Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

readme fixes #93

Merged
merged 1 commit into from
Jun 23, 2023
+11 −13
Merged

readme fixes #93

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 11 additions & 13 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@
[![CI status](https://github.com/BlackSamorez/tensor_parallel/actions/workflows/run-tests.yaml/badge.svg?branch=main)](https://github.com/BlackSamorez/tensor_parallel/actions)

<p align="center">
🚀 &nbsp;<b><a href="https://www.kaggle.com/code/blacksamorez/tensor-parallel-int8-llm/">Try new 20B LLMs demo in Kaggle</a></b>
🚀 &nbsp;<b><a href="https://www.kaggle.com/code/blacksamorez/tensor-parallel-int4-llm/">Try new 40B LLMs demo in Kaggle</a></b>
</p>

Run large PyTorch models on multiple GPUs in one line of code with potentially linear speedup.
Expand Down Expand Up @@ -43,14 +43,14 @@ For best memory efficiency, call `tp.tensor_parallel` while the model is still o

Here are a few use cases:
- [`examples/training_flan-t5-xl.ipynb`](./examples/training_flan-t5-xl.ipynb) - fine-tune full FLAN-T5 model on text summarization
- [`tensor_parallel int8 LLM`](https://www.kaggle.com/code/blacksamorez/tensor-parallel-int8-llm/) - inferencing a large language model with LLM.8bit + tensor_parallel
- [`tensor_parallel int8 LLM`](https://www.kaggle.com/code/blacksamorez/tensor-parallel-int8-llm/) - adapter-tuning a large language model with LLM.8bit + tensor_parallel
- __TBA__ - defining custom parallelism strategy


Advanced parameters to `tensor_parallel`:
- `device_ids: List[device]` - which devices to use; defaults to all available GPUs
- `output_device: device` - model outputs will have this device
- `tensor_parallel_config: tp.Config` - use custom parallelism strategy, see [`slicing_configs.py`](./tensor_parallel/slicing_configs.py)
- `tensor_parallel_config: tp.Config` - use custom parallelism strategy, see [`slicing_configs.py`](./src/tensor_parallel/slicing_configs.py)
- `distributed: bool` - if True, use torch.distributed backend instead of threading (requires `torchrun`)
- `sharded: bool` - if True, find all trainable parameters that weren't split by Tensor Parallelism and split them using [ZeRO-3 algorithm](https://deepspeed.readthedocs.io/en/latest/zero3.html).
- weights will be split between GPUs and re-assembled before each forward pass
Expand Down Expand Up @@ -113,11 +113,11 @@ tensor_parallel_state_dict = tp.convert_state_dict(
# Dispatch the partial state_dict (load_state_dict doesn't work with meta so here I use accelerate)
device_map = tp.infer_sharded_device_map(model)
for param_name, param in state_dict.items():
module_name = param_name
while len(module_name) > 0 and module_name not in device_map:
module_name = ".".join(module_name.split(".")[:-1])
param_device = device_map[module_name]
accelerate.utils.set_module_tensor_to_device(model, param_name, param_device, value=param)
module_name = param_name
while len(module_name) > 0 and module_name not in device_map:
module_name = ".".join(module_name.split(".")[:-1])
param_device = device_map[module_name]
accelerate.utils.set_module_tensor_to_device(model, param_name, param_device, value=param)
```

With this no more than one part of the model needs to be loaded into memory at once.
Expand Down Expand Up @@ -145,18 +145,16 @@ Why use `tensor_parallel` ...
- DeepSpeed has many parallelization strategies, but requires careful configuration
- tensor_parallel has one strategy that works with 1 line of code
- tensor_parallel works in a jupyter notebook
- v.s. [MegatronLM](https://github.com/NVIDIA/Megatron-LM)?
- v.s. [MegatronLM](https://github.com/NVIDIA/Megatron-LM)
- MegatronLM has _great_ tensor parallelism for one model architecture
- tensor_parallel has _good_ parallelism for any architecture
- tensor_parallel is way easier to install
- v.s. [parallelformers](https://github.com/tunib-ai/parallelformers)?
- parallelformers implements a fixed [list of architectures](https://github.com/tunib-ai/parallelformers/tree/main/parallelformers/transformers)
- tensor_parallel works for any architecture automatically
- v.s. [parallelformers](https://github.com/tunib-ai/parallelformers)
- parallelformers is inference-only, tensor_parallel supports training
- v.s. [`alpa`](https://github.com/alpa-projects/alpa)
- alpa is a powerful tool for automatic distributed training / inference in JAX
- tensor_parallel works with PyTorch
- v.s. [`Model.parallelize()`](https://huggingface.co/docs/transformers/model_doc/gpt2#transformers.GPT2Model.parallelize)?
- v.s. [`Model.parallelize()`](https://huggingface.co/docs/transformers/model_doc/gpt2#transformers.GPT2Model.parallelize)
- both are easy to use, both fit large models
- in parallelize, one GPU works at a time
- in tensor_parallel, GPUs work in parallel
Expand Down