| Vijay Vasudevan | 10e62dc | 2015-12-11 23:03:16 -0800 | [diff] [blame] | 1 | # Changes since last release |
| 2 | |
| 3 | ## Breaking changes to the API |
| 4 | |
| 5 | * `AdjustContrast` kernel deprecated, new kernel `AdjustContrastv2` takes and |
| 6 | outputs float only. `adjust_contrast` now takes all data types. |
| 7 | * `adjust_brightness`'s `delta` argument is now always assumed to be in `[0,1]` |
| 8 | (as is the norm for images in floating point formats), independent of the |
| 9 | data type of the input image. |
| 10 | * The image processing ops do not take `min` and `max` inputs any more, casting |
| 11 | safety is handled by `saturate_cast`, which makes sure over- and underflows |
| 12 | are handled before casting to data types with smaller ranges. |
| Geoffrey Irving | cbff45c | 2016-01-12 08:06:56 -0800 | [diff] [blame] | 13 | * For C++ API users: `IsLegacyScalar` and `IsLegacyVector` are now gone from |
| 14 | `TensorShapeUtils` since TensorFlow is scalar strict within Google (for |
| 15 | example, the shape argument to `tf.reshape` can't be a scalar anymore). The |
| 16 | open source release was already scalar strict, so outside Google `IsScalar` |
| 17 | and `IsVector` are exact replacements. |
| Josh Levenberg | db7478e | 2016-01-20 14:54:50 -0800 | [diff] [blame] | 18 | * The following files are being removed from `tensorflow/core/public/`: |
| 19 | * `env.h` -> `../platform/env.h` |
| 20 | * `status.h` -> `../lib/core/status.h` |
| 21 | * `tensor.h` -> `../framework/tensor.h` |
| 22 | * `tensor_shape.h` -> `../framework/tensor_shape.h` |
| 23 | * `partial_tensor_shape.h` -> `../framework/partial_tensor_shape.h` |
| 24 | * `tensorflow_server.h` deleted |
| Geoffrey Irving | 5643775 | 2016-01-25 09:43:13 -0800 | [diff] [blame] | 25 | * For C++ API users: `TensorShape::ShortDebugString` has been renamed to |
| 26 | `DebugString`, and the previous `DebugString` behavior is gone (it was |
| 27 | needlessly verbose and produced a confusing empty string for scalars). |
| Manjunath Kudlur | c2722a1 | 2016-01-27 13:24:50 -0800 | [diff] [blame^] | 28 | * `GraphOptions.skip_common_subexpression_elimination` has been removed. All |
| 29 | graph optimizer options are now specified via |
| 30 | `GraphOptions.OptimizerOptions`. |
| Vijay Vasudevan | 10e62dc | 2015-12-11 23:03:16 -0800 | [diff] [blame] | 31 | |
| Josh Levenberg | 02dff6d | 2016-01-07 18:37:54 -0800 | [diff] [blame] | 32 | ## Bug fixes |
| 33 | |
| 34 | * The Python API will now properly set the `list` member of `AttrValue` in |
| 35 | constructed `GraphDef` messages for empty lists. The serialization of some |
| 36 | graphs will change, but the change is both forwards and backwards compatible. |
| 37 | It will break tests that compare a generated `GraphDef` to a golden serialized |
| 38 | `GraphDef`. |
| 39 | |
| Geoffrey Irving | cbff45c | 2016-01-12 08:06:56 -0800 | [diff] [blame] | 40 | |
| Vijay Vasudevan | 2c3738d | 2015-12-08 14:55:13 -0800 | [diff] [blame] | 41 | # Release 0.6.0 |
| 42 | |
| 43 | ## Major Features and Improvements |
| 44 | |
| 45 | * Python 3.3+ support via changes to python codebase and ability |
| 46 | to specify python version via ./configure. |
| 47 | |
| 48 | * Some improvements to GPU performance and memory usage: |
| 49 | [convnet benchmarks](https://github.com/soumith/convnet-benchmarks/issues/66) |
| 50 | roughly equivalent with native cudnn v2 performance. Improvements mostly due |
| 51 | to moving to 32-bit indices, faster shuffling kernels. More improvements to |
| 52 | come in later releases. |
| 53 | |
| 54 | |
| 55 | ## Bug fixes |
| 56 | |
| 57 | * Lots of fixes to documentation and tutorials, many contributed |
| 58 | by the public. |
| 59 | |
| 60 | * 271 closed issues on github issues. |
| 61 | |
| 62 | ## Backwards-incompatible changes |
| 63 | |
| 64 | * tf.nn.fixed_unigram_candidate_sampler changed its default 'distortion' |
| 65 | attribute from 0.0 to 1.0. This was a bug in the original release |
| 66 | that is now fixed. |
| 67 | |
| Vijay Vasudevan | ddd4aaf | 2015-12-08 09:58:59 -0800 | [diff] [blame] | 68 | # Release 0.5.0 |
| 69 | |
| 70 | Initial release of TensorFlow. |