Update README.md

README.md

@@ -261,9 +261,6 @@ with tempfile.TemporaryDirectory() as temp_dir:
 from collections import Counter
 import matplotlib.pyplot as plt
 
-# Load dataset
-dataset = load_dataset("xixu-me/fsl-product-classification")["train"]
-
 # Analyze class distribution
 class_counts = Counter(dataset['class_name'])
 print(f"Number of classes: {len(class_counts)}")
@@ -314,7 +311,6 @@ import random
 from collections import defaultdict
 import torch
 from torch.utils.data import DataLoader
-from datasets import load_dataset
 
 def create_few_shot_split(dataset, n_way=5, k_shot=5, n_query=15, seed=None):
     """
@@ -369,9 +365,6 @@ def create_few_shot_split(dataset, n_way=5, k_shot=5, n_query=15, seed=None):
     
     return support_set, query_set
 
-# Example usage
-dataset = load_dataset("xixu-me/fsl-product-classification")["train"]
-
 # Create a 5-way 5-shot episode
 support_set, query_set = create_few_shot_split(dataset, n_way=5, k_shot=5, n_query=15)
 
@@ -480,8 +473,7 @@ transform = transforms.Compose([
 ])
 
 # Load dataset
-hf_dataset = load_dataset("xixu-me/fsl-product-classification")["train"]
-fsl_dataset = FSLProductDataset(hf_dataset, transform=transform)
+fsl_dataset = FSLProductDataset(dataset, transform=transform)
 
 # Create episode data
 support_data, query_data = fsl_dataset.create_episode_dataloader(
@@ -752,7 +744,6 @@ def create_cross_domain_split(dataset, train_ratio=0.6, val_ratio=0.2, test_rati
     }
 
 # Create cross-domain splits
-dataset = load_dataset("xixu-me/fsl-product-classification")["train"]
 splits = create_cross_domain_split(dataset)
 
 print(f"Train classes: {len(splits['train']['classes'])}")
@@ -988,60 +979,6 @@ def create_data_splits(dataset, split_ratios={'train': 0.8, 'test': 0.2},
         print(f"{split_name.capitalize()} split: {n_samples} samples, {n_classes} classes")
     
     return splits
-
-# Example usage of utility functions
-def analyze_dataset(dataset_path="data.tzst"):
-    """
-    Complete dataset analysis workflow
-    """
-    print("🔍 Analyzing FSL Product Classification Dataset")
-    print("=" * 50)
-    
-    # 1. Verify dataset integrity
-    print("\n1. Verifying dataset integrity...")
-    is_valid = verify_dataset_integrity(dataset_path)
-    
-    if not is_valid:
-        return
-    
-    # 2. Load dataset
-    print("\n2. Loading dataset...")
-    try:
-        dataset = load_dataset("xixu-me/fsl-product-classification")["train"]
-        print(f"✅ Successfully loaded dataset with {len(dataset)} samples")
-    except Exception as e:
-        print(f"❌ Error loading dataset: {e}")
-        return
-    
-    # 3. Generate statistics
-    print("\n3. Generating statistics...")
-    stats = dataset_statistics(dataset)
-    
-    print(f"📊 Dataset Statistics:")
-    print(f"   Total samples: {stats['total_samples']:,}")
-    print(f"   Total classes: {stats['total_classes']:,}")
-    print(f"   Avg samples per class: {stats['avg_samples_per_class']:.1f}")
-    print(f"   Min samples per class: {stats['min_samples_per_class']}")
-    print(f"   Max samples per class: {stats['max_samples_per_class']}")
-    print(f"   Std samples per class: {stats['std_samples_per_class']:.1f}")
-    
-    # 4. Plot distributions
-    print("\n4. Plotting class distribution...")
-    plot_class_distribution(dataset, top_k=30)
-    
-    # 5. Save dataset info
-    print("\n5. Saving dataset information...")
-    save_dataset_info(dataset)
-    
-    # 6. Create splits
-    print("\n6. Creating data splits...")
-    splits = create_data_splits(dataset, strategy='stratified')
-    
-    print("\n✅ Dataset analysis complete!")
-    return dataset, stats, splits
-
-# Run analysis
-# dataset, stats, splits = analyze_dataset()
 ```
 
 ## Troubleshooting
@@ -1069,26 +1006,7 @@ from tzst import extract_archive
 extract_archive("data.tzst", "output/", streaming=True)
 ```
 
-#### 2. Memory Issues with Large Dataset
-
-**Problem**: Out of memory when loading the full dataset
-
-**Solution**:
-
-```python
-# Use streaming dataset
-from datasets import load_dataset
-dataset = load_dataset("xixu-me/fsl-product-classification", streaming=True)
-
-# Or load in chunks
-def load_dataset_chunked(chunk_size=1000):
-    dataset = load_dataset("xixu-me/fsl-product-classification")["train"]
-    for i in range(0, len(dataset), chunk_size):
-        chunk = dataset.select(range(i, min(i + chunk_size, len(dataset))))
-        yield chunk
-```
-
-#### 3. Non-continuous Class Labels
+#### 2. Non-continuous Class Labels
 
 **Problem**: Class labels are not continuous (0, 1, 2, ...)
 
@@ -1107,7 +1025,7 @@ def map_labels(example):
 dataset = dataset.map(map_labels)
 ```
 
-#### 4. CUDA/GPU Issues
+#### 3. CUDA/GPU Issues
 
 **Problem**: CUDA out of memory during training