Founding Engineer
- 👨🏼💻 I love coding and learning new things
🤩 I'm interested in
rusty1s
Follow
rusty1s
Follow
Achievements
Achievements
Highlights
- Pro
- 305 discussions answered
Pinned
4,217 contributions in the last year
Less
More
Contribution activity
April 2022
Created a pull request in pyg-team/pytorch_geometric that received 1 comment
Opened 21 other pull requests in 4 repositories
pyg-team/pytorch_geometric
9
merged
1
open
-
Better fallback in case
pytorch-lightningis missing -
HeteroData:num_featuresimplementation -
Replace
\Thetasymbols in documentation -
Custom
explain_messagetest - Dynamic nightly versioning
- Clean up warnings
- Optional dependencies
-
Fix
summarymatrix in DGI - Nightly build
-
Optional
pytorch-spline-convin conda
pyg-team/pyg-lib
1
open
6
merged
rusty1s/pytorch_sparse
1
open
2
merged
rusty1s/pytorch_cluster
1
merged
Reviewed 35 pull requests in 3 repositories
pyg-team/pytorch_geometric
25 pull requests
-
Add temporal sampling support to
NeighborLoader -
Unsupervised
GraphSAGEexample onPPI - Implement Lightning module for GraphGym
-
Test labels in
LinkNeighborLoader -
Fix
DimeNetenvelope bug with missing clamping -
fix minor bug in
LinkNeighborLoaderwhen data.x does not exist - add an example of pytorch ignite training with console and tensorboar…
-
Add random negative sampling to
LinkNeighborLoader -
Add
HeteroDatasupport toRemoveIsolatedNodes -
Fix
from_networkxin case where attributes are tensors - Bump installed software to newer versions
- [pre-commit.ci] pre-commit autoupdate
-
Add support for projecting features before aggregation in
SAGEConv -
Add Geom-GCN splits to
Planetoiddataset -
Convert edge types input for
LinkNeighborLoaderto canonical form. - Implement Disentangled GCN from ICML2019
- Dimenet++ implementation
-
Fix
LinkNeighborLoaderin casesrc_node_type = dst_node_type -
Link level
NeighborLoader -
MessagePassing.explain_messagefunctionality. -
Support for
LSTMaggregator inSAGEConv - Fix the original broken download url of DBP15K
- Updates to dimenet model
- Update message_passing.py
- Optional dependencies
- Some pull request reviews not shown.
pyg-team/pyg-lib
6 pull requests
rusty1s/pytorch_sparse
3 pull requests
Created an issue in pyg-team/pytorch_geometric that received 4 comments
Global Pooling via EquilibriumAggregation
4
comments
Opened 2 other issues in 2 repositories
pyg-team/pytorch_geometric
1
open
PyTorchLightning/pytorch-lightning
1
closed
Answered 31 discussions in 1 repository
pyg-team/pytorch_geometric
pyg-team/pytorch_geometric
- Node-level regression problem: take inputs of a subset of nodes to predict the behavior of all nodes
- Relating the PointNet++ classification model to the original paper architecture description (part 2)
- When 'RuntimeError: The 'data' object was created by an older version of PyG.' evokes?
- KL loss scaled
- Question about torch_geometric.transforms.RadiusGraph
- No module named 'torch_geometric.data.storage'
- Creating custom dataset from list of networkx graphs
- Gather from different graphs in batch by index
- How to implement the more advanced sampler (e.g., ClassStratifiedSampler)?
- Heterogeneous graph classification
- Hetero conv question
- DataLoader cant load batch sizes bigger than 10
- knn_graph function outputs
- PNA without edge_attributes
- Doubt about hetero conv
- Can I get the index of the sampled node in the original data after using NeighborLoader for sampling
- DataLoader cant load batch sizes bigger than 10
- Heterogeneous Graph with no features
- Specific implementation of Data and MessagePassing for undirected graphs?
- Does SAGEConv operator perform sampling?
- Dataset always rewrite
- Kernel size on Spline Conv?
- Advice in dynamic predictions where new nodes are added at each time-step; one node can have multiple edges
- Activation function in GCNConv layer.
- Discussion about graph autoencoding architectures
- Some discussions not shown.
121
contributions
in private repositories
Apr 1 – Apr 23