These 7 commits are when the Protocol Buffers files have changed:
| Commit: | 79391ab | |
|---|---|---|
| Author: | Alex Samylkin | |
Implement update node features
| Commit: | 71e5fca | |
|---|---|---|
| Author: | Alex | |
| Committer: | GitHub | |
Implement uniform sampling for temporal graphs and add timestamps to result tuples (#180) * Implement uniform sampling for temporal graphs and return timestamps --------- Co-authored-by: Alex Samylkin <alsamylk@microsoft.com>
The documentation is generated from this commit.
| Commit: | c225e86 | |
|---|---|---|
| Author: | Alex | |
| Committer: | GitHub | |
Temporal graph support (#176) Add temporal graph support and lastn sampling
| Commit: | 4992935 | |
|---|---|---|
| Author: | Alex | |
| Committer: | GitHub | |
Introduce version for binary data. (#80) Introduce version for binary files Co-authored-by: Alex Samylkin <alsamylk@microsoft.com>
| Commit: | 8ad4738 | |
|---|---|---|
| Author: | Adithya Bhonsley | |
| Committer: | GitHub | |
Neighbor Count Functionality for Memory/Distributed Graphs (#44)
| Commit: | b835f78 | |
|---|---|---|
| Author: | Alex Samylkin | |
| Committer: | GitHub | |
Implement features with variable length (#18) he code is very similar to dense features, extracting features through index look ups, but without back fills. The major difference, however, is we don't know feature dimensions in advance and can't pre-allocate arrays in Python. This means we need to use callbacks from C++(py_data.cc) that will tell python interpreter(_StringFeatureCallback) where to copy data from when all features are fetched from graph engine.
| Commit: | ed11f2b | |
|---|---|---|
| Author: | Alex Samylkin | |
Initial release