # Range for light source vertical angle [min, max] in degrees
+  # DATASET_HIGHLIGHT_DILATION:
+  #   value: 25  # Dilation kernel size (pixels) for dataset highlight masks
+  # DATASET_HIGHLIGHT_THRESHOLD:
+  #   value: 0.9  # Brightness/luminance threshold (0-1) for detecting highlights in dataset images
+  # DATASET_HIGHLIGHT_USE_LUMINANCE:
+  #   value: True  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B) for dataset highlights; if False, use simple mean brightness
+  # HIGHLIGHT_COLOR:
+  #   value: [1.0, 1.0, 1.0]  # RGB color for synthetic highlights (normalized 0-1)
+  # CLAMP_RECONSTRUCTION:
+  #   value: True  # Clamp reconstructed images to [0, 1] range if True
+
+  # ### OPTIMIZATION                                                                      
+  # LEARNING_RATE:
+  #   value: 1.0e-3  # Base learning rate for optimizer
+  # WEIGHT_DECAY:   
+  #   value: 0.0  # L2 regularization weight (0.0 = no weight decay)
+  # EPOCHS:
+  #   value: 25  # Maximum number of training epochs
+  # GRADIENT_ACCUMULATION_STEPS:
+  #   value: 1  # Number of steps to accumulate gradients before optimizer step (1 = no accumulation)
+  # WARMUP:
+  #   value: 200  # Number of warmup steps for learning rate schedule (linear warmup from 0 to LR)
+  # GRADIENT_CLIPPING_MAX_NORM:
+  #   value: 8  # Maximum gradient norm for clipping (set to -1 to disable clipping)
+  # LR_SCHEDULER: 
+  #   value: 
+  #     ONPLATEAU:  # ReduceLROnPlateau scheduler (reduces LR when validation metric plateaus)
+  #       PATIENCE: 5  # Number of epochs to wait before reducing LR
+  #       FACTOR: 0.1  # Factor by which LR is reduced (new_lr = old_lr * factor)
+  #     COSINE:  # CosineAnnealingLR scheduler (cosine annealing schedule)
+  #       N_PERIODS: 1  # Number of cosine periods over training
+  #     # STEPWISE:  # StepLR scheduler (reduces LR at fixed step intervals)
+  #     #   N_STEPS: 4  # Number of times to reduce LR during training
+  #     #   GAMMA: 0.5  # Factor by which LR is reduced at each step (new_lr = old_lr * gamma)
+  #     # EXPONENTIAL:  # ExponentialLR scheduler (exponential decay)
+  #     #   GAMMA: 0.5  # Multiplicative factor for exponential decay
+
+  # SWITCH_OPTIMIZER_EPOCH:
+  #   value: null  # Epoch number to switch from bootstrap to refining optimizer (null = no switch)
+  # OPTIMIZER_BOOTSTRAP_NAME:
+  #   value: "AdamW"  # Optimizer name for initial training phase ("Adam", "SGD", etc.)
+  # OPTIMIZER_REFINING_NAME:
+  #   value: "AdamW"  # Optimizer name for refining phase (used after SWITCH_OPTIMIZER_EPOCH)
+  # EARLY_STOPPING_PATIENCE:
+  #   value: 10  # Number of epochs without improvement before stopping training
+  # SAVE_INTERVAL:
+  #   value: 1000  # Number of training steps between model checkpoints
+  
+  # DATASET_HIGHLIGHT_SUPERVISION_THRESHOLD:
+  #   value: 0.1  # Pixel highlights above this threshold (should be low) are excluded from supervision
+
+  # ### LOSS WEIGHTS (relative to the total loss, NOT NORMALIZED LATER)                                                                              
+  # SPECULAR_LOSS_WEIGHT:
+  #   value: 0.0  # Weight for specular component reconstruction loss
+  # DIFFUSE_LOSS_WEIGHT:
+  #   value: 1.0  # Weight for diffuse component reconstruction loss
+  # HIGHLIGHT_LOSS_WEIGHT:
+  #   value: 1.0  # Weight for highlight mask regression loss
+  # IMAGE_RECONSTRUCTION_LOSS_WEIGHT:
+  #   value: 0.0  # Weight for full image reconstruction loss
+  # SATURATION_RING_LOSS_WEIGHT:
+  #   value: 0.0  # Weight for saturation ring consistency loss (around highlight regions)
+  # RING_KERNEL_SIZE:
+  #   value: 11  # Kernel size (odd number) for saturation ring dilation around highlights
+  # RING_VAR_WEIGHT:
+  #   value: 0.5  # Weight for variance matching in saturation ring loss (vs mean matching)
