v3.9.0 🌈

🧭 What's Changed

• New optional c++ based storage layer (#2733) @khustup2
• Check function passed to dataset.filter() is correctly configured (#2812) @nvoxland-al
• Avoid making http requests to validate unauthenticated users (#2810) @nvoxland-al
• Improved numpy conversion error messages (#2809) @nvoxland-al
• fixed tensor size calculation (#2808) @activesoull

⚙️ Who Contributes

@activesoull, @istranical, @khustup2, @nvoxland and @nvoxland-al

v3.8.27 🌈

🧭 What's Changed

• Don't require azure modules to be installed (#2803) @nvoxland-al
• Bump libdeeplake version to 0.0.109 (#2804) @activesoull
• Fix nan case for ingest_dataframe (#2800) @istranical

⚙️ Who Contributes

@activesoull, @istranical and @nvoxland-al

v3.8.26 🌈

🧭 What's Changed

• Azure SSO support (#2788) @dgaloop
• bump libdeeplake version to 0.0.108 (#2801) @activesoull

⚙️ Who Contributes

@activesoull, @dgaloop, @ian-activeloop and @nvoxland-al

v3.8.25 🌈

🧭 What's Changed

• Improve pre-signed URL handling for Azure (#2796) @dgaloop
• EF_Search Setting in Deeplake. (#2792) @sounakr
• Removed coco references from the "ingest_dataframe" docs (#2798) @joe-activeloop

⚙️ Who Contributes

@dgaloop, @joe-activeloop, @nvoxland-al and @sounakr

v3.8.24 🌈

🧭 What's Changed

• bump libdeeplake version to 0.0.107 (#2794) @activesoull
• Fixed tensor "sample_info" on some saved view samples (#2793) @nvoxland-al

⚙️ Who Contributes

@activesoull, @nvoxland and @nvoxland-al

v3.8.23 🌈

🧭 What's Changed

• Include path in sample decompression error (#2785) @FayazRahman
• Updated video docs (#2781) @istranical
• Generate a better error message if a list or other unsupported argument is passed to video.normalize_index (#2787) @nvoxland-al
• fix of An error occurred (400) when calling the HeadObject operation (#2789) @activesoull
• Validate htype vs dtype when detecting htype based on inserted data (#2786) @nvoxland-al

⚙️ Who Contributes

@FayazRahman, @activesoull, @istranical, @nvoxland and @nvoxland-al

v3.8.22 🌈

🧭 What's Changed

• Auto-convert numbers to strings when appending to a text dtype tensor (#2782) @nvoxland-al
• Use libdeeplake batch requests for getting bytes. (#2783) @khustup2
• Fixed allow_delete wording in messages (#2780) @nvoxland-al

⚙️ Who Contributes

@khustup2 and @nvoxland-al

v3.8.21 🌈

🧭 What's Changed

• Ensure username is stored correctly in commit messages (#2772) @nvoxland-al
• Added allow_delete flag to dataset (#2763) @nvoxland-al
• Removed use of the term "log in" and references to the CLI "activeloop login" (#2768) @nvoxland-al
• Removing support for tokens set via the CLI (#2774) @taylorstubbsactiveloop

⚙️ Who Contributes

@nvoxland-al and @taylorstubbsactiveloop

v3.8.20 🌈

🧭 What's Changed

• Update setup.py (#2770) @activesoull
• fix upcasting logic (#2773) @activesoull
• Update README to remove cli mention. (#2767) @khustup2

⚙️ Who Contributes

@activesoull, @khustup2 and @nvoxland-al

v3.8.19 🌈

🧭 What's Changed

• Removed Deeplake CLI (#2761) @nvoxland-al
• bump libdeeplake to 0.0.99 (#2765) @activesoull
• fix correct creds used (#2764) @activesoull
• Get rid of global random seed (#2759) @dgaloop
• Fixed edge cases in indra dataset adaptor. (#2762) @khustup2

⚙️ Who Contributes

@activesoull, @dgaloop, @khustup2 and @nvoxland-al