Update README.md

README.md

@@ -9,15 +9,15 @@ This dataset contains time series data paired with visual representations and na
 ### Dataset Statistics
 
 - **Level 1 (Basic)**: Single series analysis (min/max detection, trend analysis)
-  - Training samples: ~27,000
+  - Training samples: 54,000
   - Test samples: Multiple variants (minmax, multiseries, startend, subseries)
 
 - **Level 2 (Intermediate)**: Multi-series analysis and relationships
-  - Training samples: ~40,000+
+  - Training samples: 45,632
   - Test categories: global, local, numerical
 
 - **Level 3 (Advanced)**: Complex reasoning and annotations
-  - Training samples: ~1,000+
+  - Training samples: 3,515
   - Test samples: Final test set
 
 ## Data Structure
@@ -125,9 +125,9 @@ The `images` field in each sample contains URLs to visual representations of the
 ```
 AAA-HiTSR/
 ├── Train/
-│   ├── l1_train_llatisa.json      (Level 1: 27,000 samples)
-│   ├── l2_train_llatisa.json      (Level 2: 40,000+ samples)
-│   └── l3_train_llatisa.json      (Level 3: 1,000+ samples)
+│   ├── l1_train_llatisa.json      (Level 1: 54,000 samples)
+│   ├── l2_train_llatisa.json      (Level 2: 45,000+ samples)
+│   └── l3_train_llatisa.json      (Level 3: 3,500+ samples)
 └── Test/
     ├── l1_test_startend.json
     ├── l1_test_minmax.json