Add verbose flag to formatting methods and clean up comments

Author: AllenBaranov

sigllm/primitives/formatting/__init__.py

@@ -18,3 +18,4 @@
     'DigitInterleave',
 ]
 
+

sigllm/primitives/formatting/digit_interleave.py

@@ -23,7 +23,7 @@ def interleave_digits(timestamp):
             return result_str
 
         result = [
-            separator.join(interleave_digits(timestamp) for timestamp in window) + separator  # Add comma at the end
+            separator.join(interleave_digits(timestamp) for timestamp in window) + separator
             for window in data
         ]
         return result
@@ -37,13 +37,10 @@ def deinterleave_timestamp(interleaved_str):
             total_digits = len(interleaved_str)
             num_values = total_digits // width_used
 
-            # Reconstruct each original value
             values = []
             for value_idx in range(num_values):
-                # Collect digits for this value from each position
                 value_digits = []
                 for digit_pos in range(width_used):
-                    # Calculate position in interleaved string
                     pos = digit_pos * num_values + value_idx
                     if pos < total_digits:
                         value_digits.append(interleaved_str[pos])
@@ -58,7 +55,7 @@ def deinterleave_timestamp(interleaved_str):
                 deinterleave_timestamp(timestamp)
                 for sample in entry
                 for timestamp in sample.lstrip(separator).rstrip(separator).split(separator)[:trunc]
-                if timestamp.strip()  # Skip empty strings
+                if timestamp.strip()
             ]
             for entry in data
         ], dtype=object)

sigllm/primitives/formatting/multivariate_formatting.py

@@ -136,7 +136,7 @@ def test_multivariate_formatting_validity(self, data=None, verbose=False):
                 print(data.shape)
 
         string_data = self.format_as_string(data, **self.config)
-        LLM_mock_output = np.array(string_data).reshape(-1, 1) # pretend only one sample
+        LLM_mock_output = np.array(string_data).reshape(-1, 1)
         if verbose:
             print(LLM_mock_output)
         integer_data = self.format_as_integer(LLM_mock_output, **self.config)

sigllm/primitives/formatting/value_interleave.py

@@ -8,13 +8,12 @@ def __init__(self, verbose: bool = False, **kwargs):
 
 
     def format_as_string(self, data: np.ndarray, digits_per_timestamp = 3, separator = ",") -> str:
-        #print(data[0])
         max_digits = max(len(str(abs(int(v)))) for window in data for ts in window for v in ts)
         width_used = max(digits_per_timestamp, max_digits)
         self.metadata['width_used'] = width_used    
         result = [
             separator.join(''.join(str(int(val)).zfill(width_used)[:width_used] for val in timestamp)
-            for timestamp in window) + separator  # Add comma at the end
+            for timestamp in window) + separator
             for window in data
         ]
         return result