Track information about the batch size within the iteration. Parameters ---------- size : int The size of the detected batch size to be maintained in this iteration source_context Source information about where the batch size was detected for
(self, size, source_context, non_uniform_batch=False)
| 274 | |
| 275 | class _IterBatchInfo: |
| 276 | def __init__(self, size, source_context, non_uniform_batch=False): |
| 277 | """Track information about the batch size within the iteration. |
| 278 | |
| 279 | Parameters |
| 280 | ---------- |
| 281 | size : int |
| 282 | The size of the detected batch size to be maintained in this iteration |
| 283 | source_context |
| 284 | Source information about where the batch size was detected for better error reporting. |
| 285 | non_uniform_batch : bool, optional |
| 286 | Should be set to True if pure Split/Merge are used by hand and not as an implementation |
| 287 | of conditional operation. In that case there can be varying batch sizes within |
| 288 | the pipeline and we don't know how to track them, by default False |
| 289 | """ |
| 290 | self._size = size |
| 291 | self._source_context = source_context |
| 292 | self._non_uniform_batch = non_uniform_batch |
| 293 | |
| 294 | @property |
| 295 | def size(self): |