Florence-2 coordinates model - 6 epochs, val_loss=3.8039

README.md

@@ -27,9 +27,9 @@ print(coords)
 ## Training Details
 - **Base Model**: microsoft/Florence-2-base
 - **Dataset**: 74233 captcha images with click coordinates
-- **Training**: LoRA fine-tuning, 3 epochs
-- **Best Val Loss**: 3.8446
-- **Training Time**: 25.8 minutes
+- **Training**: LoRA fine-tuning, 6 epochs
+- **Best Val Loss**: 3.8039
+- **Training Time**: 50.0 minutes
 - **GPU**: Nvidia A10G 24GB (Modal.com)
 
 ## Coordinate Format

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5570d735b2c3d13a4e20dcaf27f87c8d256f2935fa1e72d51b548d27fba5b911
+oid sha256:9f9395960805d6717c4304e20840592a2e62d09ec38a59a8d301a8cb1d6d5779
 size 463326048

training_config.json

@@ -1,7 +1,7 @@
 {
   "base_model": "microsoft/Florence-2-base",
   "task_prompt": "<COORDINATE>",
-  "num_epochs": 3,
+  "num_epochs": 6,
   "learning_rate": 1e-06,
   "lora_r": 8,
   "lora_alpha": 16,
@@ -9,8 +9,8 @@
   "grad_accum_steps": 1,
   "train_samples": 70521,
   "val_samples": 3712,
-  "best_val_loss": 3.844581150239514,
-  "total_training_time_min": 25.7799010515213,
+  "best_val_loss": 3.80387744749746,
+  "total_training_time_min": 49.95302717288335,
   "coordinate_format": "Normalized 0-999 using Florence-2 <loc_XXX> tokens",
   "usage": {
     "prompt": "<COORDINATE>",