Axonn

Latest version: v0.2.0

Safety actively analyzes 681775 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

0.2.0

What's Changed
* Update README by bhatele in https://github.com/axonn-ai/axonn/pull/16
* fix evaluation bug for inter-layer by siddharth9820 in https://github.com/axonn-ai/axonn/pull/18
* Support for intra-layer parallelism by siddharth9820 in https://github.com/axonn-ai/axonn/pull/21
* add checkpointing and post backward hook support by siddharth9820 in https://github.com/axonn-ai/axonn/pull/24
* docs: fix readthedocs.org build issues by bhatele in https://github.com/axonn-ai/axonn/pull/26
* fix g_intra print by zsat in https://github.com/axonn-ai/axonn/pull/27
* Tests: convert memopt to int before bool by adityaranjan in https://github.com/axonn-ai/axonn/pull/28
* Docs: installation and running mnist test by adityaranjan in https://github.com/axonn-ai/axonn/pull/29
* add 2D tensor parallelism for FC layers by siddharth9820 in https://github.com/axonn-ai/axonn/pull/30
* readme: add slack link by bhatele in https://github.com/axonn-ai/axonn/pull/31
* CI/CD tests for intra-layer parallelism by siddharth9820 in https://github.com/axonn-ai/axonn/pull/33
* add AxoNN logo by bhatele in https://github.com/axonn-ai/axonn/pull/34
* changes to the intra-layer API for the GPT benchmark by siddharth9820 in https://github.com/axonn-ai/axonn/pull/36
* add dependencies between workflows by bhatele in https://github.com/axonn-ai/axonn/pull/41
* [WIP] ILP Conv Layer support by prajwal1210 in https://github.com/axonn-ai/axonn/pull/38
* Intra-layer - Overlap communication in backward pass by siddharth9820 in https://github.com/axonn-ai/axonn/pull/44
* [WIP] A tensor parallel API for beginners by siddharth9820 in https://github.com/axonn-ai/axonn/pull/40
* first iteration of 3D tensor parallelism by siddharth9820 in https://github.com/axonn-ai/axonn/pull/49
* Initialize layers on the GPU by siddharth9820 in https://github.com/axonn-ai/axonn/pull/51
* add option to change batch dimension in drop by siddharth9820 in https://github.com/axonn-ai/axonn/pull/52
* change outer variables by siddharth9820 in https://github.com/axonn-ai/axonn/pull/53
* A context manager to optimize communication by siddharth9820 in https://github.com/axonn-ai/axonn/pull/54
* Rebase axonn-cpu to master by Avuxon in https://github.com/axonn-ai/axonn/pull/56
* More communication optimizations by siddharth9820 in https://github.com/axonn-ai/axonn/pull/57
* Parallel transformers by jwendlan in https://github.com/axonn-ai/axonn/pull/59
* Added Depth Tensor Parallelism to Conv Layer by prajwal1210 in https://github.com/axonn-ai/axonn/pull/60
* Change overlap for depth tp and do not initialize MPI unless absolutely needed by siddharth9820 in https://github.com/axonn-ai/axonn/pull/62
* removed mpi4py dependency by S-Mahua in https://github.com/axonn-ai/axonn/pull/63
* adding parallelize context for opt by jwendlan in https://github.com/axonn-ai/axonn/pull/65
* Removing the drop and gathers in depth tensor parallelism for the easy API by siddharth9820 in https://github.com/axonn-ai/axonn/pull/66
* change parallelize context to use AutoConfig by siddharth9820 in https://github.com/axonn-ai/axonn/pull/67
* Bugfix: Initialize grad_input, grad_weight to None by adityaranjan in https://github.com/axonn-ai/axonn/pull/68
* docs: fix build issues and add sub-sections by bhatele in https://github.com/axonn-ai/axonn/pull/69
* added automatic_parallelism by S-Mahua in https://github.com/axonn-ai/axonn/pull/70
* This PR shards the Dataloader across depth and data parallel ranks both by siddharth9820 in https://github.com/axonn-ai/axonn/pull/74
* Make monkeypatching more efficient and change easy API to a single argument by siddharth9820 in https://github.com/axonn-ai/axonn/pull/72
* Add API for tensor parallel model checkpointing by siddharth9820 in https://github.com/axonn-ai/axonn/pull/77
* Changes to fix issues in IFT. by siddharth9820 in https://github.com/axonn-ai/axonn/pull/78
* AxonnStrategy for Lightning Fabric backend by anishbh in https://github.com/axonn-ai/axonn/pull/76
* initial doc for EasyAPI, Accelerate, and FT example by jwendlan in https://github.com/axonn-ai/axonn/pull/73
* User guide Changes by siddharth9820 in https://github.com/axonn-ai/axonn/pull/80
* Update advanced.rst by siddharth9820 in https://github.com/axonn-ai/axonn/pull/81
* More lightning features by siddharth9820 in https://github.com/axonn-ai/axonn/pull/82
* Supporting init_module, load/save checkpoint by siddharth9820 in https://github.com/axonn-ai/axonn/pull/83
* make no-grad-sync yield None by siddharth9820 in https://github.com/axonn-ai/axonn/pull/88
* create an engine for all things pipelining and deprecate custom mixed precision by siddharth9820 in https://github.com/axonn-ai/axonn/pull/91
* Tensor parallel embedding by siddharth9820 in https://github.com/axonn-ai/axonn/pull/93
* Improving AxoNN's memory consumption by siddharth9820 in https://github.com/axonn-ai/axonn/pull/95
* Correct url of ci tests badge by siddharth9820 in https://github.com/axonn-ai/axonn/pull/99
* reorg code and first implementation of the new easy API by siddharth9820 in https://github.com/axonn-ai/axonn/pull/96
* Minor changes for Release 0.2.0 by siddharth9820 in https://github.com/axonn-ai/axonn/pull/100

New Contributors
* zsat made their first contribution in https://github.com/axonn-ai/axonn/pull/27
* adityaranjan made their first contribution in https://github.com/axonn-ai/axonn/pull/28
* prajwal1210 made their first contribution in https://github.com/axonn-ai/axonn/pull/38
* jwendlan made their first contribution in https://github.com/axonn-ai/axonn/pull/59
* S-Mahua made their first contribution in https://github.com/axonn-ai/axonn/pull/63
* anishbh made their first contribution in https://github.com/axonn-ai/axonn/pull/76

**Full Changelog**: https://github.com/axonn-ai/axonn/compare/v0.1.0...v0.2.0

0.1.0

[AxoNN](https://axonn.readthedocs.io) is a parallel framework for training deep neural networks.

Features:
- Offers a hybrid of inter-layer parallelism with pipelining and data parallelism.
- Supports both 16-bit mixed precision and 32-bit full precision training.
- A highly efficient and scalable implementation of inter-layer parallelism with pipelining using asynchronous MPI-based communication and message-driven scheduling that achieves significant overlap of computation and communication.
- Memory optimizations that can reduce the model state memory consumption by 5x for mixed precision training using the Adam optimizer and indirectly also increase hardware efficiency.

Links

Releases

Has known vulnerabilities

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.