Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly
Sign up
Appearance settings

Commit e01ffcd

Browse files
1.0.0-beta6 release notes, round 1
Signed-off-by: AlexDBlack <blacka101@gmail.com>
1 parent 2f0e815 commit e01ffcd

File tree

1 file changed

+145
-0
lines changed

1 file changed

+145
-0
lines changed

‎release-notes.md

Lines changed: 145 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -7,6 +7,13 @@ redirect_from: "/releasenotes"
77
---
88

99
**Contents**
10+
* <a href="#onezerozerobeta6">Version 1.0.0-beta5</a>
11+
- <a href="#onezerozerobeta6-dl4j">Deeplearning4j and DL4J Keras Import</a>
12+
- <a href="#onezerozerobeta6-nd4j">ND4J and SameDiff</a>
13+
- <a href="#onezerozerobeta6-datavec">DataVec</a>
14+
- <a href="#onezerozerobeta6-rl4j">RL4J</a>
15+
- <a href="#onezerozerobeta6-arbiter">Arbiter</a>
16+
- <a href="#onezerozerobeta6-nd4s">ND4S</a>
1017
* <a href="#onezerozerobeta5">Version 1.0.0-beta5</a>
1118
- <a href="#onezerozerobeta5-dl4j">Deeplearning4j and DL4J Keras Import</a>
1219
- <a href="#onezerozerobeta5-nd4j">ND4J and SameDiff</a>
@@ -63,6 +70,144 @@ redirect_from: "/releasenotes"
6370
* <a href="#five">Version 0.5.0</a>
6471
* <a href="#four">Version 0.4.0</a>
6572

