* Fix for #63

examples/classification/IMDbSparseAbsorbingTextCategorizationDemo.py

@@ -89,7 +89,7 @@
 
     for e in range(60):
         start_training = time()
-        tm.fit(X_train, Y_train)
+        tm.fit(X_train.astype(np.uint32), Y_train)
         stop_training = time()
 
         start_testing = time()

tmu/models/classification/vanilla_classifier.py

@@ -371,7 +371,7 @@ def fit(
             **kwargs
     ):
         metrics = metrics or []
-        assert len(X) == len(Y), "X and Y must have the same length"
+        assert X.shape[0] == len(Y), "X and Y must have the same number of samples"
         assert len(X.shape) >= 2, "X must be a 2D array"
         assert len(Y.shape) == 1, "Y must be a 1D array"
         assert X.dtype == np.uint32, "X must be of type uint32"

tmu/util/encoded_data_cache.py

@@ -1,6 +1,8 @@
 from typing import Optional
 import xxhash
 import numpy as np
+from scipy.sparse import issparse
+
 
 class DataEncoderCache:
     def __init__(self, seed: int):
@@ -9,9 +11,28 @@ def __init__(self, seed: int):
         self.array_hash: Optional[str] = None
         self.encoded_data: Optional[np.ndarray] = None
 
-    def compute_hash(self, arr: np.ndarray) -> str:
-        """Compute a hash for a numpy array."""
-        return xxhash.xxh3_64_hexdigest(arr)
+    def compute_hash_csr_matrix(self, csr_mat):
+        # Convert the components of the csr_matrix to bytes
+        data_bytes = csr_mat.data.tobytes()
+        indices_bytes = csr_mat.indices.tobytes()
+        indptr_bytes = csr_mat.indptr.tobytes()
+
+        # Concatenate the bytes representations
+        total_bytes = data_bytes + indices_bytes + indptr_bytes
+
+        # Compute the hash on the concatenated bytes
+        hash_value = xxhash.xxh3_64_hexdigest(total_bytes)
+
+        return hash_value
+
+    def compute_hash(self, arr):
+        """Compute a hash for a numpy array or csr_matrix."""
+        if issparse(arr):
+            # It's a sparse matrix, handle specially
+            return self.compute_hash_csr_matrix(arr)
+        else:
+            # It's a dense array, proceed as before
+            return xxhash.xxh3_64_hexdigest(arr.tobytes())
 
     def get_encoded_data(self, data: np.ndarray, encoder_func) -> np.ndarray:
         """Get encoded data for an array, using cache if available."""