graphlearn-for-pytorch

We are thrilled to announce the release of GraphLearn for PyTorch v0.2.3. This update includes some enhancements focusing on:

• Distributed support for vineyard as an integration with GraphScope.
• Optimizations such as graph caching, and some experimental features including support for bf16 precision and all-to-all communication.
• Some bug fixes.

What's Changed

• IGBH: Add Dockerfile and some minors by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/128
• IGBH: adjust mllogger tag position by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/130
• Distributed support for v6d GraphScope by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/116
• add graph caching support for distributed training by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/132
• add bf16 precision support to utilize Intel's AMX accelerations by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/133
• fix: data path for s/c case by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/135
• add all2all support to replace p2p rpc, using gloo as backend temporarily by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/134
• Fix table dataset init graph by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/136
• Verify hops as dict with 0 by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/138
• fix: several bugs on distributed mode by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/139
• [fix] explicitly call share_memory_ in Feature in cpu mode by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/142
• upgrade pytorch and cuda versions by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/141

Full Changelog: https://github.com/alibaba/graphlearn-for-pytorch/compare/v0.2.2...v0.2.3

We're excited to announce the release of GraphLearn for PyTorch v0.2.2. This update brings numerous fixes and features enhancing the framework's functionality, performance, and user experience. We extend our gratitude to all contributors who have made this release possible.

What's Changed

• [Fix] ensure consistency between the seeds added to and retrieved from the multiprocessing queue using the put and get methods by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/65
• [Fix] skip sampling empty inputs by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/67
• [Fix] try to fix tensor.nbr by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/71
• Fix igbh example by using proper parameters by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/70
• [Feat] put input data on server and allow for n to n connection between servers and clients by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/59
• [Build] adjust setup.py and create an ext_module util function by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/73
• [Feat] add "trim_to_layer" support to igbh example by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/74
• [Feat] Add edge weight sample for cpu by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/72
• [Fix] Skip mem-sharing the feature tensor in cpu mode by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/75
• [Fix] fix empty out_cols calling torch.cat by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/78
• [Fix] enable gc when evaluation by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/81
• [Feat] supports node split for both Dataset and DistDataset by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/82
• [Feat] load dataset from vineyard by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/80
• [Feat] Refactor RPC connection in server-client mode by @Jia-zb in https://github.com/alibaba/graphlearn-for-pytorch/pull/83
• [Feat] Add fields parsing for GraphScope side by @Jia-zb in https://github.com/alibaba/graphlearn-for-pytorch/pull/84
• [Build] split building of glt and glt_v6d by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/85
• [Feat] Multithread partition by @Jia-zb in https://github.com/alibaba/graphlearn-for-pytorch/pull/88
• [Fix] fix get GRAPHSCOPE_HOME in os.environ by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/89
• [CI] glt v6d ci by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/90
• [Feat] add trim_to_layer support to igbh distributed training by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/87
• [Fix] fix test: check partition dir exists before test by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/91
• [Feat] Update IGBH example by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/92
• [Fix] Fixes the build failure on MacOS and the compile flags settings in CMakeLists.txt by @sighingnow in https://github.com/alibaba/graphlearn-for-pytorch/pull/93
• [Feat] enable continuous downloading for large dataset by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/94
• [Feat] support two-stage partitioning by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/95
• [Fix] update the label index range for igbh-full dataset by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/96
• IGBH: synchronize after evaluation completes by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/97
• IGBH updates by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/98
• IGBH: persist feature when using FP16 by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/99
• Fp16 support by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/100
• [Fix] fix GPU allocation while splitting training and sampling in distributed training by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/101
• [Fix] Large file process by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/103
• [Feat] Refine IGBH preprocessing by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/105
• [Feat] Expose random seed configuration for single-node and distributed training by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/106
• [Fix] ML Perf code freeze minors by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/108
• [Fix] Fixes include path resolution on MacOS by @sighingnow in https://github.com/alibaba/graphlearn-for-pytorch/pull/109
• [Fix] Use a lock to protect the critical path of sampler initialization in neighbor sampler by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/110
• [Fix] adjust the lock location by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/111
• [Fix] add argument of channel size by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/113
• [Feat] IGBH: add MLPerf logging and control of evaluation frequency by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/114
• [Feat] Add gpt example by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/115
• [Feat] IGBH: support specifying the fraction of validation seeds by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/117
• [Fix] delete unused code by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/121
• [Feat] Separate training batch size and validation batch size in IGBH by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/122
• [Feat] Add mechanism of save/load checkpoint by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/123
• [Fix] add random seed para for link and subgraph loader by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/124
• [Fix] properly handle drop last in distributed sampler by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/125

New Contributors

• @sighingnow made their first contribution in https://github.com/alibaba/graphlearn-for-pytorch/pull/93

Full Changelog: https://github.com/alibaba/graphlearn-for-pytorch/compare/v0.2.1...v0.2.2

We are delighted to bring a number of improvements to GLT, alongside the 0.2.0 release. This release contains many new features, improvements/bug fixes and examples, which are summarized as follows:

1. Add support for single-node and distributed inbound sampling, provide users options of both inbound and outbound sampling.
2. Add chunk partitioning when partitioning graphs with large feature files, reduce the memory consumption of feature partitioning.
3. Add examples for the IGBH dataset
4. Fix bugs and improve system stability

What's Changed

• fix: clone the id chunk before pickle dump to avoid dumping the entire tensor by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/44
• Update figure by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/45
• Feature: In bound sampling of single machine by @Jia-zb in https://github.com/alibaba/graphlearn-for-pytorch/pull/48
• fix bug 'index out of bounds for partition book List' for igbh-large … by @kaixuanliu in https://github.com/alibaba/graphlearn-for-pytorch/pull/49
• Fix igbh rgnn example by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/50
• Add distributed in-sample functions by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/51
• [Example] clarify the setting of the number of servers and clients by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/52
• Fix igbh rgnn example by @Jia-zb in https://github.com/alibaba/graphlearn-for-pytorch/pull/53
• [bug] fix invalid configuration argument when samplers return torch.empty by @husimplicity in https://github.com/alibaba/graphlearn-for-pytorch/pull/54
• [Example] Separate server and client launch scripts for server-client mode distributed training by @Zhanghyi in https://github.com/alibaba/graphlearn-for-pytorch/pull/56
• Add IGBH multi-card single-node example & bug fix when mem-sharing graphs by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/62
• Update the igbh readme of single-node multi-GPU training by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/63
• Bump version to 0.2.1 by @LiSu in https://github.com/alibaba/graphlearn-for-pytorch/pull/64

New Contributors

• @Jia-zb made their first contribution in https://github.com/alibaba/graphlearn-for-pytorch/pull/48
• @Zhanghyi made their first contribution in https://github.com/alibaba/graphlearn-for-pytorch/pull/52

Full Changelog: https://github.com/alibaba/graphlearn-for-pytorch/compare/v0.2.0...v0.2.1

We are pleased to announce the first open release of GLT v0.2.0!

• GLT provides both CPU-based and GPU-based graph operators, including neighbor sampling, negative sampling, and feature lookup. The GPU-based graph operations significantly accelerate computation and reduce data movement, making it suitable for GPU training.
• For distributed training, GLT implements multi-processing asynchronous sampling, pin memory buffer, hot feature cache, and utilizes fast networking technologies (PyTorch RPC with RDMA support) to speed up distributed sampling and reduce communication.
• GLT is also easy to use, with most of its APIs compatible with PyG/PyTorch and complete documentation and usage examples available, GLT focuses on real-world scenarios and provides distributed GNN training examples on large-scale graphs.