73+
74+
# <a name="onezerozerobeta6">Release Notes for Version 1.0.0-beta6</a>
75+
76+
## Highlights - 1.0.0-beta6 Release
77+
78+
* SameDiff optimizations
79+
* Deeplearning4j UI - Play framework replaced with Vertx; deeplearning4j-ui dependency now no longer has Scala dependency or Scala version suffix
80+
* OpenMP replaced with thread pool c++ parallelism framework; enabled c++ parallelism for platforms without threading
81+
* ND4J namespaces
82+
83+
84+
85+
## <a name="onezerozerobeta5-dl4j">Deeplearning4J</a>
86+
87+
### Deeplearning4J: Features and Enhancements
88+
89+
* DNNL (MKL-DNN) upgraded to version 1.1
90+
91+
### Deeplearning4J: Bug Fixes and Optimizations
92+
93+
* KDTree implementation optimized [Link](https://github.com/KonduitAI/deeplearning4j/pull/7)
94+
* Deeplearning4j zoo models and datasets hosting location updated [Link](https://github.com/eclipse/deeplearning4j/pull/8292)
95+
* Fixed nIn validation for Deconv2D layer [Link](https://github.com/eclipse/deeplearning4j/issues/8225)
96+
* Fixed an issue with incorrect Deconvolution2d results for Keras import models [Link](https://github.com/eclipse/deeplearning4j/issues/8298)
97+
* Added DNNL/MKLDNN support for batch normalization layer [Link](https://github.com/KonduitAI/deeplearning4j/pull/14)
98+
* Fixed various integer casts to avoid overflows for very large arrays (with dimensions or length > Integer.MAX_VALUE) [Link](https://github.com/KonduitAI/deeplearning4j/pull/15)
99+
* Fixed an issue with UNet non-pretrained model architecture (last layer kernel size) [Link](https://github.com/eclipse/deeplearning4j/issues/8214)
100+
* Deeplearning4j SameDiff layers now use DL4J workspaces for better performance and reduced memory consumption [Link](https://github.com/KonduitAI/deeplearning4j/pull/23)
101+
102+
103+
### Deeplearning4j: Transition Guide, 1.0.0-beta5 to 1.0.0-beta6
104+
105+
106+
107+
### Deeplearning4j: 1.0.0-beta6 Known Issues
108+
109+
110+
## <a name="onezerozerobeta6-nd4j">ND4J and SameDiff</a>
111+
112+
### ND4J/SameDiff: Features and Enhancements
113+
114+
* DNNL (MKL-DNN) upgraded to version 1.1
115+
* Added SameDiff.calculateGradientsAndOutputs method [Link](https://github.com/eclipse/deeplearning4j/issues/8318) [Link](https://github.com/KonduitAI/deeplearning4j/pull/21/)
116+
* Additional SameDiff single batch .output method overloads for DataSet/MultiDataSet added [Link](https://github.com/SkymindIO/deeplearning4j/pull/253)
117+
* TensorFlow import ops coverage enhanced (significant number of additional ops supported) [Link](https://github.com/SkymindIO/deeplearning4j/pull/254), [Link](https://github.com/eclipse/deeplearning4j/pull/8341), [Link](https://github.com/KonduitAI/deeplearning4j/pull/25)
118+
* PRelu op added [Link](https://github.com/eclipse/deeplearning4j/pull/8247)
119+
* adjust_contrast, igamma and igammac ops added [Link](https://github.com/KonduitAI/deeplearning4j/pull/1)
120+
* ND4J/SameDiff: BitCast, CompareAndBitpack, DivideNoNan, DrawBoundingBoxes, FakeQuantWithMinMaxVarsPerChannel ops added [Link](https://github.com/KonduitAI/deeplearning4j/pull/2)
121+
* non_max_suppression_overlaps op added [Link](https://github.com/KonduitAI/deeplearning4j/pull/9)
122+
* ImagePreProcessingScaler now supports segmentation use cases [Link](https://github.com/eclipse/deeplearning4j/issues/8135)
123+
* concat operation now supports the concatenation axis being specified via the last input array [Link](https://github.com/eclipse/deeplearning4j/issues/8285)
124+
* Added Gamma and Poisson RNG distributions [Link](https://github.com/KonduitAI/deeplearning4j/pull/27)
125+
* SameDiff's use of DeviceLocal for variables/constants etc is now configurable [Link](https://github.com/KonduitAI/deeplearning4j/pull/32)
126+
* Uniform distribution op now supports random integer generation, not just random floating point generation [Link](https://github.com/KonduitAI/deeplearning4j/pull/30)
127+
128+
### ND4J/SameDiff: Bug Fixes and Optimizations
129+
130+
* OpenMP replaced with ThreadPool abstraction, enables parallelism for platforms without OpenMP support [Link](https://github.com/KonduitAI/deeplearning4j/pull/8)
131+
* SameDiff memory management overheauled for (in some cases significantlny) reduced memory consumption and improved performance [Link](https://github.com/KonduitAI/deeplearning4j/pull/10)
132+
* Switched to Clang instead of gcc for OSX compilation to avoid compiler-related issues [Link](https://github.com/KonduitAI/deeplearning4j/pull/8)
133+
* Removed `SameDiff.outputs()` "best guess" output inference due to being unreliable, in favor of explicit `SameDiff.setOutputs(String...)` call [Link](https://github.com/eclipse/deeplearning4j/issues/8265)
