Releases: deepforge-dev/deepforge
Releases · deepforge-dev/deepforge
v2.5.0
Overview
- Update code editor to use monaco and LSP #571#1800
- Support for reification of artifact provenance #1734#1946
- Improvements to documentation #1895#1896
- Improve output parsing from operation definition #1815#1890
- Improvements to execution index and graph visualization #1907#1909#1975#1976#2027#2028
- Bug fixes to TwoPhaseCommit logic #1955#1956#1953#1954#1960#1961
- Library import capabilities (from extensions) #1993#1994
- Job execution bug fixes #1973#1974#1978#1979#2000#2001
- Miscellaneous code cleanup and improvements #1874#1877#1879#1880#1882#1883#1878#1881#1930#1931#1972#2029#2030
- Save operation metadata in output directory #1917#1919
Interactive Editors
- Add interactive training editor and workspace (disabled on stable) #1808#1816#1938#1941#1926#1943#1933#1952#1963#1964#1965#1966#1967#1968
- Add to existing session; show loading status #1885#1937
- Miscellaneous other improvements #1939#1940
Compute
- Provide job name when submitting to SciServer Compute #1830#1889
- Purge support #1724#1922
- Add account linking support #2005#2006#2011#2012#2009#2010#1999#2008#2013#2014
Interactive compute
v2.4.1
v2.4.0
Overview
- Record data provenance in artifacts, job output #1705#1708#1728#1732
- Improved data handling (streaming) #1702#1703#1777#1814#1840#1841
- Ensure data is not stored on DeepForge when uploaded #1704#1717
- Allow artifact type editing #1681#1682
- Minor CLI improvements #1678#1686
- Fixed bug with GME Blob storage with authenticated users #1645#1838
- Minor improvements to SciServer compute adapter #1790#1791#1822#1823#1826#1828#1849#1850#1855#1864
- Store PlotlyJSON in model rather than own representation #1745#1747#1805#1806#1845#1846#1857#1858
- Deployment/Documentation improvements #1721#1722#1736#1737#1755#1756#1757#1758#1762#1763#1780#1782#1740#1786#1787#1801
Bleeding Edge Updates (not fully released)
- Add support for exploring tensors interactively #1692#1694#1845#1846#1842#1843 (not available yet from the default UI)
- Add interactive compute capabilities #1700#1701#1691#1783#1784#1810#1811#1817#1820#1844#1847#1851#1852
- Add custom actions to action button #1742#1741#1772#1773#1776#1781
- Add interactive editing/exploration #1768#1770#1774#1775#1818#1821
Misc
- Improvements for internal dev tooling #1711#1712#1706#1707#1713#1716#1729#1730#1752#1754#1760#1771#1788#1789
- Misc code cleanup #1698#1699#1709#1710#1696#1697#1714#1715#1723#1735#1738#1739#1750#1751#1797#1798#1829#1831
- Minor bug fixes #1684#1685#1687#1688#1726#1727#1731#1733#1743#1744#1748#1749#1764#1765#1779#1792#1793#1802#1803#1833#1834#1835#1837#1827#1836#1853#1854#1859#1860
v2.3.1
v2.3.0
Overview
- Improvements to SciServer storage and compute adapter:
- Import existing artifact from storage backend #1608#1630#1660#1661
- Improvements to config dialogs #1650#1651#1658#1659#1653#1664
- Misc bug fixes and logging improvements #1628#1632#1633#1643#1644#1654#1655#1663#1662#1667
- Deployment updates #1626#1627#1640#1641#1642
v2.2.1
v2.2.0
Overview
- Add storage, compute adapters enabling usage with existing cyberinfrastructure #1186#1227#1245#1255 ...
- Add scatter, 3D plots, and image support to matplotlib backend #1572#1308#1329#1218#1373 ...
- Add support for custom python dependency management w/ conda #1466#1514#1528#1532 ...
- Misc code clean up and refactors #1233#1306 ...
v2.1.1
v2.1.0
Overview
- Added option for job caching (disabled by default) #1204#1203
- Improved error messages #1201#1202#1159#1196#1193#1194
- Added compatibility for use with default webgme visualizers #1171#1172#1177#1180
- Record data origin #1168#1169
- Improved support for working with libraries #1173#1174#1178#1179
- Improved UI for showing time/duration of executions #1187#1188
- Misc cleanup #1182#1183
Bug Fixes
- Improved job handling when involuntarily exits #1205#1206#1207#1208#1209#1210
- Fixed naming of copied operations #1160#1161
- Ensure no race-conditions when working with pointers #1166#1167
- Fixed graphing issues #1191#1192#1189#1190
- Added missing python libs to docker image #1197#1198
- Documentation updates #1158
v2.0.0
Overview
The neural network architecture modeling language has been decoupled from the base DeepForge application and packaged as an extension (such as https://github.com/deepforge-dev/deepforge-keras). This allows for multiple deep learning frameworks to be used and allows for even other types of extensions.