| Vijay Vasudevan | 2c3738d | 2015-12-08 14:55:13 -0800 | [diff] [blame] | 1 | # Release 0.6.0 |
| 2 | |
| 3 | ## Major Features and Improvements |
| 4 | |
| 5 | * Python 3.3+ support via changes to python codebase and ability |
| 6 | to specify python version via ./configure. |
| 7 | |
| 8 | * Some improvements to GPU performance and memory usage: |
| 9 | [convnet benchmarks](https://github.com/soumith/convnet-benchmarks/issues/66) |
| 10 | roughly equivalent with native cudnn v2 performance. Improvements mostly due |
| 11 | to moving to 32-bit indices, faster shuffling kernels. More improvements to |
| 12 | come in later releases. |
| 13 | |
| 14 | |
| 15 | ## Bug fixes |
| 16 | |
| 17 | * Lots of fixes to documentation and tutorials, many contributed |
| 18 | by the public. |
| 19 | |
| 20 | * 271 closed issues on github issues. |
| 21 | |
| 22 | ## Backwards-incompatible changes |
| 23 | |
| 24 | * tf.nn.fixed_unigram_candidate_sampler changed its default 'distortion' |
| 25 | attribute from 0.0 to 1.0. This was a bug in the original release |
| 26 | that is now fixed. |
| 27 | |
| Vijay Vasudevan | ddd4aaf | 2015-12-08 09:58:59 -0800 | [diff] [blame] | 28 | # Release 0.5.0 |
| 29 | |
| 30 | Initial release of TensorFlow. |