134+
* Fixed an issue with Nd4j.hstack on 1D arrays [Link](https://github.com/eclipse/deeplearning4j/issues/8218)
135+
* SameDiff no longer allows empty arrays for variables [Link](https://github.com/eclipse/deeplearning4j/issues/8209)
136+
* Fixed an issue with Nadam updater LR schedules not being cloned [Link](https://github.com/eclipse/deeplearning4j/pull/8243)
137+
* Cleaned up IActivation interface [Link](https://github.com/eclipse/deeplearning4j/pull/8261)
138+
* Added new LSTM op implementation with DNNL/MKLDNN support (forward pass only so far) [Link](https://github.com/KonduitAI/deeplearning4j/pull/4)
139+
* SameDiff API cleaned up; deprecated methods removed [Link](https://github.com/KonduitAI/deeplearning4j/pull/12)
140+
* Switched SameDiff variable initialization to non-lazy, to avoid unexpected behaviour when mixing execution and ND4J RNG seed setting [Link](https://github.com/eclipse/deeplearning4j/issues/8248)
141+
* SameDiff.zero and .one methods now create constants, not vairables [Link](https://github.com/eclipse/deeplearning4j/issues/8224)
142+
* Moved CUDA build version and device logging to Java logging, from c++ stdout to enable disabling logging (via ND4J config or slf4j config) [Link](https://github.com/eclipse/deeplearning4j/issues/8270)
143+
* Added DNNL/MKLDNN support for batch normalization [Link](https://github.com/KonduitAI/deeplearning4j/pull/14)
144+
* SameDiff: Fixed an issue where listeners weren't being called for gradient calculation [Link](https://github.com/eclipse/deeplearning4j/issues/8319)
145+
* Added DNNL/MKLDNN support for deconv2d/3d operations [Link](https://github.com/KonduitAI/deeplearning4j/pull/24)
146+
* Fixed an issue with biasadd_bp operation and NHWC data format [Link](https://github.com/eclipse/deeplearning4j/issues/8280)
147+
* Fixed an issue with certain strided slice backprop configurations [Link](https://github.com/eclipse/deeplearning4j/issues/8342), [Link](https://github.com/KonduitAI/deeplearning4j/pull/29)
148+
* Fixed an issue with LogSumExp reduction operation backprop for along dimension case [Link](https://github.com/KonduitAI/deeplearning4j/pull/35), [Link](https://github.com/eclipse/deeplearning4j/issues/8360)
149+
150+
151+
152+
### ND4J: Transition Guide, 1.0.0-beta5 to 1.0.0-beta6
153+
154+
* `SameDiff.outputs()` now requires user to call `SameDiff.setOutputs(String...)` first; previous "best guess" output inference was unreliable [Link](https://github.com/eclipse/deeplearning4j/issues/8265)
155+
* SameDiff.zero and .one methods now create constants, not vairables [Link](https://github.com/eclipse/deeplearning4j/issues/8224)
156+
157+
158+
### ND4J: 1.0.0-beta6 Known Issues
159+
160+
## <a name="onezerozerobeta6-datavec">DataVec</a>
161+
162+
### DataVec: Features and Enhancements
163+
164+
165+
### DataVec: Bug Fixes and Optimizations
166+
167+
* NDArrayScalarOpTransform now supports modulus operator [Link](https://github.com/eclipse/deeplearning4j/pull/8330)
168+
169+
170+
## <a name="onezerozerobeta6-rl4j">RL4J</a>
171+
172+
### RL4J: Features and Enhancements
173+
174+
* Added AsyncTrainingListener [Link](https://github.com/eclipse/deeplearning4j/pull/8072)
175+
* Replaced multiple uses of java.util.Random with ND4J Random [Link](https://github.com/eclipse/deeplearning4j/pull/8282)
176+
177+
### RL4J: Bug Fixes and Optimizations
178+
179+
* Refactored RL4J video recording to separate VideoRecorder class [Link](https://github.com/eclipse/deeplearning4j/pull/8106)
180+
* Fixed an issue with target for DQN [Link](https://github.com/eclipse/deeplearning4j/pull/8250), [Link](https://github.com/eclipse/deeplearning4j/issues/8107)
181+
* Refactoring for DQN and double DQN for improved maintainability [Link](https://github.com/eclipse/deeplearning4j/pull/8267)
182+
* Internal refactoring and various bug fixes [Link](https://github.com/eclipse/deeplearning4j/pull/8303)
183+
184+
185+
## <a name="onezerozerobeta6-arbiter">Arbiter</a>
186+
187+
### Bug Fixes and Optimizations
188+
189+
### Arbiter: Known Issues
190+
191+
192+
## <a name="onezerozerobeta6-nd4s">ND4S</a>
193+
194+
### ND4S Features and Enhancements
195+
196+
197+
198+
## PyDataVec
199+
200+
### PyDataVec Features and Enhancements
201+
202+
* PyDataVec TransformProcess now supports non-inplace operations [Link](https://github.com/eclipse/deeplearning4j/pull/8326)
203+
204+
205+
---
206+
---
207+
---
208+
209+
210+
66211
# <a name="onezerozerobeta5">Release Notes for Version 1.0.0-beta5</a>
67212

68213
## Highlights - 1.0.0-beta5 Release

0 commit comments

Comments
(0)

AltStyle によって変換されたページ (->オリジナル) /