Tutel v0.3.2

What's New in v0.3.2:

1. Add --use_tensorcore option for benchmarking in tutel.examples.helloworld.
2. Read TUTEL_GLOBAL_TIMEOUT_SEC from environment variable to configure NCCL timeout setting.
3. Extend tutel.examples.helloworld_custom_expert to explain the way to override MoE with customized expert layers.

How to Setup:
python3 -m pip install -v -U --no-build-isolation https://github.com/microsoft/tutel/archive/refs/tags/v0.3.2.tar.gz

Tutel v0.3.1

What's New in v0.3.1:

1. Enable 2 additional collective communication primitives: net.batch_all_to_all_v(), net.batch_all_gather_v().

How to Setup:
python3 -m pip install -v -U --no-build-isolation https://github.com/microsoft/tutel/archive/refs/tags/v0.3.1.tar.gz

Tutel v0.3.0

What's New in v0.3.0:

1. Support Megablocks-style dMoE inference (see README.md for more information)

How to Setup:
python3 -m pip install -v -U --no-build-isolation https://github.com/microsoft/tutel/archive/refs/tags/v0.3.0.tar.gz

Tutel v0.2.1

What's New in v0.2.1:

1. Support Switchable Parallelism with example tutel.examples.helloworld_switch.

How to Setup:
python3 -m pip install --user https://github.com/microsoft/tutel/archive/refs/tags/v0.2.1.tar.gz

Tutel v0.2.0

What's New in v0.2.0:

1. Support Windows Python3 + Torch Installation;
2. Add examples to enable Tutel MoE in Fairseq;
3. Refactor MoE Layer implementation, letting all features (e.g. top-X, overlap, parallel_type, capacity, ..) be able to change at different forward interations;
4. New features: load_importance_loss, cosine router, inequivalent_tokens;
5. Extend capacity_factor value that includes zero value and negative values for smarter capacity estimation;
6. Add tutel.checkpoint conversion tools to reformat checkpoint files, making it able to use existing checkpoints to train/infer with a different world size.

How to Setup:
python3 -m pip install --user https://github.com/microsoft/tutel/archive/refs/tags/v0.2.0.tar.gz

Tutel v0.1.5

What's New in v0.1.5:

1. Add 2D hierarchical a2a algorithm used for extremely-large scaling;
2. Support different parallel_type for MoE computation: data, model, auto;
3. Combine different expert granularities (e.g. normal, sharded experts, megatron dense ffn) into same programming interface & style;
4. New features: is_postscore to specify whether gating scores are weighed during encoding or decoding;
5. Enhance existing features: JIT compiler, a2a overlap with 2D.

How to Setup:
python3 -m pip install --user https://github.com/microsoft/tutel/archive/refs/tags/v0.1.5.tar.gz

Contributors: @abuccts, @yzygitzh, @ghostplant, @EricWangCN

Tutel v0.1.4

What's New in v0.1.4:

1. Enhance communication features: a2a overlap with computation, support different granularity of group creation, etc.
2. Add single-thread CPU implementation for correctness check & reference;
3. Refine JIT compiler interface for flexible usability: jit::inject_source && jit::jit_execute;
4. Enhance examples: fp64 support, cuda amp, checkpointing, etc.
5. Support execution inside torch.distributed.pipeline.

How to Setup:
python3 -m pip install --user https://github.com/microsoft/tutel/archive/refs/tags/v0.1.4.tar.gz

Contributors: @yzygitzh, @ghostplant, @EricWangCN

Tutel v0.1.3

What's New in v0.1.3:

1. Add Tutel Launcher Support based on Open MPI;
2. Support Establishing Data Model Parallel in Initialization;
3. Support Single Expert Evenly Sharded on Multiple GPUs;
4. Support List of Gates and Forwarding MoE Layer with Specified Gating Index;
5. Fix NVRTC Compatibility when Enabling USE_NVRTC=1;
6. Other Implementation Enhancements & Correctness Checking;

How to Setup:
python3 -m pip install --user https://github.com/microsoft/tutel/archive/refs/tags/v0.1.3.tar.gz

Contributors: @ghostplant, @EricWangCN, @guoshzhao.

Tutel v0.1.2

What's New in v0.1.2:

1. General-purpose top-k gating with {'type': 'top', 'k': 2};
2. Add Megatron-ML Tensor Parallel as gating type;
3. Add deepspeed-based & megatron-based helloworld example for fair comparison;
4. Add torch.bfloat16 datatype support for single-GPU;

How to Setup:
python3 -m pip install --user https://github.com/microsoft/tutel/archive/refs/tags/v0.1.2.tar.gz

Contributors: @ghostplant, @EricWangCN, @foreveronehundred.

Tutel v0.1.1

What's New in v0.1.1:

1. Enable fp16 support for AMDGPU.
2. Using NVRTC for JIT compilation if available.
3. Add new system_init interface for initializing NUMA settings in distributed GPUs.
4. Extend more gating types: Top3Gate & Top4Gate.
5. Allow high level to change capacity value in Tutel fast dispatcher.
6. Add custom AllToAll extension for old Pytorch version without builtin AllToAll operator support.

How to Setup:
python3 -m pip install --user https://github.com/microsoft/tutel/archive/refs/tags/v0.1.1.tar.gz

Contributors: @jspark1105 , @ngoyal2707 , @guoshzhao, @ghostplant .