+  # RING_TEXTURE_WEIGHT:
+  #   value: 1.0  # Weight for texture consistency term in saturation ring loss
+  # HLREG_W_L1:
+  #   value: 1.0  # Weight for L1 loss in highlight regression
+  # HLREG_USE_CHARB:
+  #   value: True  # Use Charbonnier loss (smooth L1) instead of standard L1 if True
+  # HLREG_W_DICE:
+  #   value: 0.2  # Weight for Dice loss in highlight regression (for mask overlap)
+  # HLREG_W_SSIM:
+  #   value: 0.0  # Weight for SSIM loss in highlight regression
+  # HLREG_W_GRAD:
+  #   value: 0.0  # Weight for gradient loss in highlight regression
+  # HLREG_W_TV:
+  #   value: 0.0  # Weight for total variation loss in highlight regression
+  # HLREG_BALANCE_MODE:
+  #   value: "auto"   # Class balancing mode for highlight regression: 'none' | 'auto' | 'pos_weight'
+  # HLREG_POS_WEIGHT:
+  #   value: 1.0      # Positive class weight (used only if BALANCE_MODE == 'pos_weight')
+  # HLREG_FOCAL_GAMMA:
+  #   value: 2.0      # Focal loss gamma parameter (0.0 = standard BCE, 1.0-2.0 helps with gradient vanishing)
+  
+  # WEIGHT_TOKEN_INPAINT:
+  #   value: 1.0  # Weight for token-space inpainting loss (L1 + cosine similarity in feature space)
+  # WEIGHT_CONTEXT_IDENTITY: 
+  #   value: 0.0  # LEAVE TO 0.0: Weight for L1 loss on context (non-masked) regions (identity preservation) 
+  # WEIGHT_TV_IN_HOLE:
+  #   value: 0.0  # LEAVE TO 0.0: Weight for total variation loss inside masked/hole regions
+  # RING_DILATE_KERNEL:
+  #   value: 17  # Dilation kernel size (odd number) for creating ring mask around highlights
+  # WEIGHT_SEAM:
+  #   value: 0.5  # Weight for gradient matching loss on saturation ring
+  # SEAM_USE_CHARB:
+  #   value: True  # Use Charbonnier loss instead of L1 in seam loss (smooth L1 for boundary consistency)
+  # SEAM_WEIGHT_GRAD:
+  #   value: 0.0  # Weight for gradient matching term inside seam loss (0.0 = disable gradient term)
+  # TOKEN_FEAT_ALPHA:
+  #   value: 0.5  # Mixing factor for token feature loss: alpha * L1 + (1-alpha) * (1-cosine_sim)
+
+  # ### DIFFUSE HIGHLIGHT PENALTY
+  # WEIGHT_DIFFUSE_HIGHLIGHT_PENALTY:
+  #   value: 0.1  # Weight for penalty loss on highlights in diffuse decoder output (0.0 = disabled)
+  # DIFFUSE_HL_THRESHOLD:
+  #   value: 0.8  # Brightness/luminance threshold for detecting highlights in diffuse (0.0-1.0)
+  # DIFFUSE_HL_USE_CHARB:
+  #   value: True  # Use Charbonnier loss instead of L1 for diffuse highlight penalty
+  # DIFFUSE_HL_PENALTY_MODE:
+  #   value: "brightness"  # Penalty mode: "brightness" (penalize brightness/luminance above threshold) or "pixel" (penalize RGB values directly)
+  # DIFFUSE_HL_TARGET_BRIGHTNESS:
+  #   value: null  # Target brightness/luminance for penalized pixels (null = use threshold value)
+  # DIFFUSE_HL_USE_LUMINANCE:
+  #   value: False  # If True, use perceptually-weighted luminance (0.299*R + 0.587*G + 0.114*B); if False, use simple mean brightness
+
+  # ### LOGGING, RESULTS AND WANDB                                                                 
+  # LOG_INTERVAL:
+  #   value: 1  # Number of training steps between console log outputs
+  # WANDB_LOG_INTERVAL:
+  #   value: 1  # Number of training steps between WandB metric logs
+  # IMAGE_LOG_INTERVAL:
+  #   value: 10  # Number of training steps between image logging to WandB
+  # NO_WANDB:
+  #   value: False  # Disable WandB logging if True (useful for local debugging)
+  # MODEL_WATCHER_FREQ_WANDB:
+  #   value: 50  # Frequency (in steps) for logging model parameter histograms to WandB
+  # WANDB_ENTITY:
+  #   value: "unreflect-anything"  # WandB organization/entity name
+  # WANDB_PROJECT:
+  #   value: "UnReflectAnything"  # WandB project name
+  # NOTES:
+  #   value: "448 Test"  # Notes/description for this training run