Sep 22 16:34 data: add fixed all-node evaluation negatives 16:32 docs: distinguish fixed negative candidate sets Aug 1 19:36 chore: exclude oversized MAG artifacts Jul 30 17:53 chore: remove redundant node role artifacts 02:40 docs: explain manual MAG artifact download Jul 29 22:59 docs: generate dataset catalog from manifests 22:09 refactor: remove unused state labels 21:59 fix: publish explicit temporal evaluation artifacts 17:47 chore: move dataset tooling to FROST Jul 27 08:51 Replace GDELT int_full with official TGL artifact 08:08 Add MAG edges 07:14 Add GDELT dataset Apr 15 01:23 Update default macro batch size for LASTFM dataset Apr 14 08:01 Update default macro batch size for WIKI and WIKIPEDIA datasets Lei Zhao
06:20 Update recommended max_macro_batch_size calculations and values for datasets Lei Zhao
Apr 10 17:46 Add dataset-specific training defaults for macro batch sizes Apr 8 20:35 Update default macro batch sizes for datasets in dataset_defaults.sh Apr 7 16:26 Update default macro batch sizes for datasets in dataset_defaults.sh Apr 6 16:57 Add new scripts for CSV validation and dataset processing, including checks for column values and generating node roles Mar 30 08:45 Add node_role.npy files for LASTFM, MOOC, REDDIT, and WIKIPEDIA datasets 08:39 Add new node_role.npy files for LASTFM, MOOC, REDDIT, and WIKIPEDIA datasets Mar 29 09:15 rename expected speed note to English DyGFormer runtime guide and link it from DATA README Mar 28 07:07 add deterministic precomputed negative sampling by eid 06:25 Update edges.csv files to reflect new file sizes and hashes 00:04 Fix case sensitivity in dataset directory names in README.md 00:03 Add new datasets and features for various projects - Added edge features and edges CSV files for CANPARL, CONTACTS, ENRON, FLIGHTS, LASTFM, MOOC, REDDIT, SOCIALEVO, UCI, UNTRADE, USLEGIS, and WIKIPEDIA.
- Included full graph files with reverse edges for all mentioned projects.
- Introduced node features for LASTFM, MOOC, REDDIT, and WIKIPEDIA, linking to respective node role files.
- Updated ban labels for MOOC and REDDIT datasets.
- All new files are tracked with Git LFS for efficient storage and versioning. Mar 27 23:39 modify csv header 23:05 Update README.md to clarify dataset structure and normalization notes 22:58 Update README.md to correct filename for topology cache 22:56 Update .gitignore and add full_graph_with_reverse_edges.npz files for multiple datasets 22:31 regenerate graph 22:09 Update edges.csv files with new file sizes and hashes for multiple datasets 22:01 Update edges.csv files with new file sizes and hashes 20:13 Add edge and node feature files for various datasets and update corresponding CSV files 19:48 Add full_graph.npz files for multiple datasets and update README.md for clarity 18:12 remove raw_data 09:58 Update README.md to enhance dataset documentation and structure 08:54 Add FROST datasets snapshot 07:30 initial commit 
