Commit message (Collapse) | Author | Age | |
---|---|---|---|
* | Modularize TF Lite interface definitions and reorganize file structure | Pete Warden | 2018-09-07 |
| | | | | PiperOrigin-RevId: 212064501 | ||
* | Consolidate refactoring of runtime shapes. | A. Unique TensorFlower | 2018-08-30 |
| | | | | PiperOrigin-RevId: 210945714 | ||
* | SpaceToBatchND should pad with zero_point when inference_type is uint8 | Suharsh Sivakumar | 2018-07-26 |
| | | | | PiperOrigin-RevId: 206265356 | ||
* | Enhance error reporting. | A. Unique TensorFlower | 2018-05-21 |
| | | | | PiperOrigin-RevId: 197390052 | ||
* | Making GetInput from kernel_util.h return a pointer to const data. | A. Unique TensorFlower | 2018-05-11 |
| | | | | PiperOrigin-RevId: 196340200 | ||
* | Remove redundant calls to realloc dynamic tensors. | A. Unique TensorFlower | 2018-02-06 |
| | | | | PiperOrigin-RevId: 184682942 | ||
* | Make TFLite SpaceToBatchND op have parity with TF SpaceToBatchND op. | Nupur Garg | 2018-01-29 |
| | | | | PiperOrigin-RevId: 183734695 | ||
* | Support SpaceToBatchND in TFLite. | Nupur Garg | 2018-01-02 |
The internal implementation only supports 4D tensors for now. The dimension has to be 1 batch + 2 spatial + 1 other. The most common format within this restriction is NHWC. PiperOrigin-RevId: 180592870 |